272. Zeroes and Ones: Into The Depths of Computation | Jim Keller

[0] Hello, everyone.

[1] I'm pleased to welcome Jim Keller to my YouTube channel and podcast today.

[2] Jim is a microprocessor engineer known for his work at digital equipment, AMD, Apple, Tesla, and Intel.

[3] He was co -architect for what were among the earliest of 64 -bit micro processors, the EV5 and EV6 digital alpha processors designed in the 90s.

[4] In the later 90s, he served as lead architect for the AMD K8 microarchitecture, including the original Athlon 64, and was involved in designing the Athlon K7 and Apple A4 through seven processors.

[5] He was also the co -author of the specifications for the X86 -64 instruction set, and hyper -transport interconnect.

[6] From 2012 to 2015, he returned to AMD to work on the AMD K -12 and Zen microarchitectures.

[7] At Tesla, he worked on automotive autopilot hardware and software, designing the hardware -3 autopilot chip.

[8] He then served a senior VP of Silicon Engineering, heading a team of 10 ,000 people at Intel.

[9] He is presently, president and CTO at Tens Torrent, building AI computers.

[10] He's also my brother -in -law.

[11] And we've talked a lot over the last 20 years.

[12] He was a friend of mine before he married my sister.

[13] And we've known each other for a very long time since we both lived in Boston.

[14] So I'm very happy to have you to talk to you today, Jim.

[15] I'm really looking forward to it.

[16] So thanks for agreeing to do this.

[17] Sure thing.

[18] So let's walk through your career first.

[19] It takes some unpacking.

[20] Your resume takes some unpacking to be comprehensible, I would say.

[21] So let's start with digital equipment.

[22] You were working on very early stage sophisticated microprocessors.

[23] So tell me about that.

[24] Well, the long story is, you know, I graduated college.

[25] as an electrical engineer with a bachelor's and I took a job in Florida because I wanted to live on the beach and it turned out to be a really interesting job was at Harris and I spent like two years working in the labs fixing me up electrical equipment and doing some networking stuff and some digital design and at some point a friend told me I should work at digital so I read about digital and I literally read the computer architecture manual for the back 780 on the plane on the way to the interview And then I interviewed them with a whole bunch of questions because I just read this architecture spec, which I didn't know that much about, to be honest, but I was kind of a wise ass with a kid.

[26] And they hired me because they thought it was funny.

[27] And so I got my, you know, architectural education working on the Vax 8800, working for a guy named Bob Stewart, along with some other really great architects.

[28] So I spent about seven years in that group, you know, learning to be a processor.

[29] architect.

[30] And then I spent a little time at digital research in California for six months and then went back and joined the Hudson team where they were building alpha processors and became co -architect with Pete Bannon on EV5 and then with Dirk Meyer on EV6.

[31] Now those chips.

[32] It was about 15 years and we worked on, I would say, three very successful shipping products and then a couple other products that didn't get to market.

[33] So those chips from what I remember were remarkably ahead of their time, but that didn't seem to save digital equipment corporation.

[34] Is that fair to say?

[35] That's definitely fair to say.

[36] The digital literally made the world's fastest computer in the years they were going out of business.

[37] And there was a complicated market dynamic.

[38] Digital is very successful building mini computers in the mini computer age, which replaced the mainframes to a large extent.

[39] But they missed the boat on the PC revolution and the, let's say, workstations.

[40] And they had a big, expensive computer mindset rate when computer prices were falling in a lot.

[41] So we were building really fast processors for server and highing workstations.

[42] And the market had kind of moved on.

[43] And let's say a lot of crazy things were going on inside the company at the time.

[44] Yeah, well, we're going to return to the topic of crazy things going on inside computers.

[45] But it's interesting to note right there.

[46] so that that's a situation where a company has a great product but doesn't know how to launch it into the marketplace or is blinded by its own preconceptions.

[47] It can't even see necessarily what it happened.

[48] Gordon Bell was CTO and he was a really brilliant computer architects, but he also had really good, let's say, observational skills.

[49] And like during midway through the Vax 80800, he decided the technology we were using was too little too late and redirected the program and it became a very successful product because he knew what was going on and he made decisions like that.

[50] When he left, I think Digital became an argument between business unit managers, not about technology.

[51] And Alpha was a great technology, but it went into business units that were aiming at high prices and high margins, not market penetration and not basically keeping up of the software revolution that was happening in the time.

[52] So, Ken Olson was a great manager, but he wasn't a technical leader.

[53] and without Gordon Bell, the company kind of lost its way.

[54] When companies lose their way, they failed.

[55] And they failed fast, too.

[56] You know, they went from record profits to losing a billion a year, and they kind of rectified that for a little while, and then Shale dropped off again and over.

[57] Yeah, well, one of the things I've been struck by watching your career and talking to you over the years is exactly that.

[58] The rate at which a company that appears dominant can disintegrate and disappear is really quite something quite stunning.

[59] I think Fortune 500 companies tend to last no more than 30 years.

[60] That's approximately the span.

[61] And that's not a, that's not a tremendously long period of time.

[62] So there's always dominant companies and always a handful of dominant companies, but the company that's dominant tends to shift quite quickly.

[63] Well, we should return to this a couple of times as the, you know, the classic escrow in economics.

[64] You start out low, you solve a problem, you ramp up, the plateau, and then you fail.

[65] And this dominates business and dominates humanity at some level.

[66] And it, you know, plays out over and over.

[67] Now, it sounds like you didn't have.

[68] So how is it that you managed to do this job?

[69] You intimated when we were talking that you weren't really trained for it.

[70] And so you were trained as an engineer.

[71] You had an, is it a bachelor's degree in engineering?

[72] And how prepared were you as a consequence of your degree for any of the jobs that you undertook?

[73] Training is highly overrated.

[74] So a good engineering degree is math, physics, basic understanding of science and some smattering of communication skills.

[75] You can probably do a great engineering degree in two or three years if you're dedicated to it.

[76] You know, the things that stretch my brains the most when I went to college is with math and mechanical engineering.

[77] And Penn State, I went to Penn State.

[78] And they used mechanical engineering and masses too of the weed out courses to find out if you had the chops or the gumption to get through engineering.

[79] So there was a fairly high failure rate there.

[80] But mechanical engineering is a really interesting discipline because you have to think, think about solving mass problems spatially, like, you know, do things like how do you calculate the force on a rotating, accelerating, accelerating object?

[81] Like, it's somewhat complicated, and it makes you really think.

[82] So you have to learn to think.

[83] And in engineering school, you never answer a multiple choice question.

[84] You learn stuff and film is and stuff, but then you, calculate and to understand what the result is.

[85] And the problem sets are like little design exercises.

[86] How much of it do you think is pure screening, let's say, well, it wouldn't be pure screening, but fundamental screening for conscientiousness and IQ and how much of it is learning to think how much of the education process is that?

[87] Like if you're going to hire an engineer, are you hiring fundamentally on the of IQ and you get smarter people from the top schools?

[88] Or do you think that the engineering training actually does prepare people for a technical career?

[89] Well, it depends on the engineering depends on the school and depends on their approach.

[90] So my IQ isn't super high compared to really smart people.

[91] I mean, it's high enough.

[92] When I went to college, it took me about a year and a half to learn how to think properly.

[93] And I found for me personally, I had to do the work on a regular basis early.

[94] I didn't study for finals.

[95] I wasn't the kind of person that could pick up a book understanding and get an aid the next day and then forget about it.

[96] I'm not, I can't do that.

[97] So I had to learn how to do the work, go through the mechanisms, automatize some of the basics, so I didn't have to think about them so hard, but literally let my brain work on this stuff so then I could use them to go problem solve.

[98] And especially on engineering, there's lots of different kinds of engineering.

[99] There's like highly technical stuff where you turn to crank, like a skilled lawyer might.

[100] But there's other stuff where you have to be really creative.

[101] You have an unsolved problem and nobody solved before.

[102] And as an engineer, you have a skill set, right?

[103] But you have to apply it creatively.

[104] And there's lots of high IQ people who aren't creative and there's, you know, low IQ people that are creative.

[105] And you find in a big engineering team, there's a real diversity of personality types.

[106] There's open -minded people, conscientious people, you know, gregarious people.

[107] And it takes many different kinds of people working together to do something sophisticated.

[108] I'd say, you know, like some of my senior classes in engineering was just going a little deeper on stuff I already knew.

[109] Like I could have left it after three years and been just fine.

[110] But I do think the work I did did help me, you know, be an engineer.

[111] But then the problems I saw, you know, I worked on after I graduated college.

[112] Like, in school, most of the problems you're given, there's a known answer because they're in a book, you know, and you're in a room with 20 people and they're doing the same stuff.

[113] When you're an engineer working in the company, they never have two people the same thing to do because that's a waste of money, right?

[114] And when you start engineering, you're given relatively small tasks by, you know, your manager or supervisor.

[115] But as you go along at some point, depends on again, on who you are, you're working on stuff that you don't even know how to deal with.

[116] You know, there's no answer in the book.

[117] So it's, but it's not like physics, right?

[118] Like, physicists are a funny bunch.

[119] I realized this the other day that physicists, they're supposed to work on stuff that's unsolved.

[120] Whereas engineers, you know, there's a big repertoire of engineering, and then it's reduction in practice.

[121] And then the world's complicated.

[122] So you, you know, when you go build a new bridge that's never been built before, it's not like bridges are unsolved problem.

[123] This particular bridge hasn't been solved before.

[124] You know, maybe unique challenges to it.

[125] But it's not like physics where you're looking for an unknown particle or, you know, there's a pretty big dividing line between engineering and pure science.

[126] Engineers typically work in the names where there's many, many knowns and the unknowns are problems of the combination of, you know, reality, you know, complexity.

[127] Whereas physics, in principle they're working on stuff that's fundamentally on it.

[128] As soon as it's known, they have to move on because then it's engineering.

[129] Like this is, you know, translate the unknown into engineering and engineering applies known concepts to unknown problems.

[130] Okay, so you, okay, so now you, you went from digital to AMD and you learned how to design microprocessor.

[131] So at AMD, you worked on the K -8.

[132] Yeah.

[133] And at that point, AMD was losing ground to Intel.

[134] Yes.

[135] And so how did you fix that?

[136] So basically, Dirkmeyer and I were co -architects of EV6, the third alpha chip.

[137] He left about, he left a digital TAMD about a year before I did.

[138] He started the K7 project.

[139] When I joined, I started the K -8 project and then helped and then work with him significantly on K7 as well.

[140] And how do we do it?

[141] Yeah, so those were 64 -bit chips that you guys designed to compete with the Intel chips that had dominated the home computer market at that point.

[142] Well, so there's a funny thing, which is, like, at some level, building fast computers isn't that part, right?

[143] So you have to have a goal.

[144] So a lot of designers, so they have a design.

[145] And then the easiest thing for the next one is to go look at that design and make a like 10 % better or 20 % better, right?

[146] But every one of those designs has limitations built into it.

[147] Like, it's sort of like, if you buy a two -bedroom house, you can add one bedroom.

[148] You can't add eight bedrooms, right?

[149] If you want an eight -bedroom house, you have to build a different kind of house.

[150] Right.

[151] So every design you built has kind of a, you know, a range that it can play.

[152] And, you know, you build the first one and you know you can make some improvement.

[153] but at some point the improvements don't really help that much, right?

[154] And so AMD, they had a design called K5, which, for complicated reasons, didn't work out that well, and they lost ground to Intel.

[155] Before that, they had literally the 386 and 46 AMD copied Intel's designs.

[156] They were clone manufacturer.

[157] The K5 was their first design, and it didn't work out that good.

[158] And then they bought a company called NextGen, which had K6, which is an okay design, but it wasn't competitive against Intel.

[159] and then K -7, Dirk was the chief architect of, and he designed a computer that was competitive and the head of Intel.

[160] And some of that came from our work at Digital on UD5 and UB6.

[161] Turk worked on UB4 as well.

[162] And some of it was just saying in this day, we have this many transistors because you get more transistors every generation.

[163] So you can basically imagine you're building the house, suddenly you have way more bricks and way bigger steel beams.

[164] So your idea about what to build has to scale with that.

[165] And then K7 was a 32 -bit chip, and then K -8 was a 64 -bit chip.

[166] You know, somewhat related to that, as it turned out, but also it was built to be bigger.

[167] And what I did is I wrote the performance model.

[168] I came up with the basic architecture, and I started to organize a team around building it.

[169] And while we were doing that, we also wrote the thing called hypertransport spec, which became the basis of essentially all modern server computers or what's called two socket servers.

[170] We wrote that in 98 and 2002, or 2022, they're still building them that way.

[171] And when you say you wrote it, what does that mean?

[172] What does the process of writing that entail?

[173] What is it that you're writing and how do you do that?

[174] I'm dyslexic.

[175] So I wrote a complete, you know, protocol spec about how two computer chips talk to each other.

[176] And 18 pages, right, which is relatively terse.

[177] And there's a couple of pictures, and, you know, computer protocols are pretty straightforward.

[178] There's a command.

[179] There's the address you're talking to.

[180] There's the data you're moving.

[181] There's some protocol bits that tell you how to exchange commands, right?

[182] And then Dirk took the spec and said, you mind if I slush it out a little bit?

[183] And three days later, he sent me a 50 -page version of it, which clarified all the little bullets.

[184] And then that specification we literally used to build the interface between.

[185] K -8 ships.

[186] Right.

[187] So there's a couple levels of design.

[188] What sort of impact did that have, what sort of impact did that have on the, on the broader world?

[189] What's the significant?

[190] It's very difficult for non -engineers to understand any of this.

[191] It's so underground.

[192] AMD market share and server went from 0 % to 35%, which was a huge impact to the business.

[193] And it became essentially the standard because apparently Intel had a version of that, but it didn't go to market.

[194] But after Opturan came out to market, Intel built a similar version, similar protocol about how to connect a small number of processors together with that kind of interconnect.

[195] And then that, let's say, design framework became standard in the industry.

[196] So if you go into a Google data center and you pull it out, there'll be two sockets with an interconnect between the two of them, and each socket will have memory attached to it.

[197] And they call it a 2P server or two processor server.

[198] And it had a really big impact.

[199] We didn't do it because we thought it was going to have a big impact.

[200] We did it because we thought it was a better way to build computers.

[201] And at A &D, we were somewhat resource constrained.

[202] So we couldn't build a thing that looked like a big IBM server.

[203] So we built was basically a small server with the minimal amount of interconnect between it.

[204] So it was a little bit of creativity by constraint, Steve Jobs line.

[205] And what function does those servers have, again, in the broader world?

[206] What are they doing now for people?

[207] Well, it's basically the entire cloud.

[208] It's all Google, all Amazon, all Facebook, all Microsoft, Azure.

[209] But here's the interesting thing.

[210] When we built them, the big server guy, servers used to be backclones like this big, with multiple CPU slots, multiple memory cards, multiple I .O. slots.

[211] And the server manufacturer thought the server was oriented around the back.

[212] So IBM, HD, Dell, they all turned to.

[213] down.

[214] But all the little started at the time like Google were using PCs as low -cost servers.

[215] And we made this, basically, you could take a PC board instead of putting one computer on it, you could put two, which radically saved the money.

[216] So when A &D made those kinds of servers, it was a way lower entry point for server class technology.

[217] And the little startups that the Pine used it, and then over 15 years disrupted all the big server manufacturers.

[218] So it's, you know, it's one of those, I couldn't say we planned it, like the constraints that we had a target market.

[219] We didn't know that it was going to become essentially how servers we built, you know, for 20 odd years, but it happens.

[220] After AMD, you went to Apple, you worked on the A4 through seven processors.

[221] Well, I was, I worked at two startups that did processors for networking, Sidebite and T .Semi.

[222] and that was probably about five or six years.

[223] And then I joined Apple in 2008.

[224] So I guess I was, no, it must have been eight years.

[225] I was AMD 98, 98, 99, 2000.

[226] And then I worked at startups for about eight years.

[227] And then I went to Apple.

[228] Yeah, and I worked on mobile processors.

[229] And so tell me about those chips and what you did at Apple.

[230] The first of the funny part is I had some friends that were working in Apple, and they wouldn't tell me what I was going to work on.

[231] So when I interviewed there, they said, oh, you should have come here.

[232] It would be fun.

[233] And I didn't actually know what I was going to work on.

[234] They had a group called Platform Architecture run by a guy named Mike Colbert, who was like the unofficial CEO of Apple.

[235] He worked for Steve Jobs.

[236] And he had a group of architects that, you know, looked at what Apple was doing and figured out what they should do next.

[237] And I worked on a MacBook Air definition, like I wrote the power management spec.

[238] and did some other architecture work, which ultimately was an end -dity chip called MCP -8 -9.

[239] And then I was one of the chief architects of, you know, four generations of SOCs, what's called A -4, 8 -5, A -6, A -7.

[240] And we did a lot of stuff there because, but the division was, you know, mobile phones.

[241] And SOCs are what?

[242] Yeah, system -on -a -chip.

[243] Oh, yes.

[244] So to pack a computer into a phone, you have a little piece of silicon about that big.

[245] and all the components, the CPU, the GPU, the I .O. are all on the same chip.

[246] And when they first started building phone chips, they were considered to be very slow, low -cost, very integrated chips.

[247] And we thought, if you looked ahead, because technology shrinks about every two years, and about six or eight years, we'd have enough transistors on a phone chip, that would be more powerful than a PC at the top.

[248] So we started architecting, computers, interconnects, and other functions so that when we had enough transistors, we could literally have, you know, a high -end desktop in a phone.

[249] And Apple's DNA is, you create the product that kills your current product.

[250] Can you create the product that?

[251] So every company has a great product, and they worry about competitors coming in and kill it.

[252] And Apple wanted to be the first to kill their own products.

[253] So Steve Jobs thought phones and tablets would replace PCs.

[254] And he wanted to be the first to do it.

[255] He didn't want somebody else to do it to him.

[256] Did you know Jobs?

[257] No. I've seen him a couple times.

[258] I said hi to him twice.

[259] I felt like I knew him pretty well.

[260] Everybody at Apple did.

[261] Like when Steve wanted something done, everybody knew the next day.

[262] My boss, Mike, talked to him every single day, multiple times sometimes.

[263] He said, Joe, we'd walk in Mike's office.

[264] and he'd be holding the phone out like this.

[265] He goes, Steve, he's pest.

[266] We're like, yeah, so what?

[267] But Mike could translate what Steve wanted into engineering stuff.

[268] And Steve trusted Mike a lot.

[269] And like he could translate to vision into engineering.

[270] And Steve's judgment on stuff like this is spectacular.

[271] So, and did you have any sense?

[272] Do you have any sense of why that is?

[273] I mean, Jobs was famously, obviously originated Apple and then was famously brought back into save them when they were in danger of extinction and then, in fact, did seem to save them.

[274] And you never know when you hear about these things from the outside, how much of that is sort of a mythologization of a person and how much of it is, you know, this person was really singular and unique.

[275] Yeah, definitely singular and unique.

[276] And your psychological parlance, as we talked about, he would be considered high in openness and disagreeableness.

[277] right and I think negative emotionality like he was a very difficult person but a solution to you know things could go really bad and being disagreeable was I'm going to make it as great as possible and he was willing to take the risk for that you know his public persona was very well practiced Mike used to say the worst the practice for the you know Apple keynotes the better they would go off.

[278] Like he was throwing iPhones, you know, at one of the iPhone, you know, pre -launch practices because nothing was right.

[279] But then when he showed up and, you know, his persona of, you know, technical explainer, let's say, you know, that was very real.

[280] That's what he wanted to portray.

[281] We believed every single bit of it.

[282] You could tell.

[283] So, you know, any engineers, when I joined Apple, I watched some of his early keynotes when he came Apple and changed the Macs, you know, it's inspiring.

[284] But, you know, it's also super tough, right?

[285] Because he went into a company that was very dysfunctional, had a whole bunch of, you know, engineering groups doing basically random stuff.

[286] Let's say senior managers who felt like they owned their product lines and knew what they were doing.

[287] And, you know, Steve wanted them to do what he wanted them to do and they didn't want to do it.

[288] And I'm pretty sure he cleaned housepiece early.

[289] And he famously reduced the product lines and, you know, who knows how many products is like four.

[290] You know, there was consumer and professional.

[291] So do you think it was that disagreeableness?

[292] I mean, we hear all the time now in the modern world about the necessity for empathy and so forth.

[293] And that's the agreeableness dimension.

[294] And you're making the claim that Jobs was low in agreeableness and that he was able to kill off malfunctioning projects.

[295] And that's not exactly a nice thing to do.

[296] So imagine if you go to a room full people.

[297] People have dedicated their life in the last five years of their work, five years of their careers, building products that you can sell.

[298] And you say, we have to do something completely different.

[299] And everybody's, you know, every day as an engineer you're working on something, you embrace it, you love it, care about.

[300] Like, engineers are very emotional people somewhere in their pointy little souls, right?

[301] So, but if it's not working out for whatever reason, you have to do something different.

[302] And if you listen to everybody, you'll never change anything, right?

[303] It's difficult to get people reoriented.

[304] Now, another line you came out of where it came from is you run fastest when you're running towards something and away from something.

[305] Yeah, that was from animal experimental literature.

[306] If you threaten a rat and offer to reward simultaneously, it will run faster towards the reward than if you just reward it.

[307] Because you get all your motivational systems on board that way.

[308] Yeah, yeah.

[309] So Steve was very good at the vision.

[310] We are going to build this beautiful computer.

[311] And you better goddamn build it now or you're going to die.

[312] So that was his, okay, so the openness, the openness.

[313] That's the creativity dimension.

[314] That gives him the vision.

[315] He's extroverted.

[316] Can he communicate enthusiastically?

[317] He can certainly put on the act.

[318] I have no idea if he was an expert or not.

[319] I never saw him be extroverted in any of that.

[320] central setting.

[321] You know, I've seen him walk around.

[322] Like I said, we used to see him in the cafeteria.

[323] He was visible on the Apple campus even until his last days.

[324] He didn't like to be bothered.

[325] Like he didn't go up to Steve and say, hey, Steve, how's it going?

[326] Right.

[327] Well, that would be reflective of basic disagreeableness too, right?

[328] You know, it's hard with, it's hard.

[329] Sometimes people can communicate very effectively, communicate a vision because they are high in openness.

[330] Extroverted people are enthusiastic and assertive.

[331] So they tend to be, verbally dominant and can inspire people because they generate a lot of positive emotion, but that can be mimicked by openness.

[332] So you can also remember, like Steve was part of Pixar and very much part of Hollywood and, you know, creating movies and creating personas and characters and archetypes.

[333] So he was super well grounded in how that stuff works and what works and doesn't work about it.

[334] Right.

[335] And he had an unerring eye for beauty and elegance.

[336] And he cared about it.

[337] And he would fight for that.

[338] And that's hard.

[339] It's very hard to fight for beauty and elegance.

[340] And I suspect it's particularly hard.

[341] Perhaps maybe I'm wrong about this.

[342] But I would think that would be a hard sell to at least a subset of engineers.

[343] Yeah.

[344] So another, this is explained to my boss at Tesla was, I worked both for Elon and for a guy named Doug Field.

[345] And Doug said there's this, no, there's this productivity graph versus order.

[346] So at the origin is zero productivity and chaos.

[347] Right.

[348] And then as you add order to your design methods, your productivity will go up.

[349] And what happens with engineers is they understand as they get better processes, they get better trained, they get better working together, every single thing that makes the whole organization more orderly improves productivity.

[350] Unfortunately, that peaks at some point.

[351] And then too much order, productivity goes down.

[352] And so then as I say, any idiot can see, you should be at the peak, you know, enough order to really be effective, but not so much order you grind to the halt.

[353] But why can't you stay there?

[354] And the reason is, is once order takes over the organization, it's unstoppable.

[355] Right?

[356] It feels good.

[357] You get even better at doing what you're doing.

[358] You get even more organized.

[359] You micromanage your time even better.

[360] You crows out all of creativity.

[361] You're not open to change.

[362] A whole bunch of bad things happen.

[363] You shut out the disorderly people.

[364] who actually know how to make a change and do something creative, and the organization dies.

[365] So successful...

[366] Do you think the jobs was conscientious as well?

[367] Do you know, like, was he in early?

[368] Was he working 18 -hour days?

[369] I know he was up in the middle of the night because he called Michael on that stuff.

[370] Well, the point is both Steve and Elon were counterforces to order, right?

[371] You have to be really strong to avoid the organization getting captured by order.

[372] Well, order also has its remarkable error of moral virtue, right?

[373] Because it's pure and it's efficient.

[374] Everything about it feels good, you know.

[375] But it's like alcohol.

[376] The first drink feels, right?

[377] The second feels okay.

[378] The third one, not that good.

[379] But, you know, you keep remembering what the first one did.

[380] So you drank it.

[381] Right.

[382] There's lots and lots of processes, you know, where some is good, too much is bad.

[383] But the counterforce, the more is weak.

[384] And that's the thing that, you know, so Steve was interesting because he was simultaneously super creatives and had visions, which could inspire people.

[385] But he also prevented the company from being over -organized and preventing him for doing what he was doing.

[386] And that's hard because people, like I said, they get committed to what they're doing.

[387] Yeah, well, it's an open question.

[388] Like, imagine that the creative process has a productive component and then a culling, component and the productive component looks like it's associated with openness, but what the culling component is is open to question.

[389] And it does seem to me that at least upon occasion, it's, it's low agreeableness.

[390] It's the ability to say, no, we're going to dispense with that and to not let anything stand in the face of that decision, which would also include often human compassion.

[391] Yeah, and people have different approaches to it.

[392] Like jobs would call things so they weren't beautiful or they weren't great.

[393] You know, Elon Musk was famous for getting the first principles and really understanding their fundamentally and culling from like a standpoint of knowledge.

[394] Yeah, and you've asked me like what makes an engineer great.

[395] Like so you have to have the will to creativity.

[396] Like now there's lots of engineering jobs that aren't creative.

[397] Like you need a skill set, you can exercise the skill set.

[398] But if you're going to build new things, you need to be creative.

[399] But you also have to have to have a, filter good enough to figure out what's actually good and bad.

[400] Like, I know a lot of really creative engineers and they find a new thing that's excited and go down the rabbit hall on it and they, you know, they can work on it for six months and nothing to show for it.

[401] So you have to have that conscientiousness.

[402] I don't know if it's conscientiousness, this agreeable, this, you know, that taste on how...

[403] Well, the conscientiousness would keep you working in the direction that you've chosen and doing that diligently and orderly.

[404] The low agreeableness, well, that's the, that's the open question because agreeableness is such a complicated dimension.

[405] There's obvious disadvantages and advantages at every point on the distribution.

[406] I mean, disagreeable people are often harder to work with because they don't care much about your feelings.

[407] But one thing I've noted about working with disagreeable people is you always know what they're thinking.

[408] And if you want someone to tell you what's stupid and wrong, they're perfectly willing to do that.

[409] Yeah, I used to watch, I used to wonder, so Dirk Meyer was a disagreeable manager.

[410] But he could tell you what was wrong with what you were doing in the way.

[411] You would go, okay.

[412] Like he was very unemotional about it.

[413] Like he'd go, Jim, I really like this and this.

[414] But this isn't working for shit.

[415] Like, what are we going to do about it?

[416] And you just would just be offshut to matter of facts.

[417] I would say when I was younger, I was a lot less disagreeable.

[418] You know, I'm fairly open minded.

[419] And, you know, I like the creating new stuff, kind of stuff, things.

[420] But then I saw enough things fail over the years because we didn't.

[421] make the, you know, let's say the hard choices about something.

[422] And then, you know, you hate to work on something for two years and have to go away because at some point you realize you're doing a couple wrong things.

[423] And you didn't do something about it when you could.

[424] And so as a, you know, as a manager and a senior leader, I'm somewhat famously disagreeable.

[425] Part of it's an act to get people to move.

[426] And part of it's, you know, my beliefs that I can't have people dedicate themselves to doing bad things for very long because it'll it'll bite us.

[427] Yeah, well, we've talked a little bit about this too, about the moral dilemma between agreeableness and conscientiousness.

[428] They're both virtues.

[429] Agreeableness seems to me to govern short -term intimate relationships like that between a mother and a child, and it involves very careful attention to the emotional reactions of another person and the optimization of those in the short term, but consciousness.

[430] againstness looks like a longer term virtue and they come into conflict at some point because sometimes they come into conflict in the midterm yes right yeah it's you know could just be you know how our brains see the future but it's like you know if you're managing the group and you have to fire somebody it's hard right but do you want to fire five people now or everybody later like once you've internalized that and taking responsibility for that decision then making you know management leadership position choices is always hard, but it's so much better to make them and then succeed, then it is to fail because you couldn't make the hard cause.

[431] Yeah, well, and it isn't obvious at all who's got the upper hand, you know, someone who fires early out of necessity, but is accurate and looking carefully or someone who, you know, is willing to let people drag on.

[432] I'll give you, I'll give you two counter examples of that.

[433] So Jack Walsh in this book, straight from the gut, a weird thing.

[434] He said, once you have a doubt on somebody, you never act fast enough, which, you know, it took me years to really believe that.

[435] And then the other weird one is people say, hey, I have this organization of 100 people and there's five, five people are not working out, but I'm not sure who they are.

[436] So I'm going to be really careful because I don't want to accidentally fire a good person.

[437] Right.

[438] That makes sense, right?

[439] You've got five bad people, you know, maybe you figure out how two or three of them are, but there's this other group of five or ten, you're not sure which ones are the wrong ones.

[440] Here's the sad truth.

[441] There's a lot of people in the world.

[442] You're better firing too many than too few.

[443] And how did you come to terms with that emotionally?

[444] I mean, look, we have a mutual friend who fires people with quite great regularity, and I've talked to him, and he scores very high in disagreeableness.

[445] And I talked to him about firing, which he's done a lot of.

[446] And he was actually quite positive about it.

[447] He said, I don't fire anyone who I don't think is causing.

[448] more trouble than preventing.

[449] And so by firing the person that I'm firing, I'm actually doing a very large number of people, including potentially that person of favor.

[450] It didn't bother him.

[451] But he was temperamentally wired that way, I would say.

[452] But I would say, you know, digital equipment went bankrupt because they had bad people who didn't fire.

[453] I've seen many groups fail because they couldn't clean house.

[454] Right.

[455] And the impact on, you know, the greater good equation, easy.

[456] You want to save 90 people or, you know, lose 100.

[457] So that's true.

[458] The thing that took me a while to realize that the world needs shaking up all over the place and individuals do, right?

[459] A lot of people who are not doing too good, they need a wake -up call.

[460] You give them a bad review and they kind of shrug and they're like, what are you going to do about it?

[461] You know, it's like a spoiled kid, nothing, right?

[462] But when they actually get fired, they really have to do some soul searching.

[463] And then the fact that if you're doing something good, there's always a queue outside the door and more people.

[464] Now, here's another way to think about it.

[465] Take a group of 100 people and rank them from top to bottom.

[466] Human beings, by the way, are really good at this.

[467] You have four managers in the group, except for the manager's individual friends.

[468] They'll tend to rank the 100 people the same way.

[469] I've done this experiment many times.

[470] So we're really good at ranking.

[471] And there's a little bit of what are you ranking for?

[472] You're ranking for creativity, productivity, conscientiousness.

[473] But if you set the criteria right, people rank pretty well.

[474] If you have 100 people in your group and there's 50 people outside, the distribution of those 50 people is around the average of the team, right?

[475] So there's this idea that you fire the bottom 10 % of a team because the random people you hired will be better on average than the bottom 10 % of your team.

[476] Right?

[477] It's just math.

[478] Unfriendly statistics.

[479] But yes, I get the argument.

[480] Problem is that every company that does that, first it gets game because managers hire bottom 10 percenters.

[481] So when they get to fire the 10 percent, they don't have to fire their friends.

[482] Right.

[483] And it also really is hard on morale.

[484] Like people bond.

[485] And there may be people in the bottom 10 percent of your group that are the social glue of the organization.

[486] So you may be inadvertently taking out the stuff that makes the team.

[487] work.

[488] Right.

[489] Well, that's a measurement error too, right?

[490] It means that your criteria for competence aren't broad enough.

[491] Yeah.

[492] That's tough.

[493] That's a good point.

[494] Maybe you're ranking a little wrong, but impact on morale is high.

[495] Teams, generally speaking, Rory Reed was CEO of A &D when I joined, and we had a big layoff, which we had to do because we were running out of money.

[496] We're broke.

[497] And when we all just settled, we landed on just the right amount of people for the money we had.

[498] And you basically read us to riot act.

[499] He said, you said, guys, teams have to grow.

[500] When you cut, you always cut further and then you grow.

[501] People aren't happy unless they're growing, right?

[502] It's like when you prune bushes and stuff.

[503] You don't prune the bush to where you want the bush.

[504] You prune the bush past that point, so it grows out and it looks nice.

[505] Things have to grow.

[506] It's really an amazing dynamic.

[507] Yeah, well, and it's never.

[508] it's never clear how much death there is involved in growth.

[509] And the pruning analogy is exactly that.

[510] And this is harsh stuff, obviously, but you're looking at one collapse or another.

[511] It's right.

[512] That's the thing.

[513] It's not harsh because it's beautiful when you prune your bush and the growth back beautifully.

[514] It's great when you rebuild an organization that's really strong and powerful because you made the right calls, right?

[515] Like, this isn't just negative stuff.

[516] It's hard stuff to do that great.

[517] something really great.

[518] Like, when I joined AMD in, what was it, 2013 or something, like, they had two product lines, you know, Bulldozer and Jaguar, and they were both failing.

[519] And I had, I, I can't, I canceled both products.

[520] Okay.

[521] And so what was the human cost of that?

[522] I would say both to you and, and all sorts of the people that were involved.

[523] Well, I did, I did the math on it.

[524] It's like, you know, we needed to be building eight bedroom houses and we're trying to add six bedrooms to a two -bedroom house in one case.

[525] It wasn't never going to work.

[526] Yeah, so you saw that as doomed to failure.

[527] And the other one was structurally screwed up.

[528] It seemed to be the right ballpark for the performance we should get, but the way it was engineered and built was sort of like, you know, you let the plumber do the architecting and the house looked like shit.

[529] And it was difficult.

[530] You know, for a complicated reason, technical reasons, there was no path out of where they were.

[531] And when I realized I had to cancel them, yeah it was sleepless nights here we are we had we had revenue on that we had people committed to it people really liked it when i canceled it um especially on the jaguar team a significant number people quit because they were angry about it um it was some pretty big organizations there was some management we had to let go uh the best architect at the time well one of the best architects at amd was he really was my way or the highway and he was you know he he could not communicate what he was doing so i let him go which is a strange thing to do.

[532] So if he'd ranked the organization, you were ranking near the top, and I let one of those guys go because he wasn't ineffective working with the team.

[533] And how did you justify that to yourself?

[534] And how did you check yourself against stupidity and ignorance and, you know, self -interest and how did you know that what you were doing was right?

[535] Well, I'm a little lower in conscientious than they should be for a senior leader, first of all.

[536] So I knew this wasn't going to work.

[537] So you're in a space.

[538] The direction I'm going is not going to work.

[539] So you know how mosquitoes work?

[540] Mosquitoes are fun.

[541] So they detect two things.

[542] They detect water, vapor, and carbon dioxide.

[543] So mosquitoes will fly along in a direction as long as the water vapor and carbon dioxide are staying the same or going up.

[544] But as soon as it starts to go down, they change direction in a random direction.

[545] Right.

[546] And within a couple of turns, they're aiming right for a mammal.

[547] They can fight.

[548] It's gotten colorful.

[549] So if you know you're going in the wrong direction, a change in direction can maybe is just as bad, but there's some chance it's good, especially you're somewhat smart and you have some experience.

[550] Right.

[551] So there was a whole bunch of things.

[552] So even a random move is better than no move if the outcome is certain failure.

[553] And so that is some justification for taking a risk.

[554] Now, there's an infinite number of fail directions, but you're somewhat in.

[555] formed, right?

[556] And then the other problem is, like, when you build a house, it has a foundation.

[557] Once the foundation is built, it's very difficult to change the top of the house a lot.

[558] Like, if you have a foundation for a two -story house, it's hard to make it into an eight -story house.

[559] So when we cancel those projects, we consciously reset some design methodologies, you know, some team organizations, some leadership, some let's say, you know, we said we're going to, have the best in class, leadership, design methodology, and some of the architectural tools.

[560] We're just going to take those as given.

[561] Now that the land has been cleared, we had the opportunity to go back to do that.

[562] And it was interesting.

[563] And the design teams, it turns out there was some very good pieces, you know, in the two processors they had.

[564] But they weren't working together organically like they should.

[565] And say the framework of the design wasn't big enough.

[566] And then the tools over the years have involved into lots of little local improvements, but it wasn't really the right tool set.

[567] Now, AMD leapt forward when you did this.

[568] And they were the only competitor to Intel in a realistic sense.

[569] And so these actions on your part were part of what made that company thrive and kept competition within the microprocessor world.

[570] So these didn't have, these decisions didn't have trivial outcomes.

[571] Oh, no, it had really great outcomes.

[572] And there was, the really cool thing was, you know, when we did that, we didn't really bring in outsiders.

[573] Like, that Zen design was entirely based on people who worked at AMD at the time.

[574] Right.

[575] So what we needed was to clear the plate a little bit to reestablish some, you know, first principles about how we were doing things to have a better goal.

[576] There was a little bit of head knocking on getting, like, the methodology is straightened out.

[577] you know, I was, let's say, fairly disagreeable about how we were going to get through that because people kept saying, oh, it's too hard to do this.

[578] Well, is it any good?

[579] No. Well, if it's no good, it doesn't matter how hard it is, you have to do it, right?

[580] If you're going to drown, you don't go, well, a mile is too far to swim, so I'm just not going to swim.

[581] You're going to drown, right?

[582] If you're a mile offshore, you're drowning.

[583] Swimming a mile is the requirement, right?

[584] and I've explained that like a million different ways when something is pretty good you know the world's you know divided into three things things are good so you're happy and things are bad you fix it and then there's the middle ground where it's not great but it's not killing those are the ones where human beings have a really hard time improving right so by canceling the projects and declaring everything bad everything could be improved all right so what was it It would bug me about it.

[585] It was like, Jim, this wasn't that bad.

[586] Well, was it great?

[587] Was it going to win?

[588] All right, it's bad.

[589] I defined everything that's not great bad.

[590] If I moved on the continuum, everything 5 % or more away from great was bad, period.

[591] And how did you, like, how did you come to decide that was a good criteria?

[592] What's that?

[593] Why did you decide that was a good criteria that?

[594] All the time we were competing with Intel and on a whole bunch of metrics, like, like, literally their CPU had twice the frequency, twice the performance per clock.

[595] A whole bunch of metrics are so good.

[596] So we plotted them all.

[597] Like, you know, I had a whole bunch of data on this stuff, but also it's a mindset.

[598] Like if it wasn't best in class, like a computer has a whole bunch of things in it, there's something called a branch predictor, which predicts which way branches are going.

[599] There's was way better than ours.

[600] We could measure it.

[601] So we did.

[602] There's a thing called, you know, a memory system.

[603] Their memory system was way better.

[604] It was twice as fast.

[605] So it was like, well, we need to be within 10 % of them on everything or we're going to get our assets kick.

[606] And it would be really nice if we were better on some things.

[607] So we measured all their stuff.

[608] Like if you're going to compete, you know, like in basketball, you don't just players, you know, play yourself.

[609] You know, you try to see what the other teams are doing.

[610] Whether they good at, what are they bad at?

[611] You know, every coach is great at doing analysis of all the competition.

[612] And then, you know, you win two A's, you meet the competition with what they're good at, and then you have some secret plays that you're better at than surprising.

[613] Given recent SCOTUS wins, it feels like the pendulum may be swinging back to a time when the nuclear family was situated at the center of American life, where real conversation, learning, and growth began at home.

[614] President Ronald Reagan said in his farewell address that all great change in America begins around the dinner table.

[615] Well, all great meals in America begin with Good Ranchers.

[616] Good Ranchers cares deeply about providing families with steakhouse quality beef, chicken, and seafood meat at a reasonable price.

[617] Their mission is to bring people to the table, making those shared moments with your loved ones easy, accessible, and delicious.

[618] Good Ranchers ships 100 % American meat, born, raised and harvested in the U .S., right to your door.

[619] Plus, when you subscribe, your price is locked in for the life of your subscription.

[620] Great food creates great conversation, and great conversation.

[621] makes great change.

[622] So start bringing people back to the table with high quality American meat.

[623] Go to good ranchers .com slash Peterson to get $30 off your first order plus free shipping.

[624] That's good ranchers .com slash peterson.

[625] So you went from AMD to Tesla.

[626] So, and we talked a little bit about Elon Musk.

[627] I want to ask you about him and your experiences at Tesla, but also and what you did there.

[628] Sure.

[629] So let's talk about Elon Musk to begin with.

[630] Sure.

[631] Well, he's a pretty public guy.

[632] So I don't know that I can add that much.

[633] Well, kind of what you said about Steve Jobs, was it true?

[634] Yeah, that's mostly true.

[635] Like Elon's the real deal.

[636] He's a really good engineer.

[637] He has this belief that he can learn deeply about anything fast, and he's practiced it and does it.

[638] I see him do it.

[639] Like he gets into lots of details.

[640] He has done five impossible.

[641] things.

[642] Yeah, yeah, it's pretty respectable.

[643] And he likes the details and he has a good eye for it.

[644] Usually before a Tesla, when you do a technical presentation, usually done, you have some kind of methodology for resenting the problem to an executive.

[645] What's the problem?

[646] What's, you know, the background of that?

[647] Elon is solution person.

[648] If I don't like your solution, I don't care about the stupid problem and then your data, you know, just forget it.

[649] You know, what's the solution?

[650] There's a solution?

[651] Great.

[652] Great.

[653] Then everything else is backup.

[654] You know, what's the data behind that thing?

[655] Oh, we've got the data.

[656] What was the problem?

[657] Like, how did you figure this out?

[658] Oh, here was the original problem we found.

[659] Like, he likes, you know, he has a reverse order for most people.

[660] Most people tell you his story.

[661] It's a problem.

[662] Now, here's what we figured it out.

[663] Here's the background data.

[664] Here's all we built the solution.

[665] Here's the solution.

[666] Here's the next step.

[667] Like, that's a typical technical presentation.

[668] It's not a bad idea.

[669] Elon hated this.

[670] Like, stop with the bullshit.

[671] What's the goddamn solution?

[672] Can you give me an example of that?

[673] We had a problem with low -resolution camera images, right?

[674] And we were trying to improve, like, how the computer perceived roads in, like, low -light conditions.

[675] So we started with, well, here's the resolution of the camera.

[676] Here's the light sensitivity.

[677] Here's the images we're getting.

[678] And he was like, what the thing?

[679] You have a solution to that?

[680] well yeah we do we have the software that does this well page 12 why is it on page 12 page one of it in a really good place here's the old image here's the new image here's how we did it right that's what he wanted now partly is a high bad so did that what would would so what I'm wondering is what was effective about that was it that there was an ethos then that the most important thing that you had to bring forward wasn't a problem but it was a solution and so that people were striving constantly to generate and communicate solutions, which seems like a good strategy?

[681] Well, no, I think he's worried.

[682] So, engineers see the problem and they start investigating it.

[683] And then as you're investigating, you develop understanding.

[684] And your understanding the problem is good.

[685] Like, you would think that, right?

[686] You think that.

[687] Lots of people think that.

[688] Elon doesn't think that.

[689] Elon wants a solution, right?

[690] And if you're falling in love with your understanding and you're following in love with your little details that you're well that also feels like work you know you mentioned earlier that just because it's hard doesn't mean it's useful right and and focus you want to know that you are focused like crazy on the solution the best way to do that so then you have two states right if there's a problem there's two states you have a solution and you don't have a solution if you have a solution it's on page one we're all happy you well i've seen in you don't have a solution why the hell are you talking to me, why aren't you finding the goddamn solution?

[691] Yeah, yeah.

[692] I've seen in the software projects, other projects that I've been involved in too, that focusing on a solution, I think this is along the same lines as what you're discussing is, well, then you get a, you get to product a lot faster.

[693] It's like this thing has to exist and work.

[694] Maybe it won't solve.

[695] And the problem with the problem is that you can indefinitely investigate the problem and expand it and also that that feels like work, but it's not saleable.

[696] Yeah.

[697] Yeah, it puts in the forefront of your mind, you know, that constant, you know, you need to be creative pursuing the problem, but also make sure you're really on track of the solution and you're just not falling in love with the problems.

[698] People call it admiring the problem.

[699] Some engineers are great at admiring problems.

[700] Like, I've worked lots of people to come in with a 12 page presentation.

[701] And when I'm done, I'm like, did you guys just give me a whole bunch more data on the problem?

[702] He's like, yeah, we're really getting to the bottom of it.

[703] It's like, no, you're not.

[704] You're not getting anywhere.

[705] Well, the bottom of a problem is a solution because why would you just investigate the problem, right?

[706] I mean, your destination point is.

[707] They just keep getting deeper and deeper into problem admiration and nothing happens.

[708] Okay.

[709] I'd say three quarters of engineers would be perfectly happy to do that their whole life.

[710] Because, so you explain this to me, complex mastery behavior, right?

[711] So humans are very, you know, we like to learn, right we don't like to do dumb repetitive things right but we like to do things that are complicated that we've mastered that takes skill and you know you know insight but you can have complex mastery behavior just analyzing problems right definitely right and there's careers for that some people find they're really good at it they're a problem they're analysts they generate lots of data but if you're in charge of solving problems you know that that that that period needs to be focus, short, and concise and you need to move on to solutions, right?

[712] I think that's probably why I'm not so temperamentally fond of activists.

[713] Are they problem admirers?

[714] Well, that's what it looks like to me. It's like, well, this is one of the reasons I admire Elon Musk.

[715] It's like, well, you're concerned about the environment.

[716] Well, why don't you build an electric car then?

[717] Right.

[718] Well, so a large number of people, and this is my favorite thing I learned, and it was working with mechanics, engineers of Tesla.

[719] Because they think the world's made out of silly pudding.

[720] Right.

[721] They used to design, when we were building Model 3, they had designed a part, and then we'd joke about how they're going to make it.

[722] Are they going to see and see it like Millett?

[723] Or are they going to injection mold it, 3D print it, stamp it, make it with a hammer, you know, cut it out with scissors, you know, carve it out of a block.

[724] They had this cool machine that could carve 3D models out of clay.

[725] Like, it was funny.

[726] Like, so they could design things in their heads and on computer.

[727] and then go build any damn thing they want.

[728] If you ever look at it, complicated mechanical assembly, there'll be some extrude aluminum thing.

[729] It would be milled somewhere and then drilled and there's screws going through it.

[730] There'll be some little tab sticking off of it that holds another thing.

[731] Like, they can make stuff, right?

[732] They think they can make anything.

[733] Right.

[734] And there's a whole bunch of people in the world that don't think they can make anything.

[735] They don't, they think the world is what it is.

[736] I had a friend He had a rattle in his dashboard And he didn't know what to do about it And I was asking him where the rattle was And I was thinking that I was talking to him about like how the dashboard's made And he goes Oh I get it You think the dashboard's made of a whole bunch of parts That are put together some like I thought the dashboard It's just the dashboard Like he couldn't conceptualize it As there's his outer piece And there's inner brackets it's in your radio, and there's just these things.

[737] But mentally, I can't help but see the whole thing in 3D.

[738] And then I'm wondering which and which piece is loose and where it is.

[739] Right.

[740] And then how to fix it.

[741] Like, I'm not mechanical engineering creative, but I'm visual.

[742] Like, and for people who get stuck on activism as problem description, they don't think the world can change.

[743] Which at some level makes sense.

[744] you know, for human evolution, like, you know, it was pretty much the same for a million years.

[745] Like, it's weird how good we are at change.

[746] And my best theory on it is from zero to 20.

[747] Like, your brain is going through radical change because you're going from, you know, silly putty and not knowing much to being pretty smart.

[748] So you have to change and adapt really fast.

[749] And then humans are adapted to deal with each other and humans are fairly crafty.

[750] And, you know, you have to deal with that.

[751] But, but, you know, the lifestyle of most people from, you know, you know, say it's 30 deaths is fairly static.

[752] And, you know, so we have this funny capacity for learning rapidly, exponentially, and then dealing with slowly changing environments, but we're not naturally adapted to rapid change in the modern world is, well, especially in engineering, is rapid change.

[753] And so It was just so funny You never knew what they were up to Like one day Like they were working on the interior For the car And they made this crazy looking model Which kind of looked like a car But it turned out It was a thing you could move around And had the attachment points For all the interior parts So you could basically It looked like a weird skeleton But it had the attachment points That you could adjust And you could build all the interior parts and put a Tesla material together right in the middle of where the engineering desks were.

[754] It was really cool and let them go build it and think about it.

[755] And then it was in the CAD model and the computer, you could see it, but it didn't always work out in real life.

[756] Like we have a scale problem.

[757] When you look at something that's small, even if you scale it up perfectly when it's big, sometimes that's just what you thought and sometimes it doesn't work.

[758] And so, you know, you want to do, you know, computers like to change things fast, but like real scale models that you sit in and live in and get a human experience for it.

[759] And it was really fun for that to just show up and be like, holy cats.

[760] If I did a similar thing, we took all the electrical subsystems and motors and laid them out on two tables, two big tables, covered with all the electrical parts of a Model 3.

[761] We stared at it, and once you see them all together, it's crystal clear.

[762] That could be a lot better.

[763] Because, you know, there's three motors that look almost the same.

[764] Why isn't that one motor?

[765] There's these two parts that are completely separate assemblies, but if you build it together, you could have one thing, do both things really much more naturally at lighter weight.

[766] Right.

[767] And by laying that all out in front of you, you didn't have to do the mental work of representing that.

[768] You could do the mental work of seeing how all the parts interconnected and what might be.

[769] Yeah.

[770] Doug Clark is an architect I worked with when I was a kid.

[771] A digital call it the interocular traumatic test.

[772] When you look at it, doesn't bug you.

[773] And a lot of things, when you really lay them out like that, you go, oh, we're not doing this right.

[774] And Elon liked that kind of stuff.

[775] Like, you know, you're almost afraid to show them.

[776] Like, when you laid it all out, you looked at it and go, this is crazy.

[777] So it was like, do we show Elon or not?

[778] Because he'll look at it.

[779] I think this is crazy.

[780] Like, we built this car who did this.

[781] So I wanted to talk to you, too, to walk.

[782] I want to talk to you, like I'm saying.

[783] someone very stupid.

[784] And in this particular regard, I am, I really don't understand how computation works.

[785] And you're a microprocessor architect and you build computers.

[786] And so I'm, and I listened to a discussion you had earlier, earlier this month.

[787] And there was a lot of it I couldn't follow.

[788] I thought it might be helpful and interesting for you just to walk through for me and for my audience, how a computer actually works, what it does and how you build it and then what it would be like to design and to architect a microprocessor.

[789] Well, it's somewhere hard to describe, but there's a couple simple things.

[790] So let's start with the easiest thing.

[791] The computers have three components, really.

[792] Memory, programs, and input and output.

[793] Right?

[794] Those are the three basic things we always built.

[795] So memory is like the DRAM or the disk drive, place where you store data.

[796] And it's just stored.

[797] And it can have different representations.

[798] We currently use ones and zeros.

[799] So you can take any bit of information and describe it as a sequence of ones and zeros.

[800] And it's stored in silicon in either static and dynamic memories or on disk drives, which, you know, there's a couple of technologies for that.

[801] So does memory make sense to you?

[802] A place to store information.

[803] Well, it does, although I have some difficulty in understanding exactly how the transformation is undertaken to represent things in zeros and ones.

[804] I mean, I'll give you a simple example.

[805] So if you shine a light on a photo, a photo detective, right?

[806] So the light comes in, and it's a stream of photons, right?

[807] And the photo detector counts the photons, literally.

[808] So every photon that hits it or a couple photons hit it.

[809] They cause some electric charge to move, and that causes a circuit to wake up and say, I saw some photons.

[810] So say, you're trying to evaluate how strong that light is.

[811] So it can be anywhere from nothing to super intense, right?

[812] And then you might say, well, let's put that in a range of numbers from zero to a thousand.

[813] Right.

[814] And then you're trying to lay on it.

[815] And so you count the photons for, say, you know, a microsecond.

[816] And then you translate how many photons you counted into the number.

[817] Right.

[818] So, and so just imagine those as the light varies up and down, the count, the number coming out of your photo detector is varying between zero and a thousand.

[819] Right.

[820] And we use base 10, but you can translate that to binary, which is base two.

[821] And now you have.

[822] have ones and zero.

[823] So you've basically now translated a light ray optical information to a count.

[824] And so virtually everything can be represented by account, apparently.

[825] So just think of your computer, it has a camera, which is essentially doing just that.

[826] When you get different colors, but first you go through color filter, so you have a green filter and a red filter and a blue filter that's enough to represent the color spectrum and then you have a little little photon counting underneath that and it counts for a little while and then you ship out the number and you reset it and count again those of light varies you're getting that there's a pretty big grid like a modern camera is 12 million you know photo detectors in it actually 36 million because it's got different colors but well it depends on how they build it like they might have different pixels be different colors and then interpolate the colors Like a keyboard is really simple.

[827] So the fundamental issue to begin with is that you reduce everything to a count, and then you represent that count in base two, base two.

[828] It's base two, right?

[829] Zero and one.

[830] And you can do the same thing with sound.

[831] You can have a sound detector that basically counts no intensity of sound waves.

[832] And the keyboard, well, you have a little grid of keys.

[833] And when you push a key, it sends which key was that?

[834] That was key number 27.

[835] So now you can code.

[836] And it's either on or off.

[837] On or off, you know.

[838] So D might be count number 26, and F is 27, and G is 28.

[839] So everything gets encoded into a number.

[840] And then computer is like binary numbers, you know, for technical reasons, but that's not a big good.

[841] So memory, so input and output is the first thing.

[842] So the computer is built around the memory.

[843] So the input and output systems, so the input is all your photo detectors, sound detectors, keyboard detectors.

[844] and there's amazing numbers of sensors these days.

[845] You can detect gravitational waves, maybe.

[846] You can detect photons.

[847] You can detect electrical waves, sound waves.

[848] You can take temperature.

[849] You know, it's very varying.

[850] You turn all that stuff into a number, and then your input device writes that into memory.

[851] And memory stores information.

[852] You know, memory used to be small and expensive, and now it's big and cheap, but it's still just memory.

[853] And nothing happens to it in the memory.

[854] It just gets stored.

[855] If you look inside the computer, it's a big memory, and it's full of numbers.

[856] Now, the person who's writing the program is, you know, is telling, like, the input output.

[857] Like, here comes the input video stream.

[858] Put the video stream address, one million.

[859] So all the memories address.

[860] And the address typically starts in zero, and modern computer goes up to billions.

[861] Okay, so walk through that again, the addressing.

[862] So what's exactly the function of that?

[863] Well, you want to know where the memory is.

[864] Okay, right.

[865] You need to, okay, fine.

[866] So basically your phone probably, I don't know, has eight or 16 gigabytes of memory in it, maybe at four or eight.

[867] So billion, you know, eight billion bytes of information in there.

[868] So, and when you're designing your programs, you kind of lay out, well, here's our operating system is going.

[869] Here's what the input and output buffers are.

[870] Here's memory we're going to use to run some program.

[871] And all that's addressed.

[872] So you can think of the address is, it's just like a post office, right?

[873] So, you know, every house has a postal address, you know, it's a street address and then your house address.

[874] And so you can find every person.

[875] And the address corresponds to the physical location, in some sense, to the physical location of the, of the, and how are the zero and ones represented in the memory?

[876] It's literally a voltage that's either high or low, and zero is using ground, which is.

[877] is zero volts.

[878] And modern, you know, DRAM cells probably stored at 1 .1 volts.

[879] And, you know, in a DRAM cell, it's a capacitor that's holding electrons.

[880] So basically, when you store the cell and either you drain all the electrons out, so it's zero volts, or you put a bunch of electrons in so that it holds a one bolt.

[881] So it's literally a number of electrons in there.

[882] There's a couple ways to make memory cells.

[883] There's another way, which is called a bi -stable element where you have what's called cross -couple inverters, but that's too complicated to explain.

[884] And then the memories are usually built in rays, so there's an X, Y. So you take the number and you say, I'll take the bottom half of the number and figure out which row it's in, and the top half the number of which column it's in, and where the column in a row overlap, then I'll write my new date of a 1 and 0 in that spot.

[885] It's literally that simple.

[886] So if you look at a memory chip, you'll see this array of bits with little blocks on two edges.

[887] Usually, you know, one side's the row, one side's the column, and then the bottom is what they call the sense and you read it back out again.

[888] So a memory process is you activate the roll and column to a spot, which gives you the address of that bit, and then you drive the bit in and charge up or discharge that cell.

[889] And then it holds them.

[890] And it's super simple.

[891] You could build a memory with a pegboard.

[892] You could build a memory.

[893] I mean, literally did, you know, way back when there was something called Clure memory where they had, essentially the XY grid.

[894] And at each little place, there was a little magnetic bead, which when you put the current through, you could put the current in the same direction.

[895] You could make it be north to south and the opposite direction south and north.

[896] So the, you know, you basically remagnetized little beads.

[897] So there's lots of ways to make memory, but currently the really dense memories are called dynamic memories, where you literally put charge in there.

[898] And then there's fun stuff that happens, like flash cells, the cells got so small that the electrons, you know, from the quantum effects, pungle out occasionally.

[899] So you put 25, because there's some doubt about where the electron actually is.

[900] Yes, and sometimes it could literally jump out of the cell once it jumps out and it doesn't come back.

[901] So, so, so they got down to like 25 electrons in a cell.

[902] cell and they would wander off over a couple hours and you have to refresh them.

[903] So periodically, you go back and you read the data before too much of its escape and you write it back in.

[904] So it's called refreshing the memory.

[905] But DRAMs hold more charge of math.

[906] And then the flash guys figure out how to stack the cell.

[907] So modern flash chips, there's an X Y grid, but there's also a C dimension.

[908] They're like 256 layers thick now.

[909] So it's like a three -dimensional memory.

[910] but the simple thing still is it's a linear range of addresses where you put some data.

[911] Okay, so that's memory.

[912] So the next component is programs.

[913] This is the compute part.

[914] So a simple program, simple program is A equal B plus C, right?

[915] So the data at address A, so when you write the program, you tend to use what they call variable names, A, B, and C. But there's a tool called compiler, which will sign A is to say address 100, and B, the address 101, and C, the address 102, right?

[916] And then the computer, when it's running, says, do what I told you to do.

[917] So you see this program, A equal B plus C. So you get B, you get C, you add them together, you put it in A. and typically what happens is you have what's called a local memory or a register file.

[918] So you get the data from memory into the register file.

[919] You do whatever operation we're told to do, like ads, and then you put C -back in number.

[920] And what are the range of operations, or is that too broad a question?

[921] What are the fundamental operations apparently are arithmetic?

[922] I've done this.

[923] Like the number of operations that a computer does, like the instruction sets can have 100 or 500 or 1 ,000 different instructions.

[924] But the most common ones are load data from memory to the processor, or the program ones.

[925] Store memory back.

[926] Those are your first two instructions.

[927] And then add, subtract, multiply, divide, you know, clear, you know, very simple.

[928] It's written, you know, then there's what's called logical operators and or not.

[929] It's stunning to me conceptually thinking through this that, and computers which can produce whole worlds in some sense can do that as a consequence of zeros and ones and arithmetic operators.

[930] Sure.

[931] Well, your brain is doing something interesting like that.

[932] There's no magic to it.

[933] So the key to programs is abstraction layers, right?

[934] So at some low level, you know, like I understand computers from atoms up to operate, which is, you know, fairly broad range, but there's lots of people who can do that.

[935] Yeah, and I understand them like the surface of the keyboard.

[936] Yes.

[937] Yes, monkey with military helicopter, basically.

[938] Pete Bannon had an interview question.

[939] So, you know, computer scientists would say, tell me what happens when I move, when I type a key, right?

[940] Because you can talk all day about it.

[941] You know, because the key is at a position which encoded the number, which got sent into the memory.

[942] There's an interrupt delivered to the processor to say there's new data and then we go take a look at it.

[943] But you can describe that at many, many levels.

[944] Right.

[945] So it's a good place to start.

[946] It's not bad.

[947] As an interview question.

[948] As an interview question.

[949] Yeah, right.

[950] Some people, by the way, are stumped.

[951] They go to college and they can't tell you what happens.

[952] You know, the key is click, which is weird.

[953] So back to the computer.

[954] So it's the basic operations.

[955] That's subtract.

[956] divide, you know, clear, set the one, and or not XOR, you know, you can take a number and you can shift it around, you can mask it.

[957] And so different architectures.

[958] How were those operators discovered, Jim?

[959] I mean, I know there's their arithmetic operators.

[960] And is that just the question of how was their arithmetic discovered?

[961] But I mean, there's a logic.

[962] Way after math.

[963] So computers, at some level, they're doing arithmetic.

[964] Like, it's not very sophisticated.

[965] And I'll get to a little more complicated version of this.

[966] And by the time we invented computers, people had pretty good idea of number theory.

[967] They'd figured out that base 10 was just one of the bases.

[968] You could have two, three, four, five, six, seven, eight.

[969] People had, the philosophers have worked out what logic is.

[970] You know, if this is true and this is true, then this is true, this is true, or this is true.

[971] Like, the logical operators are real.

[972] Like, there's a whole bunch, there was a whole bunch of them.

[973] Yeah, it's the real, it's the realism of them that's stunning to me. Yes.

[974] So there's the basic operator set, and then there's something called control flow.

[975] So computers typically, they put a program like add, you know, A equal B plus C, you know, D equals E plus F, F equals E plus A, you know.

[976] And you typically put that in what's called program memory, but it's just part of the memory of the computer.

[977] And you have a program calendar, which, I don't know what's called calendar, but the thing that points at the next instruction to execute.

[978] And it's the fault thing is do this instruction and then do the one right after it.

[979] That became like the way computers were built.

[980] That's an arbitrary choice, by the way.

[981] You could have every instruction tell you where to get the next instruction.

[982] There's a bunch of things you can do.

[983] But for simplicity, people said, this piece of memory has programs in it, start at the first instruction and then do the next one and the next one the program counters but that's not good enough because then you would just start at the first one and then you go to the end of memory and be done so there's something called control flow so a program called sorry called control flow so imagine you wanted to add up a list of ten numbers so your first instruction says I'm on the first instruction and then you say the sum equals the current sum plus the next number, increment the counter of how many instructions I had, increment, count by one, and then test.

[984] Is the counter equal to 10?

[985] If yes, keep going straight.

[986] If no, go back to get the next number.

[987] Right.

[988] So you created a little loop.

[989] Right.

[990] So, and it turns out computer scientists and invented a whole bunch of kind of loop, what they call control flow constraints.

[991] Do this while X is true.

[992] Do this until the counter gets to a number.

[993] Right.

[994] So there's, so you can create little, you know, basically subprograms in the program.

[995] Right.

[996] And then there's a couple, you can, you can test like, hey, I need to decide if this is a dog or a cat.

[997] So if it's a one, go look at here.

[998] If it's a zero, go look at that.

[999] Right.

[1000] So that's a conditional branch with loop branches.

[1001] And then somebody famously invented subroutines.

[1002] You notice how he was writing the program, he'd write this little routine, but it would be used a bunch of different times.

[1003] So rather than put it in the code in multiple times, it was like define a word.

[1004] And then whenever I need that use that word, I don't have to put the whole definition for the word.

[1005] I just put the word.

[1006] So subroutine is like a local definition.

[1007] of something or a local computation that's used multiple times.

[1008] So your top level program might be go to the subroutine that counts up numbers, thumbs up numbers, and come back.

[1009] Now go to the subroutine that checks whether it's your bank balance or not, come back.

[1010] So the program, this becomes sequential operation, control flow like doing loops.

[1011] Let's say, do this until something's done.

[1012] and then conditional branches that says, depending on value, do this or that, and then subroutines to do something atomic.

[1013] And that's essentially all the program.

[1014] Operations, loops, conditional branches, and subroutine.

[1015] That's it.

[1016] Now, why can computers construct worlds?

[1017] So I still remember when, so if you look at your screen, you know, the computer in front of you, probably has two or four million pixels on it, it seems like a lot, right?

[1018] And when they first started, you know, television when they lit up screens, you know, they were scanning a little electron beam across a phosphorescent surface and lighting and modulating the intensity of the electron gun to make the little fosters brighter little.

[1019] Right, and it was writing.

[1020] It writes one line at a time, wrote one line at a time at an incredibly fast rate by human standards, and we saw that as continual And then the phosphor was designed to decay at the rate, so by the time you came back to it, it just gotten a little dimmer, and then you wrote it with the next value, so it didn't flicker.

[1021] So your eye has some persistence.

[1022] So the electron hits it, it makes the phosphor light up with some photons of the right color, and then it slowly decays, and you scan down, and it gets back there and writes it again before it's too dim.

[1023] And so the screen on a phosphor -based television, is it analogous?

[1024] It's analogous in some sense to the binary representation?

[1025] The dot is gone off?

[1026] That's entirely analog.

[1027] So it's digitized in the sense, it's discrete, I'd say, in the sense that each little pixel, you can see the little phosphors on the screen.

[1028] Right.

[1029] Especially notice that when they went to color TVs because they have a red, green, and blue thing in there.

[1030] Right, right.

[1031] And they would hit them.

[1032] And they're essentially either on or off?

[1033] Well, they have a range so that that beam is a variable intensity.

[1034] Right.

[1035] So no modern computers work differently.

[1036] So the screen in front of you has a little, it literally has an X, Y grid, and it can address each one of those things, right?

[1037] So you don't shoot a beam at it anymore.

[1038] You have an X, Y, you decode.

[1039] You know, it's almost like the screen looks like a big flat memory, but instead of storing ones and zeros, it's storing color.

[1040] But they have the same kind of decay property, and you write the new color and there's a bunch of stuff.

[1041] Now, here's the wild thing.

[1042] Computers are now so fast, you can run a 10 ,000 line program for every single pixel on that screen.

[1043] So what does that imply?

[1044] Well, it turns out for a whole bunch of reasons, like if you want to make something look really good on the screen, so the world's relatively continuous, right?

[1045] So if you look at it, there's all this light reflecting around, there's all these things going on, there's no little pixels in the surface of your table, right?

[1046] To make a discrete grid look that way, you have to, you know, combine the colors of, you have to do a whole bunch of stuff.

[1047] You have to pretend you're shining lights on it.

[1048] You have to, you know, like there's a reflection from one surface to the next one.

[1049] And it turns out when you have thousands of instructions per pixel, you can start to make those pixels look realistic.

[1050] right the operations and you go look in the pixel program like it looks so beautiful you think that's incredible but if you look in the pixel program it's load the data into the register add it to a number tested against the number subtract something or something called clipping like make sure the pixel doesn't get brighter than this and dimmer than that it's all simple operations like there's nothing in the computer that's like do a you know pixel operation, right?

[1051] Well, there may be a subroutine, you name that, but underneath it, it's just the same old stuff.

[1052] Computers always do, load, store, add, subtract, multiply, divide, branch.

[1053] So did we, okay, so how far have we got, I'm listening to so many things, I'm having a hard time keeping track of the order.

[1054] You mentioned earlier that computers consist of four elements, I believe that's what you said.

[1055] Memory?

[1056] Yep.

[1057] input and output.

[1058] And compute.

[1059] Okay, three.

[1060] So I was counting input and output separately, but okay.

[1061] And have we gone through all three of them?

[1062] Yep.

[1063] Okay.

[1064] Okay.

[1065] So memory is just the place to store bits.

[1066] Yep.

[1067] Input and output is typically a way to, you know, it depends on what you're doing.

[1068] You might just send bits one place to another, but it might also be, you could say input, you know, input and output in the computer and sensors are slightly different things.

[1069] Like sensors, you know, turn analog real world signals into bits.

[1070] Into digital.

[1071] Right.

[1072] And then programs basically transform the data in some way.

[1073] And programs are basically, you know, operations like adds and track divide, and then branches that either let you do it loops or make decisions.

[1074] And then the hardware to let you do subroutines to break the program into pieces.

[1075] And that's pretty much it.

[1076] So to some degree, you take the world, you transform it into on -off or yes, no. billions of those, and then you manipulate the yes and noes or the zeros and ones, and that can produce almost any sort of phenomenon that you can imagine.

[1077] Yes and no, it's not a very good, you know, ones and zeros is better because then it's a, it's a mathematical representation, you know, a digital representation of an analog reality.

[1078] Something like that.

[1079] And is the analog reality, analog all the way down, or is it digital at the bottom?

[1080] quantum.

[1081] Quantum at the bottom.

[1082] So there's something called the fine constant, which makes the universe look discreet, but it's a very, very small number.

[1083] Right.

[1084] So, and there's a fun fact, which is...

[1085] Is that the plank length?

[1086] Is that associated with the plank length?

[1087] Yeah.

[1088] And that's the smallest possible length, I believe.

[1089] Like the mass of the universe is 10 to the 40th, and the plank length is 10 to the minus 40th.

[1090] And there's a, there's a physics thread about the, the mystery of why those things are 10 to the 40s and 10 to the 40s.

[1091] All right.

[1092] So let's move from that to, I'm going to ask you, these are the questions.

[1093] I just want to say so.

[1094] Yep.

[1095] The thing that makes computers do what they do is abstraction layers.

[1096] So at the bottom, there's atoms.

[1097] So there's engineers you know how to put atoms together in a way that makes switches, which we call transistors, right?

[1098] So, and those guys are expert at that stuff, right?

[1099] And they just, they can operate at that level.

[1100] Then there's another thing where you take multiple transistors together and you basically wake what's called logic gates, which literally do the ans and ores and inversions, right?

[1101] And then that's an abstraction layer.

[1102] We call it, you know, the physical design library or something like that.

[1103] And then people take those and they make them up into adders and subtractors and multipliers.

[1104] This is a well understood bullion mass. So how do you add two binary numbers?

[1105] So you make those.

[1106] And then there's another abstraction layer that says, are I going to take multiple operation units and put them together to make, you know, part of the computer, right?

[1107] And then you make, there's a bunch of those blocks.

[1108] And then that thing runs a program very simply.

[1109] And there's a small number of people who write programs at the low level, but then there's people who use what's called libraries where they, you know, they're doing some higher level program and so they're going to do a matrix multiplying and do this and that, but they don't actually write that low -level code.

[1110] So there's a stack of abstractions, and when something gets too complicated, you split the abstraction layer into two things.

[1111] It used to be when people wrote a program, there's a program called a compiler that translated your C -program or four -term program into the low -level instructions.

[1112] But it turns out there's too many languages up here, and there's too many instructions here, So now they translated from the high -level language into an intermediate representation, which is sort of, let's say, a generic program.

[1113] And then there's another thing that translates the intermediate representation and the specific computer you have.

[1114] But that just keeps going higher and higher.

[1115] Like a lot of programmers, they use frameworks that can do amazing things.

[1116] Like you could literally a later program that says, search the internet for a picture of a cat, sword by color, output to my printer.

[1117] Like, there's a language or that's a program.

[1118] Search the internet.

[1119] Holy cow, that runs a trillion lines of code on 100 ,000 computers.

[1120] Find a cat.

[1121] That's a really expensive.

[1122] That's a really complicated program.

[1123] So how much of the radical increase in computation power is a consequence of hardware transformation and how much of it is a consequence of the increasing density, let's say, of these abstraction layers?

[1124] Well, so this is where, you know, there's a really creative tension.

[1125] or dynamic interplay.

[1126] So when computers first started, they were so slow, you ran really simple programs.

[1127] A equal B plus C times D, right?

[1128] And we've been going up the math hierarchy.

[1129] So then you could run a program that did what's called, you know, matrix math, like or linear algebra, systems of big equations, and then matrices, and then more complicated ones.

[1130] So as computational power went up, you could dedicate more and more stuff to, you know, of that kind of computation.

[1131] And then similar thing happened on abstraction layers.

[1132] It used to be, if you bought a million dollar computer, you hand wrote every line of code because you didn't want to waste time on the computer with like overhead.

[1133] But today, you know, that million dollar computer costs 10 cents.

[1134] You don't really care how many cycles you use, you know, parsing a cat video or something.

[1135] And so the computation capacity let the abstractions at the programming level increase a lot.

[1136] So somebody who made it, had a graph about how many bytes is this take to store the letter A?

[1137] Like, it used to be one.

[1138] And then word for Windows, it's like 10 kilobytes per letter.

[1139] Because the letter has a font, it has a color, has a shadow.

[1140] You know, there's a whole bunch of, you know, and that's fine.

[1141] Like, the computer with a million dollars for, you know, a thousand bytes of memory, you wouldn't store letter A like that, you put it in one byte.

[1142] But now you have gigabytes and terabytes of storage.

[1143] You probably already know that there are data brokers out there selling your internet data off to companies who want to serve you a targeted ad.

[1144] But you might be surprised to learn that they're also selling your information to the Department of Homeland Security and the IRS.

[1145] Mask your digital footprint and protect yourself with ExpressVPN.

[1146] One of the easiest ways for brokers to aggregate data and tie it back to you is through your devices, unique IP address.

[1147] But when you're connected to ExpressVPN, your IP address is hidden, making it much more difficult for data brokers to identify you.

[1148] ExpressVPN also encrypts 100 % of network traffic to keep your data safe from hackers on public Wi -Fi.

[1149] You can download ExpressVPN on all your devices, your phone, your computer, even your home Wi -Fi router.

[1150] Just tap one button and you're protected.

[1151] Make sure your online activity and data is protected with the best VPN money can buy.

[1152] Visit expressvPN .com slash Jordan right now and get three extra months free.

[1153] That's expressvPN .com slash Jordan.

[1154] Okay, so you walk us through the basics of computation.

[1155] Now, can you shed some light on?

[1156] Like, I don't understand what you do as a computer architect.

[1157] Like, when you go to work, when you're working on a project, what is it that you're actually involved in doing?

[1158] I make ads go faster.

[1159] So I'm a fairly low -level engineer, you know, low -level in terms of the abstract engineers.

[1160] Like, I understand the higher ones.

[1161] But, you know, I talk to the people who make transistors and N -gates and or gates.

[1162] And they talk to the people who know the atoms.

[1163] Right.

[1164] So I hardly ever talk to the atom people, but I know something about atoms.

[1165] So I build the, you know, when I'm architect and stuff, the functional units and then how they operate together at the low level that runs programs.

[1166] So I don't write programs.

[1167] I'm an architect of the computer that runs programs.

[1168] And then it used to be you could look at a computer and you know how program works.

[1169] you run the first line, the second line.

[1170] If there's a branch, the branch unit, then branch would have that in.

[1171] It's such an instruction, load the data, do the operation, if there's a branch, execute the branch, if necessary, change the program counter.

[1172] So, you know, people, you know, there was a period of time where computers had like five stages in it, and each one of them could say that's the branch, that's the fetch unit, that's the load unit, that's the ad unit.

[1173] And that's the branch of it.

[1174] But monitoring computers are more complicated than this, right?

[1175] Because computers like that would do one instruction every five cycles.

[1176] And modern computers, the fastest one I know about it, just doing 10 instructions, 10 instructions for cycle in parallel.

[1177] Right.

[1178] And this is difficult.

[1179] So the best way to - unpack that, unpack that.

[1180] So if you write a program since you write, right, when you write, you write, you write linear narrative.

[1181] You write a sentence that makes sense, followed by another sentence.

[1182] And so as you're writing along, sometimes the one sentence defines the meaning of the next sentence, right?

[1183] And then group it in the paragraphs.

[1184] You might call those subroutines, right?

[1185] And sometimes the paragraphs have to be ordered, and sometimes the paragraphs, the order doesn't matter.

[1186] Right.

[1187] So programs are written by human beings and they're written in the same linear narrative.

[1188] So if you want to go faster than parsing the instructions one at a time in order, you have to do some analysis to say, all right, I got two sentences.

[1189] Are they dependent or not?

[1190] If they're dependent, I do them in order.

[1191] If they're not dependent, I can do them in parallel or any order.

[1192] Right?

[1193] And you start, so with the modern computers, when they're reading the programs out, they're analyzing the dependencies and deciding what can happen in order, what has to happen in order for correct understanding, and what can be reordered.

[1194] And then it turns out there's many places where you say, if there's an error, go here, but there's hardly ever an error.

[1195] And you can predict that really well.

[1196] So you say, I'm going, you're reading along, and you say, there's a point where I'm not sure which should I read the next sentence or should I jump from the next paragraph.

[1197] So a modern computer predicts that.

[1198] It doesn't wait for you to fully understand all the sentences up to that point, so you know exactly where to read to.

[1199] So imagine, so now you're reading this book, and you're reading sentences in dependency order, which means you haven't, so you get to a branch, and you haven't read all the sentences before that and understood them.

[1200] So you don't know where to read the next paragraph for the next chapter.

[1201] But we predict what's going to happen, and we just keep on going.

[1202] And how does that tie into the process of designing the...

[1203] So the goal of modern computers is to go fast.

[1204] Well, let me say there's three kinds of computers.

[1205] There's computers that run very simple programs in order, right?

[1206] They just do exactly what you told them to do.

[1207] And they tend to be small and simple.

[1208] But they're so small and simple, you can make a chip with 1 ,000 of those computers.

[1209] computers on.

[1210] So when you build a GPU that does a little program for every pixel on your screen, each one of those pixels gets its own program.

[1211] It's very simple.

[1212] But you sort of say the first thousand pixels you run on these thousand computers.

[1213] So like a modern GPU has currently like six or eight thousand processors in it.

[1214] And they literally, you do the first 6 ,000 pixels and then the next 6 ,000 pixels and the next 6 ,000 pixels.

[1215] And they do that fast enough that you can run a fairly big program on every pixel on the screen for every screen refresh on.

[1216] So you have simple computers that do stuff in order, right?

[1217] And then you have, let's say, computers that are designed to run complicated long programs as fast as possible, right?

[1218] And that's where you parse the instructions carefully and you figure out what order you can do them in and when possibly you reorder it.

[1219] And the reason to reorder it is because if this doesn't depend on this, I can do in parallel.

[1220] Now I can do two things at a time.

[1221] The next thing I can predict that I can do it in parallel.

[1222] I can do three things.

[1223] And, you know, like I said, the computer in your desktop is probably doing three to five things at a time, and the best I know it is 10.

[1224] Right.

[1225] And that's because, and there's other sophisticated predictors in there.

[1226] So to do that, you have to fetch large groups of instructions at the time.

[1227] You have to figure out where the, like, the sentence boundaries are, figure out if they're dependent or not, figure out if you can predict where the next instructions are coming from when you hit branches.

[1228] And it turns out that's fairly complicated.

[1229] The difference between a little computer that does, let's say, one instruction in the time, a complicated one that does 10 instructions at times, it's 100 times more complicated.

[1230] Right.

[1231] And from a, what's the best way to do lots of instructions?

[1232] Complicated computers are not efficient.

[1233] But there's so many applications where people care how fast it is.

[1234] So when you're like clicking on your web page, you want that to come up as fast as possible.

[1235] So the part of it's that's, let's say, what's it called, you know, the logic of the web page.

[1236] It's probably a serial narrative written by a human being.

[1237] So you have to, you run that on a complicated computer that, you know, does it out of order and predicts what to do as fast as possible.

[1238] but when you render the screen itself, that runs on large numbers of simple computers to make all the pixel.

[1239] Right.

[1240] And then there's a third kind of computer, which we're starting to invent, which is AI computers.

[1241] And that's what you're working on now.

[1242] Yes.

[1243] For Tens Torrent.

[1244] Yeah.

[1245] And there's a really good talk by Andre Carpathie called Software 2 .0.

[1246] So the first two kinds of computers, simple computers and complex out -of -order computers.

[1247] they're running programs written by humans, right?

[1248] And if you look at the code, it's literally a declarative statement about operations and where to go.

[1249] And it's serial.

[1250] It's a linear narrative.

[1251] The different thing about AI computers is you use data to train the weights and neural networks to get you the desired result.

[1252] so instead of the programs are no longer written by humans now it turns out there's components of the AI stack that are written by humans but at a high level you use data to train them so they have a big neural network and you want to detect cats so you put a cat picture into the network when you start training and the output is gibberish and you compare gibberish to what a cat is and you calculate the difference in what the network said versus the desired result, which is the word cat.

[1253] And then they do something called back propagation, which is mathematically sophisticated, but essentially take the error and partition it across the layers of the network such that you've sort of bumped each neuron a little closer to saying cat next time by taking the bigger at the end, distributing it across that's called back propagation.

[1254] And then you put another cat in, and if you have the right size network and the right training methods, After you show the network a million cats, when you put a cat in, it reliably says cat, and when you put a picture, it's not a cat, it reliably says it's not a cat.

[1255] Right.

[1256] And you never wrote any code that had anything to do with cats.

[1257] And can you understand what it is that the computer is doing now that it's recognizing cats?

[1258] A little bit.

[1259] So people for years worked on visual computing, and they were trying to detect things like cats.

[1260] right and cats have a whole bunch of artifacts they have round eyes they have pointed ears and fluffy hair so you could detect it is called feature detection you would say this will be a cat if I see the following colors the following amount of fluffiness the following number you know two point of ears not three one or two round eyes depending on the view right so you could write code and the problem with that is well now the cat has an arbitrary orientation so you have to you do your feature detect on the picture and the features have to search the whole image and you have to rotate around, you know, and it's sort of, and every single thing you want to detect, you have to write a unique program for it.

[1261] You're done with cats, now you go to dogs.

[1262] And then what about the dog that has suddenly pointy ears?

[1263] These dogs have round ears and cats have pointy ears.

[1264] You know, so it was sort of endless thing.

[1265] Right, right.

[1266] Same thing with speed.

[1267] Endless, endless detail by detail, construction.

[1268] I had a friend who worked on speech recognition years ago.

[1269] So you break speech into, you know, the phonings, so you can see those, and then they have frequency characteristics, and you can differentiate vowels from consonants.

[1270] So those people working on speech were doing a whole bunch of analysis of analog waveforms that sound.

[1271] And they were making some progress, but it never really worked.

[1272] and then they train neural networks by you put the word in and you have what's called supervised learning so you play at language where you know what all the words are and you keep telling the network how to correct and with like a billion samples and a big enough neural network it can recognize speech just fine and if you train it with a broad variety of accents it can it's it can work across accidents.

[1273] And then it turns out the bigger they made these networks and more information they could put it.

[1274] And then on the cat one specifically, they found so when they first they first had a neural network cracked the cat problem.

[1275] I forget it was like 50 layers deep.

[1276] And if you looked in the layers, you could see that it was detecting between years and eyes.

[1277] But it was also detecting a lot of other things.

[1278] And some things we don't Yeah, well, if we see the back end of a cat walking away, we still know it's a cat.

[1279] And it pretty much lacks eyes and point of ears from that perspective.

[1280] If you take an object, like in light, right, take a phone, you can project the phone onto a flat surface.

[1281] Say, that's a projection, right?

[1282] And as you move it around, you get different.

[1283] It's a shadow.

[1284] But think of it as a projection.

[1285] Mm -hmm.

[1286] Right.

[1287] So that's a projection of a light source on a flat plane.

[1288] It's a fairly simple projection.

[1289] But what if you had a light shaped like a cat and you signed that on the phone?

[1290] What was the projection look like?

[1291] And it turns out mathematically there's an arbitrary number of projections.

[1292] You can, like we think of projections in three dimensions because we're three -dimensional preachers.

[1293] Right.

[1294] But there can be lots of projections.

[1295] And then you can have the projection project on another plane.

[1296] So that the neural networks are doing is they're achieving out all the details of what that is.

[1297] And some of the projection planes give you what's called, you know, size and variance or rotation variance.

[1298] Like you could recognize a cat number, which is pointing.

[1299] Like your brain is a little specialized.

[1300] Like the faces.

[1301] Right.

[1302] It likes them to be vertical.

[1303] Right side up.

[1304] Yeah.

[1305] But with a little bit of work, you can, you can recognize an upside on face before.

[1306] unless you have a problem.

[1307] Okay, so we could do two things here.

[1308] We could either talk about your, no, let's go into your, you were an engineer and then you were a manager.

[1309] And you've worked in lots of companies, some of which were incredibly creative, some of which were thriving to an incredible degree, and some of which were collapsing and, and irreparable.

[1310] So what have you learned about what makes companies work?

[1311] And more importantly, what have you learned?

[1312] about what makes them not work and maybe what do you do then?

[1313] Sure.

[1314] Well, that's a fun question.

[1315] Well, first of all, there's like I've noticed and many people have noticed this is not just me that people, like in engineering fields, people kind of bucket towards, you know, technical people and management people.

[1316] And it's not that there aren't good technical managers or there's not good managers or technical people who can manage.

[1317] Right, but that's an intersection of two skills, say.

[1318] Yeah, but generally speaking, most people are one or the, and it's like when you wake up in the morning, or do you want to solve a problem or do you want to organize the problem?

[1319] Like, are you worried about your schedule and your headcounts and how things are getting done and did you hit the milestones or are you working on technical problem?

[1320] And people, and in engineering fields, it's often there's the fellow track with the technical leadership position or the director of VP track of management leadership, right?

[1321] So I'm a technical person.

[1322] But you took on management rules repeatedly?

[1323] Well, I did because I found out that if you're generally speaking the top of the organization is the manager, the VP, and as a technical person, the matter how you go, you're an advisor for that person.

[1324] And I decided consciously after I of Apple and I was going to be a VP and have everybody work for me because then I can right so then my skill set is somewhat unusual and I'm not the only one obviously but I decided to you know get on the management track so I could build the computers I wanted because sometimes when I wasn't the leader of the group some managers at some point would decide they own the next decision and they would make some random decision I'd be grumpy about it and there's something I could do about it because people work for them, not for me. So that's, you know, it was a conscious thing.

[1325] And I hired a consultant, Ben Kat Rao, who helped me reframe how I approach this.

[1326] Now, I'm still a technical person, but I found that, it turns out there's a whole bunch of really good technical managers that I like to work with.

[1327] I'd like to organize stuff.

[1328] And I would say, you know, I maintain my openness and low conscientiousness and disagreeable behavior.

[1329] I have people who work for me, work on my team, or work with people that manage better.

[1330] So even though I've been a manager, you know, at EMD was 2 ,400 people total at the end, and Intel was 10 ,000.

[1331] My staff is, you know, 15 or 20 people.

[1332] And usually half of them are real managers and half of them are technical leaders.

[1333] That's how I founders tend to be technical people, but people working for them are non -technical.

[1334] Or they're stronger on the management side than the technical side.

[1335] But for everybody, you need to decide who you are.

[1336] Like, I had a great technical manager at A &D, and one day, he was a little mess because I was looking to the, you know, a couple of really the technical heavyweights.

[1337] It's all the problem.

[1338] He said, you know, I'm pretty technical.

[1339] I said, yeah.

[1340] I know.

[1341] I said, are you technical compared to Jim and to Barb?

[1342] And he goes, I guess not really.

[1343] He said, I know.

[1344] I really like, you know, what I want you to do is you're running this project.

[1345] You have 150 people working for you.

[1346] You make all the technical decisions you can.

[1347] But when it's out of your wheelhouse, we got serious experts.

[1348] And you have two choices.

[1349] You can call them or I can call them.

[1350] And he later told me, he said, I found that it was a lot better when I called him.

[1351] And when you call them, a successful thing.

[1352] And he was technical.

[1353] He was really good at making good decisions, but he wasn't the strongest technical person in the group.

[1354] So that's the first thing is, you know, figure out who you are.

[1355] I've seen a lot of people fail in engineering because at some point they think, I'm technical, but I want to get on the management track, but they're bored by management, and they don't have a plan to deal with it.

[1356] Yeah, well, you weren't bored by management.

[1357] Well, so.

[1358] So I joke that I decided to see the organization of the computer architecture problem and treat people.

[1359] Well, that's exactly what I was going to ask.

[1360] What transformation did you have to undertake to?

[1361] So one of them was, what do I have to do to be effective?

[1362] Right.

[1363] So that's, you know, I hate to work on failed projects.

[1364] Right.

[1365] And then the next was the organizational problem itself is an architectural problem.

[1366] And then I tapped, you know, for myself.

[1367] Well, I'm a funny kind of, if something has a solution and it's being competently driven, I'm not that interested in it.

[1368] I like problems.

[1369] And so in a big organization, there's a million problems, and I start sorting them by priority and then solving some of them are handing them out to the right people.

[1370] So there's a whole bunch of technical work to do on that.

[1371] And then I'm fairly good at skill assessing people who are technical, either for management or technical positions.

[1372] and then, you know, giving them work.

[1373] I like autonomy and management.

[1374] So if somebody's competent, they can do it and they understand it.

[1375] Ben Gett gave me a bunch of books to read.

[1376] And one of the frameworks is goals, organization, contract, and teamwork.

[1377] Or capabilities, I guess, we usually solve for that.

[1378] So is the goal super clear, do we have the capability to solve the problem?

[1379] Is there a contract between me and the group of?

[1380] doing it so they know what to do and what their goals you know box are in right and do they have the you know the um is the organization that like a lot of times you know there's a joke that start up start with a problem and build an organization supported but on the second third system the organization defines the problem rather than the problem defining the organization and then it breaks up yeah then the organization becomes the problem yes yeah they constrain the problem and become the problem.

[1381] Well, we had a number of discussions while you were doing this about ethics.

[1382] And I mean, you said that you go, you look at the problems.

[1383] Well, that's hard, right?

[1384] Because you have to know enough to know what the problems are.

[1385] Then you have to be willing to look at the problems.

[1386] Well, then you prioritize them.

[1387] Like you skipped over that very quickly.

[1388] But all of that's extraordinarily difficult, I would say, both cognitively and emotionally.

[1389] Sometimes it isn't.

[1390] Sometimes it isn't.

[1391] Like when I joined A &P, the CPUs are less than half as fast as the competition.

[1392] And they had no plan to catch up.

[1393] So that wasn't that hard hard.

[1394] No, but what would be hard there, I would presume, is figuring out how it could be that such an obvious problem had gone undetected and unsolved.

[1395] And then...

[1396] No, actually, one of their architects, when I was working at Apple, told me that they believed that CPU performance had plateaued.

[1397] It wasn't going to get any faster.

[1398] And they were going to work on adding features to the rest of the chip.

[1399] and then Intel came in and said, we think computers are going to get 5 or 10 % faster every year.

[1400] And they did it.

[1401] One had one goal, which is, you know, things slow down.

[1402] The other had a different goal.

[1403] Five or 10 % isn't a lot.

[1404] But you do that 10 years in a row and, you know, the other guys weren't.

[1405] So that wasn't that complicated.

[1406] Like Elon famously said, he tells everybody's secret plans and nobody believes them and then does them.

[1407] And they still don't believe them.

[1408] And then they're like, oh, shock.

[1409] So Intel publicly said they were going 5 or 10%.

[1410] faster every year and NAMB said, no, they're not.

[1411] You know?

[1412] And the results were at some point, the gap got bigger and bigger.

[1413] You know, the people NB were committed to their plan.

[1414] I don't know why.

[1415] It's interesting how these things get internalized and then you start, even when they, you know, at some point, you know how it is.

[1416] It's cognitive distance.

[1417] You say you're going to do something different, but you've learned how to do this other things really well.

[1418] You keep doing it.

[1419] Right.

[1420] And then.

[1421] Well, and you've been able to.

[1422] whole machinery around it.

[1423] Yes, exactly.

[1424] You know, they had a big machine that did all kinds of stuff that was perfectly used.

[1425] Right.

[1426] And good people doing it.

[1427] Like I said, we didn't hire any people to build then, but we did refactor, you know, reset the goals, refactor a whole bunch of the engineering.

[1428] Okay.

[1429] So, well, at AMD, you were successful twice.

[1430] And so, and the success was both building a chip that was competitive.

[1431] So you had to put together the teams to build the chip, but also to transform the internal structure of the company so that that became possible, and then also to communicate that to your, to your customers.

[1432] And so what's the problem set there?

[1433] I didn't communicate with the customers.

[1434] I, you know, because, you know, computer world performance cells.

[1435] Okay, so that's the first thing you brought to the table, performance cells.

[1436] And here, we're going to break that down.

[1437] Here's the measurements.

[1438] And there's lots of public benchmarks.

[1439] Like, everybody tries to gain it, but generally speaking, there's a really big community of computer users, and they know what they want, and they know what's fast.

[1440] Right.

[1441] And you know exactly how a computer works, so you can actually say, once you decide that what faster is better, does that work on all of the elements of design?

[1442] I mean, there's a little complicated, too, right?

[1443] There's certain things, like, you can make it faster, nobody would care.

[1444] It's like, yeah, there's some judgment calls in it, but it's not that complicated.

[1445] Like, you know, today on phones there's a thing called geek bench, and, you know, you get a number at the end.

[1446] You know, is your geek bench score 100 or 50?

[1447] 100s better.

[1448] Right, right, right.

[1449] The people who made the benchmark tried to pick the components of your phone experience such that the geek bench number represented whether the phone is faster or not.

[1450] And then whether you care or not's another question.

[1451] Like, for the current applications that they got twice as fast, you might not notice.

[1452] but as the computer gets faster, it's a new applications are possible.

[1453] And on the phone, where it's possible, it's great, where it's not possible, it feels slow and lagging.

[1454] So performance wins in, you know, different form factors like a notebook or a desktop or a phone have different amounts of power available.

[1455] They put it within the budget.

[1456] Okay, so you had a goal, you had the measurements in place, you decompose that into tasks, you assign competent people.

[1457] What psychological factors got in the way?

[1458] Like, how did you see?

[1459] All of them.

[1460] Yeah, fair enough.

[1461] All of it.

[1462] But what did you see specifically interfeit?

[1463] Once you have a good plan in place, that doesn't necessarily mean it's going to be implemented.

[1464] And so what are the mistakes that people make that you saw in large companies that doom the companies or that stop them from transforming internally?

[1465] So there's a couple of very separate problems.

[1466] When somebody with a good set of ideas says, I need to transform this place.

[1467] like there's there's are the goals proper right and then you want to say do I have the capability in the team to do it like like I worried when I went to A &D I wouldn't have enough experts in certain things to do it I'd have to go hire 50 people to fix it but turns out there was I did plenty of you know there was plenty of good people actually some really great people so I was like you know pretty quickly checked off the capability box and then you start wondering well why how long are we doing the right thing?

[1468] Well, the problem was the goals were wrong and then your organization was wrong.

[1469] Right.

[1470] And then generally speaking, if those aren't right, so to begin, so maybe to begin with, the goals weren't unreasonable and no one knew, but then across time, the fact that one set of goals was better than the other, the belief that computers weren't going to get much faster was a bad goal in the world where the competitor believed they were going to get a lot faster.

[1471] Yes, and could do it.

[1472] And that became incrementally worse across time to the point.

[1473] where it became cataclysmic.

[1474] Yeah.

[1475] So you got to get the goals right and you've got to establish where you have capabilities.

[1476] You know, those are the kind of fundamentals.

[1477] But then the organization built him is hard because somebody will tell you so -and -so is a great manager.

[1478] Well, is he or, you know, like a lot of times there's somebody that looks like a good manager, but you just have three people working with us.

[1479] It's really good.

[1480] The problem of that is when things are going well, the empty suit manager with his good people supporting him, they think they look like they're making lots of progress.

[1481] But when they run into hard problems and the technical guys don't want to do, they go to him and makes a random decision or does something dumb or doesn't believe him.

[1482] Like that happens a lot.

[1483] Like technically, the guy goes to the MPC manager and says, you know, I think this isn't working.

[1484] You need to change.

[1485] And he says, now we're fine.

[1486] We're just a way, right?

[1487] So you get these weaknesses in the organization, because you don't have skill level.

[1488] Like I said, I've worked with a lot of really good technical managers who know when they can make to the citizen and they know when they have to come to somebody who's more of an expert.

[1489] That's great.

[1490] And it turns out some people are so good at that.

[1491] They can operate way higher than you think because they're not technically strong, super good at translating and making judgment calls like that.

[1492] So you've got to start building your organization and then there's stuff about how do you build teams.

[1493] Like some groups are what they'd call functional, but all the people who do software in one group and all the people do hardware group and all the people do Adams in another group, right?

[1494] And then the managers, but if the thing you're building needs a little of all three of those things, you know, it's called, you know, functional organization versus product organization.

[1495] You might want a team with a couple of programmers, a couple of hardware people, a couple of atom people, and the same team.

[1496] So they're all, they all have one, on goal, as opposed to the functional group says, I'm making the best software group.

[1497] Was it the right thing for this product?

[1498] They go, I don't know.

[1499] I don't work on the product.

[1500] I work on software.

[1501] So I'm generally speaking, you know, product focus.

[1502] So if you only have like five of some discipline, you tend to make a little functional team like that.

[1503] And there's a couple things in computers on which are functional.

[1504] But generally speaking, I like product focus organization.

[1505] So everybody's like, they're all working.

[1506] working together on the same thing.

[1507] They may have different disciplines.

[1508] So, AMD.

[1509] Now, you've encountered all sorts of frustrations.

[1510] Sorry, you've encountered all sorts of frustrating situations when you've gone into companies where you're trying to put together a good product.

[1511] And so what have you seen, what do you see as particularly counterproductive?

[1512] And what have you learned, like, how to conduct yourself so that you can be successful?

[1513] Weak leadership, people who can't make the technical decisions they have to.

[1514] That's a big problem.

[1515] Functional organizations where people are optimizing for the function, not the product.

[1516] Bad goals is one of the worst things.

[1517] Some organizations have real capability gaps.

[1518] Like, you know, they think they have the right people, but they don't.

[1519] You know, some managers play favorites.

[1520] They think so -and -so is really good and they're not.

[1521] Yeah, so that's a real fun.

[1522] functional analysis, the company just can't do what it needs to do.

[1523] So, so, and we're still analyzing, here's a group, they're actually from, you know, some place, there's a belief that we're going to build this product that has to be a great product.

[1524] And how do you do the, you know, basic blocking tax and how to make that successful, right?

[1525] That's different than the malaise that overtakes big successful companies, which you can generically called bureaucratic capture.

[1526] Right.

[1527] That's a different problem.

[1528] Like a company that's bureaucratically captured will manifest all kinds of bad behavior in the organization and product development.

[1529] And then, you know, some big companies where the bureaucracy is taken over, there might still be groups that are really doing a great job making great products.

[1530] You know, so there's, you know, I think there's separate spaces and I understand both of them pretty well.

[1531] And again, you know, the way you solve big, complicated problems, you have some abstractions about what you're dealing with.

[1532] So, you know, a framework like goals, organization capability and contract is a super clear method for evaluating what the hell is going on and then making changes.

[1533] You know, very specific changes to that.

[1534] Coles are clear, you know, if you're not clear, nothing else matters, get the goals clear, right?

[1535] Capabilities, are they good?

[1536] You don't have very capabilities, nothing will say if you have to have the ability to do the job you're doing.

[1537] you know, does the organization serve the goals?

[1538] That's a big problem.

[1539] That's a painful one.

[1540] That's because that's when you start changing who works for who and what the boundaries are.

[1541] But you have to do it.

[1542] Okay, so let's tackle it this way then.

[1543] So you're going to pick someone who has optimal attributes to create and operate within a highly functional organization.

[1544] What are you looking for in that person?

[1545] And what's crucial?

[1546] Well, people are fairly diverse.

[1547] That's the funny thing.

[1548] So engineers need to have this will to create if they're technical leaders, let's say.

[1549] And then they have to have the discernment to make decisions about whether they're actually making progress with the goals or just wasting their time on something cute.

[1550] Right.

[1551] That's a thing.

[1552] Technical managers, you know, they need to know how to run a program.

[1553] They need to have a hire and fire.

[1554] They need to have a structure of work.

[1555] They need to know how to evaluate how long it's going to take, how to evaluate whether people are making progress.

[1556] There's a whole bunch of things, but then people have very different style.

[1557] Some people are very extroverted.

[1558] I worked with this woman.

[1559] She was great.

[1560] She would just have these team meetings and she would really get out there and energize the team.

[1561] And another guy in the same building was very low key and he would wander around and talk to people and have a really good sense of the team, like an introvert versus extrovert style.

[1562] But they both worked.

[1563] They were both very competent.

[1564] They were both, to me, you know, really good technical competency.

[1565] They weren't my technical leads, but they were technically competent and not to make the decisions and know when they had to punt the decision up.

[1566] So who do you not want, who do you not?

[1567] Okay, so, I mean, that kind of goes along with the management literature.

[1568] You see that you want people who are intelligent, especially for complex jobs so they can learn.

[1569] You want people who are conscientious because they work hard and they have integrity.

[1570] Then with the other dimensions, it looks like there's a fair bit of variability, although too much negative emotionality can be a problem.

[1571] I think that's because it's associated with depression and too much anxiety and so on.

[1572] But there's diversity in the other personality dimensions, and that might be task specific.

[1573] But what sort of person do you not want to work with?

[1574] Fakes.

[1575] There's lots of fakers out there.

[1576] You know, they have sales attributes.

[1577] They're, you know, extroverted agreeable.

[1578] You know, they want to say everything is good all the time.

[1579] They're not sufficiently concerned about disaster and digging in the stuff.

[1580] They may have some kind of narcissistic personality problem.

[1581] So they're imposterous.

[1582] They're mimicking competence.

[1583] Mimicking competence.

[1584] That's a problem.

[1585] There are people who literally...

[1586] They take credit from other people.

[1587] Yeah, I kind of put that in a separate boat, but there's people who...

[1588] Okay.

[1589] credit for the team I realized early on there's two kinds of managers and people get up and people get it down like as a manager I often tangled with the people I work for but I always took care of the people work for me but some other managers I had this one guy who looked at me I thought it was great and then I walked by a meeting he was having he was abusing his team and they hated them but I fired him because he always said the nice things to me and, you know, you get on those people.

[1590] So it's, yeah, there's a bunch of weird stuff that happens to management like that.

[1591] Like, you have to be excited.

[1592] Like, if you're a senior manager and a high tech thing, there's many people in the group that are smarter than you.

[1593] And you have to promote them and put that forward.

[1594] You can't be uncomfortable because somebody's smarter to them.

[1595] When I was in A &D, I had six senior fellows, I think they were all smart.

[1596] They weren't, they weren't as generalists or something, you know, and they didn't have my interest in, you know, an architecture of organization.

[1597] But, man, it was smart, super good.

[1598] I could talk to them all, you know, I could keep up with them sometimes.

[1599] But, you know, I was more than happy to promote them as smart guys.

[1600] And why were you confident enough, do you think, to allow you to be surrounded by people that you...

[1601] I grew up to do you...

[1602] Like, I'm above average smart, but I met people who were so smart.

[1603] Like, I knew Butler Lampson, he was, you know, a famous unrated IQ, and his wife was smarter.

[1604] It was a joke that he spoke at half Lampson because his wife was smart when he spoke really fast.

[1605] But I, at a fairly young age, I was competent in getting things done and work with people that were smarter to me. But they, they like my, you know, I'm an engineer and I build stuff, you know, the rocket science, they think it up and then they hope somebody would build it for them because they're off onto the next thing.

[1606] So that's, you know, a belief I have.

[1607] You know, it wasn't always easy.

[1608] I still remember working on EV5 and I went to the digital research lab and there's half of those in these super smart people.

[1609] And I started describing what I was doing and I would describe something for about two minutes and then they would spend five minutes taking another part and analyzing how it could be like way better.

[1610] And then they'd ask me the next question.

[1611] And after an hour of that, I felt like, oh, my God, it was just beating the death.

[1612] And they were like, this is great.

[1613] Jim.

[1614] I was like, you thought that was great?

[1615] They're like, yeah, we're glad you're doing it.

[1616] So, and I've always had that attitude since, you know.

[1617] But yeah, it's hard on some people when they realize how smart some people are.

[1618] So what?

[1619] I make up for it because I'm open -minded and I work my ass off for many years.

[1620] And then I've dived in lots of things.

[1621] And then, you know, I'm not afraid to ask dumb questions.

[1622] I, you know, like a lot of people protect, you're trying to project who they are.

[1623] because they don't ask right questions.

[1624] They don't learn it.

[1625] And I'm like, I don't understand what the hell is going on.

[1626] I've done that in the room of 50 people.

[1627] And they're like, well, we thought you should know.

[1628] It's like, well, I don't, but I'm not going to believe until I do.

[1629] And then they give all the information.

[1630] And I'm smarter than I used to be.

[1631] And so that takes a certain, you know, mental resilience.

[1632] And sometimes it's very hard on me. But, you know, again, it's sort of like, you know, do you fire the people you have to fire to save the group and save the product?

[1633] to make me save the company.

[1634] Yeah, then that good is really high.

[1635] And that's the right thing to do.

[1636] So exposing yourself is the right thing to do, ironically.

[1637] Now, it's hard in, you know.

[1638] Well, if you admit you're stupid, then sometimes you don't have to stay that way.

[1639] Yeah.

[1640] It's hard in some sick organization.

[1641] Sometimes it's not safe, it's exposure.

[1642] And I feel for people who are in places where they would really like to be more open and can't, because, you know, organizations, they get posted.

[1643] and bureaucratic or hard on people that are actually trying to do the right thing.

[1644] I totally understand that.

[1645] It takes a while, the psychological safety thing is it gets overused and gets a bad rap, but having an organization where it's actually safe to open your mouth and talk and ask questions and occasionally look stupid and fumble a little bit and have your peers like support you with that and be happy for you and learn stuff.

[1646] that's really important it's hard to do you know and so and there's great attention because you know as a leader you have to be disagreeable enough to do the hard things or still creating an environment where people can open up and do that and I would say I'm mixed I have mixed reviews on that topic because once I feel find things that are wrong and people are doing the wrong thing you know I have to get to the bottom and I'm going to I'm going to close with some people it's never happened to him before like the people haven't really taken what apart.

[1647] You know, they got A's in college and they got good reviews.

[1648] They rose to their Peter Principal and competence point.

[1649] All of a sudden, they're doing something they're over their head and they know what to do about.

[1650] They don't have a lot of practice.

[1651] So, yeah, it's a funny, it's a funny thing.

[1652] So I'm going to close with a question about your current venture.

[1653] You're now working with a company that does AI computing.

[1654] And what do you hope to do that you can talk about?

[1655] Well, so I was an investor in this company when it first started, the Bisha Project, who's the founder, worked with me at AMD.

[1656] And I always thought he was an especially smart guy.

[1657] And I liked his approach to building AI computation.

[1658] I'm really intrigued about computers programmed by data.

[1659] I think it's more like how our brains work.

[1660] our brains are really weird, right?

[1661] Because we think in this littering our narrative, we have those old voice in our head, but we know we have 10 billion neurons and you're collecting the way, you know, exchanging, you know, small amounts of brain transmitters and electrical pulses.

[1662] You know, it's bloody hilarious.

[1663] The gap between what a neuron looks like and what a thought looks like.

[1664] And so, and there's a really interesting opportunity to make big AI computers that are actually really programmable.

[1665] So one of the things we're doing is we're building the software stack that lets you build the neural networks you want and then program and get the results you expect reasonably well, as opposed to having a very large army of people tweaking it.

[1666] And so there's a bunch of architectural, interesting things to do.

[1667] And then it's a startup, which we have chips at work, we start production, we're going to start selling them.

[1668] You know, there's a whole bunch of work to do on how to engage with customers, and a lot of customers we're talking to are super smart.

[1669] There's all these, you know, AI Soper startups and with really smart people that have some problems that, you know, basically if the computers are a million times faster, it'd be easier to solve.

[1670] So there's like a huge capacity gap on what they want to do.

[1671] So participating, that's fun.

[1672] Like, I like that kind of thinking.

[1673] And your goal?

[1674] So you go into an organization, you have a goal for the chips?

[1675] What's your goal for this organization?

[1676] Oh, we're going to be successful.

[1677] selling AI computers so the origin to the people.

[1678] You know, significantly better performance, better programability and lower cost.

[1679] And there's a bunch of innovation work to do around that to make that really possible doable.

[1680] Like the AI field is relatively new.

[1681] The computers that run AI today are relatively clunky.

[1682] And to me, you know, need a lot of work and refinement so that, you know, from the idea that you want to express from the program, the writing to the result you want better and cleaner.

[1683] Okay, so one final question for anyone who's listening who would like to pursue engineering as a career or, let's say, who wants to be successful within the confines of a big company, what advice do you have for people?

[1684] What have you learned that you can sum up?

[1685] Yeah, look straight and then asking you that.

[1686] First, you have to, you know, you have to know yourself a bunch.

[1687] Like, what are you good at?

[1688] Like, you can't get really good at something you're not into and you're not good at.

[1689] So you have to have to have some natural talents for it.

[1690] And then you have to really spend some time figuring out what you like.

[1691] Like, I read this thing.

[1692] It was interesting.

[1693] Like, people think of college of expanding their possibilities.

[1694] And the university itself has so many options you think that would expand your possibilities.

[1695] But once you pick one of them and you study it for four, eight, ten years, you've narrowed your possibilities.

[1696] possibilities, right?

[1697] You're kind of stuck with your discipline and you pick that 20, which I think is crazy, by the way.

[1698] Like, I think if you want to be an engineer, a good general engineering degree, like mechanical engineering or electrical engineering will give you thinking skill sets.

[1699] I'm not a huge fan of people getting PhDs unless they really, really know they love it.

[1700] Right.

[1701] And then take some jobs where, you know, there's an opportunity to do something for a year or two and then do something else.

[1702] Like, I work, my work, my My first job out of school was a random job, but I worked on like five different projects in two years while I was there, you know, fixing hardware, building something, debutting something.

[1703] I learned a lot in the digital life.

[1704] I had many different roles even though I set that company for 15 years.

[1705] I wrote programs, I did logic design, I did testing, I did lab work, you know, and so I got to see a lot of different things and get a feel for what I really liked.

[1706] and I work with smart people that, you know, I had a lot to learn from.

[1707] I'm working hard when you're young is really useful.

[1708] You know, some people are like, well, you know, it's like the 10 ,000 hour problem.

[1709] And if you want to be an expert, you need to do that a couple different times on different things.

[1710] And you can't do it unless you really love it.

[1711] A friend of mine's wife said, what do they put in the water?

[1712] All you guys do is talk about work.

[1713] Yeah.

[1714] So you figure out what you're competent at because you need that.

[1715] figure out what you're interested in.

[1716] I mean, men and women seem to pick different occupations, not based on their competence, but on their interest.

[1717] And so interest is a very powerful motivating factor.

[1718] I've been in a lot of places with the best engineers for women.

[1719] So, you know, we know the numbers are less, but there's plenty of really great women.

[1720] Yeah, it certainly doesn't make it impossible.

[1721] It's just an indication of the, what would you call it, of the impact of interest as a phenomenon.

[1722] It's important as well as competent.

[1723] And so a diverse range of experiences.

[1724] Don't over index on something before you're really, you know, sure that that's something that you're really going to like or be great at, you know.

[1725] Don't be afraid to ask stupid questions if you don't know what you're doing.

[1726] Yeah.

[1727] Try to work with good people.

[1728] Working organizations where like if everybody hates the company you're working in, move somewhere else, You want to work someplace where the energy is good.

[1729] People are excited about what you're doing and why.

[1730] Like sometimes you might be in a company that has something going wrong, but your group is going to change it.

[1731] That can be really fun.

[1732] But you need some camaraderie, some hope, right, or the goal is clear.

[1733] Right.

[1734] So that's an adventure.

[1735] You have a destination and the camaraderie along the way.

[1736] Yeah.

[1737] And there's so many places doing so many wild things.

[1738] being stuck in a company you don't like that's going nowhere for 10 years, man. You don't have that many 10 years last in your life.

[1739] Make sure you're actually getting, especially if you're on getting different experience.

[1740] Somebody said, you know, you have 10 years experience or one year of experience 10 times.

[1741] Right.

[1742] Now, sometimes you work on the same thing and you refine it and you become the expert, but then you should feel like you're making progress and expertise.

[1743] Right.

[1744] But if you're just kind of going through the motions over and over and doing...

[1745] Then it's time to fire yourself under those conditions.

[1746] If you're bored, you're not moving, right?

[1747] Like, engineering's not boring, right?

[1748] It's relatively exciting.

[1749] Yeah, I think that's actually a pretty good rule of thumb.

[1750] If you're bored, you're doing it wrong.

[1751] Yeah, something's wrong.

[1752] Yeah, something's wrong.

[1753] It's funny, it's like, like, in A &D, we had this group that did test, and it was kind of dysfunctional, and there was a couple managers and nobody liked it.

[1754] And at some level, the test engineering wasn't the hardest thing.

[1755] So, but I decided, that's stupid.

[1756] Why isn't, why isn't our test group the best in the world?

[1757] But we were organized around it.

[1758] We had a really great leader.

[1759] We had a good team.

[1760] I told them I wanted it to be really great.

[1761] And I told the engineers to stop complaining about it.

[1762] They had a problem come to me and we'll fix it.

[1763] Within two years, people come to me, he's like, man, the test guys are killing.

[1764] Yeah.

[1765] Yeah, they went above and beyond.

[1766] They made it something of value.

[1767] You know, it was great.

[1768] It was super fun.

[1769] that's a really good place to end cool thanks jim good seeing you man much appreciated thank you for taking the time we'll talk soon yeah sure bye