Uncle Bob - The Long Reach of Code

Robert Martin (aka Uncle Bob) is a programming pioneer and bestselling author or Clean Code. We discuss the prospect of automating programming, spotting and developing coding talent, occupational licensing, quotas, and the elusive sense of style. Watch on YouTube https://youtu.be/ldTPVd3vO9Q. Listen on Apple Podcasts https://podcasts.apple.com/us/podcast/13-uncle-bob-the-long-reach-of-code/id1516093381?i=1000500544934, Spotify https://spoti.fi/3pVcr2H, or any other podcast platform. Listen to his fascinating talk on the future of programming: https://youtu.be/ecIWPzGEbFc https://www.youtube.com/watch?v=ecIWPzGEbFc Read his blog about programming: http://blog.cleancoder.com/ Buy his books on Amazon: https://www.amazon.com/kindle-dbs/ent... https://www.amazon.com/kindle-dbs/entity/author/B000APG87E Thanks for reading The Lunar Society! Subscribe to find out about future episodes! Timestamps (0:00 https://www.youtube.com/watch?v=ldTPVd3vO9Q&list=PLd7-bHaQwnthaNDpZ32TtYONGVk95-fhF&index=1&t=0s) - Automating programming (8:40 https://www.youtube.com/watch?v=ldTPVd3vO9Q&list=PLd7-bHaQwnthaNDpZ32TtYONGVk95-fhF&index=1&t=520s) - Educating programmers (expertise, talent, university) (21:45 https://www.youtube.com/watch?v=ldTPVd3vO9Q&list=PLd7-bHaQwnthaNDpZ32TtYONGVk95-fhF&index=1&t=1305s) - Spotting talent (26:10 https://www.youtube.com/watch?v=ldTPVd3vO9Q&list=PLd7-bHaQwnthaNDpZ32TtYONGVk95-fhF&index=1&t=1570s) - Teaching kids (29:31 https://www.youtube.com/watch?v=ldTPVd3vO9Q&list=PLd7-bHaQwnthaNDpZ32TtYONGVk95-fhF&index=1&t=1771s) - Prose and music sense in coding (32:22 https://www.youtube.com/watch?v=ldTPVd3vO9Q&list=PLd7-bHaQwnthaNDpZ32TtYONGVk95-fhF&index=1&t=1942s) - Occupational licensing for programmers (35:49 https://www.youtube.com/watch?v=ldTPVd3vO9Q&list=PLd7-bHaQwnthaNDpZ32TtYONGVk95-fhF&index=1&t=2149s) - Why is tech political (39:28 https://www.youtube.com/watch?v=ldTPVd3vO9Q&list=PLd7-bHaQwnthaNDpZ32TtYONGVk95-fhF&index=1&t=2368s) - Quotas (42:29 https://www.youtube.com/watch?v=ldTPVd3vO9Q&list=PLd7-bHaQwnthaNDpZ32TtYONGVk95-fhF&index=1&t=2549s) - Advice to 20 yr old This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dwarkesh.com https://www.dwarkesh.com?utm_medium=podcast&utm_campaign=CTA_1

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Dwarkesh Patel?

  Okay, today I'm talking with Robert Martin, who needs no introduction. So let's begin with talking about the future of programming. Will GPT-25 be able to automate programming? Will I not have a job in 25 years?

• 0:12Speaker 01

  What is GPT-25? What is that?

• 0:15Dwarkesh Patel?

  GPT-3 is the program that OpenAI just released. And it seems to be able to do some basic, make some basic methods, like making a palindrome. Because they just did deep learning over a bunch of GitHub repositories. So is the future of AI basically going to be able to automate programming?

• 0:31Speaker 01

  No.

• 0:34Speaker 01

  Pretty straightforward answer. There's a fundamental reason behind this. In order to finally replicate programming, if you wanted a machine that could program like a human, that machine would have to have human sentience. And we are very, very far away from that kind of a machine. Why would we need human sentience? Because someone has to specify the way the program is going to work. And that specification is, in fact, the program. No other specification will suffice. So a business user comes along and he creates a rough specification,

• 1:19Speaker 01

  but is depending on the human intuition and the human intelligence of the programmers to fill in all the horrible little blanks that the business person omitted. We programmers are detail managers. We deal with all the mess. We're the ones who deal with the fact that text files sometimes terminate lines with backslash n, but sometimes terminate lines with backslash n backslash r for reasons that go back 50 years because of teletypes and Unix and DOS and all this crap. And we're the ones writing the dumb little if statements to deal with bizarre little exceptions that no business person wants to even think about.

• 2:06Speaker 01

  If we tried to make a, a machine, a deep learning machine that could deal with all that stuff, we would wind up with a HAL 9000, something that has human intelligence. And, and we are so far away from that, that it does not concern me in the slightest that some, some deep learning machine will eventually take over all the programmers' jobs.

• 2:31Dwarkesh Patel?

  I'm happy to hear that then. But how about this potential, which is that the development environments and the tools of programmers' tools will become so advanced, that it'll be almost like a symbiotic pair programming kind of assignment where, you know, you're, you're working with the machine to write the code instead of just like writing it on the machine.

• 2:49Speaker 01

  Sure. I mean, that, we've already had that for, for, for the entire span of programming, we've been doing that. I mean, the original programs were written in binary, right? So Alan Turing, working on the, the automated computing engine, was literally writing things in binary because there were no languages. And then eventually we made an assembler. Can you imagine the benefit of an assembler over binary? All that horrible binary math that you had to do, you don't have to do anymore? And you can write your code in symbolic form. That's already a huge advantage. And then along comes John Bacchus with the Fortran spec in 1953. And what a huge advantage that was over assembly language.

• 3:34Speaker 01

  Algal came along pretty quickly. By 1966, you've got Ole Johan Dahl and Christian Nygaard inventing objects. This is, you know, 20 years after Alan Turing was doing binary programming. You've got the first object-oriented language. C comes along in 68, two years later, C++ comes along 12 years after that. And we've got this massive progression of

• 4:01Speaker 01

  Of incredible amounts of help to the programmer. Then in the, in the late 90s, we start getting this wild stuff in our tools, the Intellisense and the refactoring tools and the inspectors and the duplication finders. So a modern IDE is a treasure trove of, of tools that allow a programmer to do immensely more complex things than we used to be able to do. So yes, I hope that machine learning continues to make our tools more powerful, but that will never alleviate the need for the programmer to be there and supervise those tools. I'm a pilot, I fly a plane.

• 4:46Speaker 01

  The plane has a beautiful autopilot. I can turn it on and the plane will fly itself all the way to my destination. And I watch that thing like a hawk because it's a, a machine and it's going to do the wrong thing at exactly the wrong time. So I sit there and I watch that autopilot and every time it's supposed to do something, I make sure it does that thing. I'm still in charge, it's not.

• 5:14Dwarkesh Patel?

  But once you get these advanced tools, will the future of programming just look incredibly different from what programming machine is like now? You've written about how the, the way that Turing wrote the first programs is not that dissimilar to how we write programs now. Is that sort of continuity, something we should expect in the future? Or is the interface going to completely get modified?

• 5:35Speaker 01

  Well, I, so I mean, the interface has been completely modified since Turing, but, but the act of programming is still sequence selection and iteration.

• 5:45Dwarkesh Patel?

  Yes.

• 5:46Speaker 01

  Programmers still arrange programs in little units of sequence selection and iteration. Will we continue to do that? Well, I, I think that's a very interesting question. And we've had some hints about other possibilities. If you've ever programmed in Prolog, for example, you see that the sequence selection and iteration gets transformed into something else equally detailed, but it's not sequence selection iteration anymore. It's, it's declarative truth.

• 6:20Speaker 01

  I'm not going to go deep into prologue, but but if you want to bend your brain around a concept that you're not familiar with, study prologue for a couple of days. But but in the end, are we going to get to a a situation similar to my autopilot? Where I can set up some parameters in a tool? And then watch the tool as it starts to develop a program? And then I, as the programmer, watching what it's doing, start tweaking parameters around the outside. Saying, no, no, not that, not that, this, this, go in this direction. Will I be able to do that? Yeah, I'd say that's probably very likely sometime in the next couple of decades.

• 7:04Speaker 01

  That that the way we program computers will be more like training a dog. Rewarding it in the right circumstances and punishing it in the others. We'll see how that goes. We'll see how that goes, but under no circumstances will there not be a programmer involved? Someone with the same skills that you and I have today? Someone who is deeply analytical and a detail manager and is watching out for all the dumb little gotchas that the machine has no clue and no strategy for addressing.

• 7:40Dwarkesh Patel?

  I'm happy to hear that. Are teams of programmers less creative than individuals programming alone?

• 7:47Speaker 01

  Teams of programmers are for more creative than programmers working alone. But they also go in many different directions at once. So they need, there's, you need a leader. A team has to have a leader. What one of the problems that we have had in the Agile community? Is that there's a certain faction of Agileists who believe there should not be a leader? That the entire team is homogeneous and equal in a mass egalitarian society where everyone is the same. And my experience is that that doesn't work at all. A team needs a leader and then the the team is generating ideas just like crazy.

• 8:28Speaker 01

  The team generates ideas and the leader is the one who acts like, you know, Jean-Luc Picard. This one, not that one.

• 8:38Dwarkesh Patel?

  Interesting. Now, let's talk about educating future programmers. I watched an interview with Linus Torvalds, or he was expressing concern that the complexity of the ecosystem has grown so much that he's not sure that somebody in his position could come up with such a grand innovation. They're just not a rock to hold on to, to like get that level of expertise at such early an age. Are you, is, do you have that kind of concern that the ecosystem has developed so much that it's harder for young people to kind of get to the frontier?

• 9:10Speaker 01

  It takes longer now because there's so much more cruft that you have to wade through. When I was a young programmer, the machine that I was working on was initially a PDP8. I've got a little mock-up of a PDP8 sitting right there. It must have gone through a power failure because it's not doing what I usually have it do. But that was a very simple machine. It was a 4K machine, 12-bit word, half-megahertz clock rate. And there were no libraries, there were no frameworks, there were no nothing, right? If, if something executed in that machine, it's because you wrote it. And a programmer in those days could know everything about the computer.

• 9:53Speaker 01

  And it wouldn't take very long. It would take a matter of months to eventually get to the point where you just knew everything. But those days are long gone. It's, it's now not feasible to point at a computer system and say, "Well, I know everything about my MacBook Pro." No, you don't. It's just too much crud in there.

• 10:17Speaker 01

  So there is a much longer learning curve to become as adept at programming as those of us who started 50 years ago and had to climb that tree while we were working. On the other hand, it is not like being a doctor yet. If you want to become a doctor, you must specialize. There's, there's no way to go forward in medicine without specializing. Even a general practitioner is a specialist nowadays because they have to exclude all this other stuff they, they cannot do. And, and we are not there yet. A programmer over a period of...

• 11:00Speaker 01

  Five years, say, can become adept at GUI programming, middleware programming, database programming, telecommunications programming, threads, all of this stuff can become well-heeled inside of a five-year experience program. Ten years, even better, right? So it's not like being a doctor or a lawyer where you absolutely must specialize and narrow your focus down to a razor. Programmers can still be real, general practitioners.

• 11:33Dwarkesh Patel?

  Gotcha. Let's talk about the factories which manufacture programmers.

• 11:39Dwarkesh Patel?

  Universities, if you had to change one thing about how computer science is taught at universities, what would it be? And how would you change it?

• 11:46Speaker 01

  I would not teach it at universities.

• 11:48Dwarkesh Patel?

  Really?

• 11:48Speaker 01

  I don't, I don't think it's a university kind of thing. Again, this is a, a discipline that is only seventy years old. There is, there is not that much in it. There's, there's far more cruft in it than there is actual knowledge, actual, you know, solid, you know, teachable knowledge. And the, the vast majority of the skills that a programmer needs are trade-like skills, not educational-based skills, right? Or, or not university kinds of skills. If you want to be an electrical engineer, you have to go to university. There's just a tremendous amount of stuff you've got to learn there. If you want to be a doctor, you've got to go to a hell of a lot more than a university, right?

• 12:32Speaker 01

  But to be a programmer, and a good programmer, this is a skill that you can learn at a community college or in a trade school over a period of a year, maybe a year and a half. Get a good job with some mentors and learn it that route. Learn it that way. I, I think that is a much better way for young programmers to become programmers. Not go to university. It's not worth a four-year education. The amount of money you'll spend on that four-year, four-year education is not going to be worth the programming knowledge you get. Now, it might be worth a whole bunch of other knowledge. It's not worth the programming knowledge you're going to get. If you want to be a computer programmer and you're just getting out of high school.

• 13:16Speaker 01

  I've been looking at a trade program, some kind of a trade school or a boot camp or something like that. Some of these boot camps are not very good, so I'd be real careful about that. But some kind of program like that where you can, you can spend a few months learning a, learning a language like Java or C sharp or something like that. And then move on to a job which has a mentorship or an apprenticeship program.

• 13:41Dwarkesh Patel?

  Uh, I'm in university studying computer science right now, so I wish I heard that two years ago. But in any case, can you explain to me how it is then that we have an industry where to get in is not that hard? In terms of just the education you need at least, maybe there's like traits that are rare. But getting in is not that hard, the demand keeps growing very fast. How is it that people are still becoming corporate lawyers? When there's a job that pays really well, that doesn't require that much education to get, that you can get into real fast? What, what explains this in equilibrium?

• 14:14Speaker 01

  I'm going to try and interpret that question carefully. There are, there are, when, when I was a young programmer, I thought anybody could do this. That was my, my fundamental hypothesis, anybody could learn this, it's ones and zeros. I mean, how hard is ones and zeros? It took me a very long time and a number of very expensive mistakes hiring people. Uh, to realize that not everybody can do this. And in fact, the, the number of people who can do this is much smaller than I had at first thought. It, it's, it is not a rare talent, but it is not a massively common talent either.

• 14:56Speaker 01

  And the talent requires a number of things, right? There's, there's this need to be a deeply analytical.

• 15:05Speaker 01

  And there is this need to be able to focus, focus on very narrow things for very long times. And there are people who simply cannot do that. It's just not in their, in their mental wiring. They cannot do it. But some programmers can, or some people can, and they can become programmers. So you asked the question, why are we putting out corporate lawyers? Well, they have a very different kind of interest and mindset. Although they're analytical and they can focus, they focus on different kinds of things. We programmers focus on Boolean decisions, ones and zeros, blacks and whites.

• 15:47Speaker 01

  Lawyers, you know, they're looking at people. People are much more complicated than computers. It's a very different kind of mindset.

• 15:55Dwarkesh Patel?

  If you had to estimate, what percent of the population do you think has a repertoire necessary to become a programmer?

• 16:03Speaker 01

  Ooh, tough question for, to ask. What percentage of the, of the human population has the aptitude for programming?

• 16:16Speaker 01

  I think there are 100 million programmers in the world right now. That's just a guess, but I think that's about right. And that would be what, 1/80 of the population? So we, we've got some demonstration that at least 1% of the population has this skill. Close to 1, 1%. Maybe it's as high as 5 to 10%, could be that high. Could be. I know I've, I've got, I've had a number of experiences trying to teach people to program and not being able to get through after about 10 minutes. You know, their brain just goes somewhere else. They can't, can't do it. And.

• 17:00Speaker 01

  That seems to be much more common than the other talent, which is where the person that you're trying to teach gets excited and focuses in with you. And that is much rarer.

• 17:14Dwarkesh Patel?

  Hmm. So if going to university is not the best idea for somebody wanting to become a programmer. If somebody wants to solve highly technical problems, they think, well, it seems like I would need a good education to be able to do this. Maybe I need to know a lot of math. Would you recommend then they go to go to university and get a math degree so that they have the technical background?

• 17:34Speaker 01

  I used to think that was true and I actually hired someone because they had a they got a master's degree in mathematics. And I thought, okay, you know, math, masters degree in mathematics, you're going to be able to think this way. You're going to be able to focus. You're going to be able to do this. And I I found myself very disappointed with the way this person became a programmer. Now, eventually they succeeded. But it was not the quick success that I had expected. It took a lot of of working and a lot of failure to get this person's mindset into the programming mindset as opposed to the mathematics mindset. So there's a difference there. Now, is it worthwhile knowing mathematics as a programmer?

• 18:18Speaker 01

  Yeah, in certain fields, it certainly is. I mean, if you're going to be doing any kind of, you know, physics work, you know, you're going to be writing the code for a particle accelerator, probably want a little bit of math, right? If you're going to be doing a really deep, interesting financial stuff, going to be one of the quants in a in a stock exchange or something like that, probably need to know a little bit of good math, right? On the other hand, if you're a run-of-the-mill programmer writing code for, oh, I don't know, maybe. Yeah. Siri just yelled at me for saying run of the mill. Um, uh, if you're going to be writing the code that runs inside my refrigerator, for example, the

• 19:03Speaker 01

  machine control software inside my refrigerator. And probably you don't need a tremendous amount of math for that. A good high school algebra would probably do enough, maybe a little bit of precal, probably enough. I don't think you need, you know, deep math theory for that kind of stuff, even the guys who do. Game work, even though that's pretty heavy physics stuff, all that physics stuff is already in the physics engines. So most of the time there, I think you don't need all that math. Is it important for a programmer? To know the business they're going into and the answer to that is yes. Right? If you're going into an insurance business, you better know the insurance business.

• 19:44Speaker 01

  If you're going to be a programmer in the telephone company, you better understand the telephone company. Uh, and, and so. Much of the early education of a programmer in a mentorship program at a company. Is really about the company and not about programming. This is what we, this is our, our field of expertise. This is what we do. This is what an insurance claim is. This is what all these other things are. This is how we think about it. Uh, and so that you can, you can effectively apply. The business knowledge. To the problem of writing code and that brings us full circle.

• 20:24Speaker 01

  Because you had asked me originally and you'd said, well, aren't we going to eventually have these machine learning things that can write code? And the answer to that is probably not because it's the, the intuitive knowledge of knowing the business you're in. That allows you to make these low level decisions that the business people don't want to make.

• 20:43Dwarkesh Patel?

  Hmm. Uh, from that, does it follow that if you want to become a very competent programmer? You should try to specialize in a specific industry where, where, where the application programming is best exercised? Or should you try to develop first a broad range of experience and expertise?

• 20:58Speaker 01

  Uh, both actually. So as you take on a job, you want to, you want to specialize in the business that you're in. For a period of N years, whatever N is, might be two, might be three. Uh, probably that short though. And and get to know that business and then see if you can move to a different business domain. And and spend two or three years on that and then try another business domain and spend two or three years. In that and by that time you've got a nice broad business domain. And that will allow you to infer other business domains. Because you know, business domains have similarities and differences, but once you've got. Three or four of them in your head, you can start to triangulate.

• 21:44Dwarkesh Patel?

  Hmm. Now, when you are considering hiring somebody as a programmer and they don't have a lot of experience, so you can't just. You can't just look for somebody to vouch for them. What is it that they can do to signal to you that they would become a competent programmer?

• 21:59Speaker 01

  That's a heck of an interesting question and one that I've struggled with for. A very long time now, a number of decades. And I have found no reliable test.

• 22:09Dwarkesh Patel?

  Oh.

• 22:11Speaker 01

  I used to make people. So back when I thought it was math. I used to make people solve Newton's method for calculating a square root. I just draw it on the board. I draw parabola on the board and say, okay, here's the slope, you know, it's the derivative of the. Function now come up with an algorithm and I'd let them use any language they wanted. And I, you know, I thought that was a pretty good approach and then and then I found people that were very good at that. But could not do other things. Then I thought, okay. I'm going to describe. The sieve of Eratosthenes, beautiful algorithm for finding prime numbers, right? I'm going to describe that in in simple terms and have them write some pseudocode.

• 22:55Speaker 01

  And show me that they can actually build a doubly nested loop, right? And that was also not reliable, although it was better. It was not reliable. So what I've, the conclusion I've come to now is that. There is no simple test. What what you need to do is spend a fair bit of time with someone and observe how well they learn. Over a period of time and how well they apply that learning, how well they can focus, how well they're. How well they can deal with a problem analytically. And that period of time is probably weeks. So the the kind of interviewing that I would prefer to do now.

• 23:40Speaker 01

  And I, by the way, I don't hire people anymore. I'm now just kind of me. But, but I know people who do hire people and one of the, one of the ways that they do it is to ask them to join an apprenticeship program. And in that apprenticeship program, they will be given a series of ever greater challenges, not all of which are programming challenges. Some of them are writing challenges. Some of them are speaking challenges. Some of them are presentation challenges, but a sequence of ever increasing

• 24:14Speaker 01

  challenges. And then at the end of that, an assessment is made by everyone that they have talked to. And they're offered a job or not. And in some cases, there's a stipend salary for that apprenticeship program. In other cases, there's not.

• 24:30Dwarkesh Patel?

  Interesting. Do you suspect that the traits that are required to become a programmer, that these are developed in your childhood through the experiences you have in early life, or maybe through high school and middle school? Or is it something that some people are probably just born with on some level?

• 24:45Speaker 01

  It's hard to know. Um, I, I, I think there's a certain amount of just neural wiring. Said, you know, some, some character defect that causes you to be a programmer. That not all people have. However, I have seen people who were not at all interested acquire the interest and acquire the talent. Um, at different stages of life. My son, for example, who is now a programmer and a very accomplished one, had showed no interest at all in early grade school. When I got interested, I was 12 years old. Right. So I, I thought, oh, okay, everybody's going to be interested at 12.

• 25:28Speaker 01

  Now, my son didn't show any interest at all until he left home. And became a, or, or close to leaving home, late, late high school, early university. Uh, and came back with this interest and skill. My daughter has just turned 30 and is suddenly very interested and is acquiring a significant amount of skill. So I've seen this happen at different stages. The switch turns on. And, and all of a sudden, they've got all the, all the motivation and the focus energy and the analytical skill they need. So I don't know, you know, it's a deep mystery to me.

• 26:11Dwarkesh Patel?

  Are there advantages to learning this material and adopting these skills early in life? Or do you get a more mature perspective if you learn it later on in life?

• 26:21Speaker 01

  Had you asked me that question 20 years ago, I would have said there's obviously an advantage to teaching young children. I mean, if you can get the young children immersed into computer as well, then obviously they're going to turn into much better programmers. And I'm not so sure that's true now. So the counterexamples are the vast number of programming toys. And they span a huge, a huge period of time. So nowadays, you've got like things like Lego Mine Blocks and or Mindstorm. And and you've got a whole bunch of other tools that try to encourage young people to become programmers.

• 27:06Speaker 01

  You know, kids that are 12, 13, 14. In the early days, we invented languages that would help young people become programmers. The logo language, for example, was all about, you know, driving a little mechanical turtle on the floor and letting it raise a pen and draw on a piece of paper. And kids, you know, they would, they would issue it commands, go forward 10, turn right 90 degrees, go forward 10, drawing squares and circles on a paper. And what we found through all of these, through all this long period of time, attempting to make tools that would appeal to children is that it did not have a profound effect.

• 27:50Speaker 01

  There was a whole bunch of children that would sit and drive the turtle around for a little while and then lose interest and never develop the, the more interesting skills of putting things in variables and calling functions. And those just went, they didn't happen. And we see the same thing with things like Lego Mindstorms. Now, there are children who do grab it. And you can tell they want to become programmers. But then the question is, well, wouldn't they have just grabbed it anyway? Why did they need this tool? Is, are these educational props actually incentivizing more people to become programmers? Or are they just helping the ones who would become programmers anyway?

• 28:35Speaker 01

  And I don't know the answer to that. I'm skeptical nowadays.

• 28:39Dwarkesh Patel?

  Interesting. So you're not bullish on mandatory coding classes in middle and high school?

• 28:45Speaker 01

  I think everybody should have a, everybody should get the experience.

• 28:52Dwarkesh Patel?

  I see.

• 28:53Speaker 01

  Probably, you know, for a four week or a six week period, everybody should do a little, you know, basic programming or Ruby programs, you know, I want to be careful there because turn people into Ruby and put, you know, it's a very complicated language when you get into it. But, you know, maybe a little, little, little bit of coding, a little scripting, a little something so that you know basically what it's about. But other than that, I don't think there's any, any need to like push a year long course. I don't think that's necessary. I think the ones who are interested are going to do that on their own.

• 29:29Dwarkesh Patel?

  Right, right. So let me ask you about the sense of style that a programmer develops over time. Is this at all related to what has been called the elusive writer's ear? Where a writer over time develops a sense of prose and style? Is that sense related to the one that one develops for programming?

• 29:50Speaker 01

  Michael Feathers coined a term 20 years ago, which he called design sense. And he, he's threatened to write a book on this for the last 20 years. I haven't seen him write it yet. But the basic idea is that, yeah, there is this,

• 30:09Speaker 01

  there is this sense that a programmer develops over a long period of time that a, a bit of code or a design of a system or an architecture of a system somehow feels right.

• 30:27Speaker 01

  And it's an elusive kind of sense, it's the the judgment of beauty. You know, it's like, oh, beautiful thing. And and the word beauty ought to be used here because there are beautiful designs. There are beautiful architectures. There are beautiful functions which can be described as beauty and understood as beauty only by people who have the experience to look and say, oh, my goodness, that's beautiful. I was watching a a program on on Netflix yesterday. It's called Queen's Gambit, it's a new one. It's about a a young girl who becomes a chess master. And in the midst of this, someone asked her, you know, why are you so interested in chess?

• 31:12Speaker 01

  And she leaned back and she said, it's beautiful, or it can be beautiful. And I think the same thing applies to software.

• 31:21Dwarkesh Patel?

  Hmm. Christopher Hitchens, who is known for his excellent writings, suspected at some point that the sense of writing was tempered by a sort of musical intelligence about cadence and prose. Do you think musical intelligence plays into your ability to code well or beautifully?

• 31:37Speaker 01

  Well, there is, and and this has been known for some time, there is a very strong correlation between programmers and musicians. People who study music or have played an instrument or have developed a musical sense often become programmers. Why that's true, I don't know. It could be the the rhythmic thing, the the mathematical layout of the of the scales, the who knows? If you look at a a musical composition, if you look at a sheet music, it is a program. It's got if statements, it's got while loops, it's got sequence, it's got everything. So maybe it's something like that. I don't know, but there's definitely a correlation.

• 32:21Dwarkesh Patel?

  Interesting. So let's talk about your idea for, or your claim that we need some sort of body within software engineering that decides who has the ethical and the technical competence necessary to be, remain a programmer. I don't know if you change your view on this, or if I, and I'll let you elaborate what you mean by this. But somebody who is concerned about this might say, well, this is exactly what is happening in other industries where we get occupational licensing that's keeping people out of the industry so that the labor pool is artificially reduced to the advantage of the people already in the industry. And there's also the added wrinkle with the increasing calls for censorship and cancel culture.

• 33:05Dwarkesh Patel?

  We could have, expect that this body might itself become a tool for removing people who challenge orthodoxies. How do you respond to these challenges?

• 33:15Speaker 01

  Well, so the problem,

• 33:20Speaker 01

  the problem is that programming has become an entirely essential job function for our society to exist. Human society can no longer exist without programmers, and that dependency is growing at an exponential rate. So that, you know, you look around the room here and you'll just see dozens and dozens of little electronic things that have code running in them that just make life doable. And all of the things that we used to depend upon that did not have code running in them all now have code running in them. So your microwave oven and your refrigerator in your car, telephone, your television, all has code running in it.

• 34:02Speaker 01

  Our society can't exist anymore without programmers. And there is no constraint on programmers to behave ethically, to have any standards, to have any disciplines. So that's an unstable situation. It can't go on like that.

• 34:20Speaker 01

  How do we address it? Do we need something akin to the AMA, the American Medical Association, who provides licensure for doctors? Or do we need something else? Do we need maybe a guild structure? Maybe a whole set of little organizations that license their own programmers with different sets of standards and ethics and disciplines? And then let the market decide. You know, those guys did better than those guys. I think we'll hire them. I don't know where that's going to come down. I, my fear is that it will be forced upon us by government

• 35:03Speaker 01

  before we can develop that solution. And I'm hoping that we can develop that solution before government sticks their little fingers in there and screws it up completely. There are a number of companies that have begun to do this. So you look out in the world out there, especially software consulting firms. And there are software consulting firms out there that have based themselves on a set of principles, ethics, and disciplines. And they advertise themselves that way. And I think there will be more and more of that. So that's kind of the direction I think the industry is going to go. At least I hope it goes that way. And I hope we get there before government sticks their foot in it and kicks us all to hell.

• 35:48Dwarkesh Patel?

  I hope so, too. I wanted to ask you why tech of all industries seems to be so political? In the sense that specific companies seem to have a clear political orientation? That doesn't seem to be true in many other areas of work. Do you have a sense of why that is?

• 36:07Speaker 01

  Um, no. Well, yes and no, I guess. And it's, it's been a deep puzzle.

• 36:19Speaker 01

  If you, let's, let's do this differently. If you look at the political bias geographically, right? There is a definite political bias in the United States. You've got the two coasts, they tend to be liberal, blue states on each coast. Cities tend to be more liberal. Rural areas tend to be more conservative, the Midwest tends to be more conservative. If you get, um, go to a company in Kansas City. And you find a company there, you know, that has a hundred programmers. Odds are those programmers are going to be conservative. It's just, you know, because of the geographical bias. Go to San Francisco, odds are the programmers there are going to be liberal.

• 37:06Speaker 01

  That's just the geographical bias. Is there anything else in software? That drives people in a certain political direction?

• 37:19Speaker 01

  Well, maybe. It turns out to be easy to become a programmer. We talked about this before. You don't really need to go to university. You don't really need to, um, to get a full, full education, right? You can become a programmer without a lot of experience. And, and because our industry continues to invent new, um, new schemes all along. Right? There are people who can jump into these schemes very early. So for example, HTML. Um, HTML was invented, you know, 20 years ago. And the, and a whole bunch of high school kids suddenly got really good at HTML. And of course they went out to get jobs.

• 38:03Speaker 01

  Um, without a lot of university training, without a lot of anything. So it's easy. There's, there's a, a low, uh, barrier to, to, um, acceptance early on.

• 38:18Speaker 01

  And that, I think, has led to a good thing, which is a bunch of marginalized folks, folks who live in marginalized communities, manage to get in very easily. And they get experienced and they develop skills, and that's also good. And then I think, and this is a guess on my part, I think they use that skill and that, and that technology that gives them a voice. And that voice has been turned.

• 38:52Speaker 01

  And in, in a way that is sometimes very detrimental. Now, I don't want to cast, you know, aspersions on every marginalized community. But I have noticed that there are some that suddenly develop a very loud voice. And in, in a lot of cases, they wind up driving the cancel culture.

• 39:12Speaker 01

  Just, just, you know, the way that looks like it's been happening from my, you know, my middle age. No, I'm not middle aged, I'm actually old now. From my old conservative white bias, right? That's what it looks like to me.

• 39:27Dwarkesh Patel?

  Interesting. Let me ask you about quotas. I, I think you laid out a really good argument in your blog about why quotas don't necessarily help the groups you're trying to help. Does the argument you laid out there, and I'll let you explain it, apply and generalize to other places where things like quotas are applied? Like universities, for example?

• 39:51Speaker 01

  I think it does. I think it always does. I mean, the idea behind a quota is, is always very well intentioned, right? There's, there's a marginalized group, they're not getting serviced properly. We need to relax the barrier of entry for that marginalized group so that more of them can join. Well intentioned, you know, everybody wants the best. But what's the end result? The end result is that you lower the standards. The standards have to come down, which hurts the entire industry. And it doesn't help.

• 40:28Speaker 01

  The marginalized folks to let them in under low standards, but expect the high standards on exit. Right? Because then you'll see this this failure and we see that in universities and we see it in in employment as well as that the dropout rate or the failure rate is higher. It doesn't help. I once sat with a woman, a very accomplished woman. And we were at a conference and I was giving a keynote and she was giving a keynote. And we were just talking about this issue and she looked at me and she said, I hope to hell that they did not ask me to do this keynote just because I'm a woman. And you know, I I could feel that angst, the devaluing

• 41:11Speaker 01

  of her of her ability to do a keynote talks because she was invited just to become a woman. So her fear, her fear was that she was being devalued by a quota system that she didn't even know that existed.

• 41:30Speaker 01

  I think in general, the only way to deal with issues like this is to do it entirely on merit. Right? If you're going to be a programmer, we will, you know, I'm perfectly happy to hire anybody. Don't I don't care what your, you know, political biases, your sexual biases, your race, your racial situation. I don't care about any of that stuff. As long as you can write the code, I'm happy to work with you. I don't want to suddenly or I don't want to relax the standards for a certain subgroup. I don't think that's fair to anybody. Just me.

• 42:04Dwarkesh Patel?

  Yeah, yeah, I completely agree. I Malcolm Gladwell was in a debate once and I actually really like Malcolm Gladwell, but he started off by saying, first of all, I think it's scandal that nobody on this stage is a woman. And there's a point there, but then I thought about it and I realized, well, imagine being the woman who just came on stage because Malcolm Gladwell wanted the woman. You know what I mean? Like, imagine being in that situation. Anyways, let me ask you finally about advice you would give to a 20 year old who wants to become a programmer.

• 42:36Speaker 01

  Well, okay, being, becoming a programmer is not particularly difficult in our, in our day and age yet. It may be at some point in time, but right now it's not. If you're, if you think you want to be a programmer and you think you have the aptitude. It is stupidly simple to get on YouTube now or get on a bunch of other websites and learn to code. And it doesn't take a lot of effort. If you've got a laptop, if you've got a computer at your disposal, if you can go onto YouTube or any of the other sites out there, just find some site that'll walk you through an initial Java program. And then experiment with that as much as you can, learn, learn how to write this code.

• 43:19Speaker 01

  Read other people's code, go onto GitHub, download their code, read it, understand it. It'll be hard, but that's okay. You know, you can do it. There's a whole bunch of books out there, tons and tons and tons of very good books that you can read. Start out with simple books like, you know, The Joy of Java or, you know, Java for, for Dummies in 10 Days. Or, you know, Learn C in 21 Days, or any one of these books that will just help you immensely. Find a mentor, someone who's been at it for a while. Most of them will be overjoyed to have someone that they can teach and work with and help.

• 44:02Speaker 01

  If you can get into a trade, trade school for programming, that would be great. Not necessary, but it would be great. If you've got the means to go to university and you want to spend the ridiculous amounts of money that universities now cost and go into horrific debt for, you know, the rest of your life. Okay, you can do that too. I'm not going to stop you. And you'll learn a whole bunch of really cool stuff there, stuff that some of which you'll even be able to apply later on in, in your career. But if you don't have that means, if you don't have the means to go to university, that's okay. I mean, you can get this, you can acquire these skills, you can become a programmer

• 44:47Speaker 01

  with a minimum amount of investment. Just find some people to mentor you along, find a company that's willing to hire a beginner. And there are a lot because the demand is immensely high right now and, and will continue very high for a long time.

• 45:03Speaker 01

  Find a way to get in and it's not that hard. Right. So that would be my advice. Read like crazy, read inhale as much as you can. Study like crazy, watch videos and do exercises. Just, you know, fill your brain with this stuff. And then it ought to be simple after that.

• 45:27Dwarkesh Patel?

  Excellent. Well, thank you very much for talking with me today. This is very fascinating.

• 45:32Speaker 01

  It's a pleasure. I enjoyed it very much.

• 45:34Dwarkesh Patel?

  Hey, if you enjoy this podcast, please consider sharing it with your friends and posting it on social media. Word of mouth is incredibly valuable for a new and a small podcast like this one. So thanks for watching.