What Would I Do Differently?

About this Episode

In this episode of the Acima Development Podcast, Mike, Will, and Kyle explore the intersection of language learning and programming. Mike kicks off the discussion with a personal anecdote about studying German in school, noting how little he’s used it since, and how learning Spanish as an adult through Duolingo was more rewarding. This transitions into a broader discussion about how the process of learning any language—human or programming—reshapes cognitive patterns and makes it easier to acquire future languages. They draw parallels to AI, explaining how training models on multiple languages makes them better at understanding new ones, suggesting that the mental “training effect” of learning is more important than the specific content being learned.

The conversation then shifts toward practical advice for aspiring developers. Will emphasizes the concept of “the lore”—the unwritten, experiential knowledge in software development that can’t be easily taught in schools or bootcamps but must be passed down through mentorship and hands-on practice. He shares his own journey from fixing computers to compiling Linux source code at 15, which gave him a massive edge later. They all agree that building projects, no matter how scrappy or simple, and putting them out into the world is the most effective path for beginners. This “build and share” approach helps attract feedback, builds credibility, and accelerates learning in a way that no tutorial can match. The group also acknowledges the tough job market, especially for juniors, and suggests finding mentorship even in “sketchy startups” as a legitimate way to gain real experience.

Finally, the hosts reflect on how the landscape of software development is evolving in the era of generative AI, emphasizing that while tools are getting smarter, the core need remains unchanged: people who can think clearly, break down problems, and express logical solutions. They argue that learning how things work under the hood—whether it’s operating systems, threading models, or build pipelines—is still essential for becoming a great developer. Will humorously champions “vibe coding” and encourages makers to lean into their curiosity and creativity, while Mike reminds listeners that even though the languages may change, the deep thinking and hands-on experience involved in building and debugging remain the bedrock of the profession.

Transcript:

MIKE: Hello, and welcome to another episode of the Acima Development Podcast. I am Mike, and I am hosting today, as I have usually been doing. With me today, I have Kyle, and I have Will Archer. Actually, I have Kyle Archer and Will Archer [laughs].

WILL: No relation.

KYLE: No relation here.

MIKE: No relation, but they sound alike [laughs]. And, sound like, speaking of sounding like, I'm going to start today by talking about something I did long ago in high school. I took a German class. I took four years of German. I started in junior high, and then got to four years of German. And you know how many opportunities I've had to use German since I got out of high school? I think the answer is zero [laughs]. I think the answer is zero.

Actually, no. I ran into...maybe four years ago, I was visiting my parents, and I was going on a walk with my parents. And my dad ran into somebody who he knew who was from Germany, and he said, “Hey, Mike, so you speak German.” And I froze. It had been, like, 30 years [laughs]. I’m like, “Probably not right now [laughs].” I didn't have much software. If I'd taken a little bit of time, I could have done something, but no.

If I see, like, on a packaging, like, shampoo bottle or something, you know, and they've got instructions in German, you see that, and you don't even think about it. You just start trying to read it. You’re like, oh, I recognize that, and you start reading and try to make sense of it. That's about my entire experience using German is reading the back packaging that happens to have a translation in German, you know, since I got out of high school.

However, a few years ago, I said, you know what? I want to learn a language that I might actually use [chuckles]. I started using Duolingo, the language learning app. I even have a paid account now, because after a while, I started using it more. I've been doing that pretty consistently. I think, daily for the last four years, I've been learning Spanish. And I actually completed the course a while back, but they give you just review, and I've been doing that. And I got everybody in my family doing it, too. So, we're doing it together. And I went through it the fastest.

And I've read this: it turns out that after you've learned one language, it's easier to learn another. And, in fact, they actually recommend that if you want to learn a language very different from your native language, for example, for an English speaker, if you want to learn an East Asian language, if want to learn Chinese or Japanese, you learn it faster by learning a European language first.

WILL: I believe that.

MIKE: Because it trains your brain to think about things in a different way, to think about language differently. And once you've done that, you're primed, and it's easier to switch to --

WILL: So, you’re saying if I wanted to learn Chinese, right, I want to learn Chinese, it has very few loanwords or commonalities, then I would go faster if I learned Spanish, and then went to learn Chinese, than I would if I had only learned Chinese.

MIKE: That’s right. Which sounds crazy, right?

WILL: It does sound pretty crazy.

MIKE: It sounds crazy. Like, how could you possibly learn Chinese faster by learning another language first? But I've read this. I don't have the source. I don't have it right in front of me, but I have read this before that it actually is faster to learn a different language that’s more similar to your native language. So, you know, again, if you were a native Chinese speaker, it might be faster to learn maybe Korean first and then go learn a European language because it changes the way your brain thinks. I think that's fascinating.

Maybe the source that I read is totally wrong and leading me astray, but there's some intuitive sense to it that it trains the way you think. And regardless of whether it's universally true, there's probably some meaningful truth there that there's some training about how to think that is different from learning the specific syntax.

To take this further, AI models have been doing...they've run some of them in lots of different languages. So, they just train a text on lots of languages, and specifically, they train on text that's been translated from one language to another. And the fascinating thing that they find is the models that have been trained on one language or on many languages actually are better at learning a new language than they were at learning the first language.

So, if you have a language that can translate between English and German, and English and Spanish, and Spanish and Turkish, and just name your language pairs, and then you you add a new language to the mix, like Igbo or something, then if you've already had those other languages, your model will be faster and even do better than it was if it was just trained on that language. And there are a couple of reasons for that.

First of all, you have more data from commonly represented language pairs. There are probably really common ones, like English-Chinese, for example, you know, where there's a lot of stuff translated out there. And so, you have a higher body of language between really commonly used languages out on the internet, but it's more than that. There seems to also be just some universality of learning to translate that makes the models better. When you've learned between a lot of languages, you actually improve them all together.

WILL: That makes sense. That makes a lot of sense, to me, actually, I think. Because you're dealing with, how do I put it, like, you're dealing with...I don't want to use the word means but ideas and a more robust mapping of linguistic phrase and the poor, I would say, human idea. We're all people, and to a degree, I think language shapes the way we think, but only to a limited degree. And so, the more mappings, the more cross references you have between a turn of phrase, or a word, or something and the fundamental idea of the human condition that you're trying to express, then I think the more, you know, evocative and robust your translation between the two things would be.

MIKE: And it's interesting you say that. You can say there's this human experience that it gets closer to you. And they actually talk about, also in AI research, the embodiment problem, which is that some concepts have been really hard for AI to learn, partly because they don't have the human experience. I believe the...I can’t remember the philosopher who did this...was it Plato? Plato and the cave? I'd have to go look it up.

WILL: That sounds about right. That sounds right.

MIKE: So, imagine you're sitting in a cave, and people are broadcasting pictures of something up in the wall in front of you. And so, your whole experience comes from that broadcast, you know, looking at this shadow show. It’d be difficult to learn some aspects of the world. You could learn a lot, but it'd still always be second-hand, and that's how these AI models are learning. They don't have the body, typically. I mean, there are some robots now with, you know, physical bodies that can go out and live or experience the world. I'll use the word live very loosely [laughter]...to go out and experience. And so, there's some ability to learn there.

And, in fact, simulation is hard for robotics. You know, it's a lot better if you use a real robot, but then you break your robot. It’s expensive [laughter]. That embodiment problem is tricky. It's hard to learn if you don't have those experiences. You're suggesting there's this human experience that is independent from the modeling we use in language, and that if you model it several different ways, you actually get closer to the human experience, which is a fascinating idea, which we won't dig too deep into, but it's still a fascinating idea you said there [laughs]. I like that.

I mean, we haven't even hit the core topic that we're talking about today because this initial idea is so interesting that leads to it. I wanted to ask, so present it, if you were going back, if you could go talk to your old self that was first learning to code, what would you tell them? How would you have them do different things differently? And I think this idea of language and this idea of experience and language being different is a fascinating analog because it's another language, kind of language, that you're using to accomplish a similar purpose, but in a totally different context.

WILL: So, it's been a long time since I've learned how to do this, and I'm always, I don't know, I'm aware of, like, you know, how much unconscious knowledge that I've just sort of absorbed and accumulated and I deploy without thinking in just the daily process because I'm learning all the time, right? The machine never stops. The treadmill never stops [chuckles]. I'm always learning new, different more horrible ways to break machines.

But I have brought on board a fair number of junior developers. And the one thing that I look back on the most is what I refer to, colloquially, as the lore of just these are things that you have to know because they're not your job. But you're never going to be able to effectively work and to make an artifact that functions without the lore, which is just, like, anecdotally, I remember...

So, my first job I was working in a computer shop. I was a computer repairman. And I got exposed to a lot of the lore, you know, back when it was really like a dark art. It was really like a guild. When I started programming computers, there was no internet. I mean, it existed, obviously, but, like, there was no internet. There was literally no Google, nothing.

And so, I got in this computer shop, and I got, like, some graybeards who taught me some stuff. And then, I got exposed to Linux, like, as it was rolling out. Like, as it was coming out, I was getting on Linux, and I had to, like, I ordered...I purchased a slug of six or seven CD-ROMs with the source code to Linux that I had to compile myself [laughter] as a 15-year-old kid to have Linux. And it was such a hateful, abusive process.

But then I went and I got into college, and I got onto the programming team. We had a little competitive programming team. And I was wildly more effective than any of my peers because I had this lore around Linux and making the shell do things, reading files from the shell like basic commands. How does a directory work? How do you mount a device? Like, all this crazy stuff that is both absolutely foundational to the process of creating a running, functioning artifact, at least for me, and is not the kind of thing that anybody would bother to write down.

And so, that, to me, is the biggest and craziest thing, and, like, how do you get? How do you get the lore? And that is the thing that almost...I don't know of another way other than being in the company of professionals. It has to be passed, almost literally, from hand to hand. You need to be, not necessarily in the room, but in a distributed workforce, we're finding you need to be either in the room or in a very, very close virtual analog. I mean, it has to be live, one-on-one transfer of information between you and me. I think a one-on-one Zoom call would be sufficient, but I think a webinar or some analog that wouldn't do it.

And so, then it sort of leads into this problem of, like, I think, in 2025, we're being extremely hard on junior developers in this year, at least. At this moment in our industry, we are being very, very, very hard on junior developers because productivity is going up. And people, you know, productivity wants to go up. Salaries want to go up, which means the demands to get a salary job are also going up.

So, if I'm starting from zero, all that in mind, this is sort of, like, the background of like, okay, this is the problem that I need to solve. I know nothing. I need to get up enough so that I can demonstrate enough value so that I can sort of start the flywheel turning where I work, and I get paid to work, and then I get to work more, and I get paid more, and then that virtuous cycle where I can sort of put energy in, and I get enough money to, like, go full-time, or part-time, or something like that.

So, what I would say to do is...I suppose that’s the goal. It does assume the goal is to go pro. The goal is to go pro, to have a job, have a career, you know, like, really do it. If you just want to go out and have fun, I think there's never been a better time to go out and have fun doing programming. I think there are a lot of places, a lot of frameworks and stuff, or a nominal fee people will be happy to just get your stuff out there in the world. I know C was a Rails shop, and, like, what are the guys who, like, you just sort of put your artifacts up, and they just manage all the service stuff for you.

MIKE: Oh, Heroku?

WILL: Yeah, Heroku. There are places like that, I mean, analogous ones, I'm sure, for, like, you know, JavaScript, or Node, or React, or whatever you want to do. And so, those kind of places are really good. You've got your generative AI sort of coding assistance that could go out and do it. So, honestly, I would just make things and get them out into the world in the dumbest way possible. And this would be really hard for me, especially as, like, a broke kid. Like, I would spend some money to, like...I'm the guy who compiled Linux from source at, like, 15 years old, so obviously, I'm a little bit of a psycho, but that's the biggest thing that I could see.

And I’ll say that I think I'm monologuing a little bit, so I should take a step back. But, like, one of the patterns that I see in terms of people who are trying to bootstrap a software engineering career successfully is people who are trying stuff, putting it out on the internet, writing a blog, making a thing, like, I'm trying to make this thing. And then, you just put it up, put it up, put it up, put it up, put it up. Here I'm building this thing; here you go. Here I'm building this thing; here you go. Here I'm building this thing; here you go. And you put it out into the world. You're proactively doing the work. You're asking good questions.

Because most developers, most engineers, are very, very, very, very, very forthcoming with helping anybody out there who’s actually doing the work. If you come to somebody like, “This doesn't work. I don’t know.” Then, you know, you may get a cool response. But if you come out like, “Hey, I was trying to do this thing, and this is what I tried, and then this thing happened. And so, then I tried this, and this, and this, but then that happened. And I'm worried about, and I'm thinking, like, this is maybe a thing that I could try. But I have this question and, like, how do I implement it?” People will do backflips for you. [crosstalk 16:54]

MIKE: These are professional problem solvers. They can't help [laughs] but want to just solve the problem.

WILL: Exactly. And so, the cycle that I'll see in terms of people doing it right is they'll go out. They'll start working. They'll put their work out into the world, at whatever level of quality that they're doing. And they'll start, and then they'll get some feedback. And they'll put some more out, and they'll get some more feedback. And they'll pull some more out. They'll get some more feedback. They'll put some more out. They'll get some more feedback. And then, the blog is done because they have a job, and they're too busy to write the blog anymore. And I've seen that happen over and over and over. And so, that's kind of what I would...

Generative AI, I think, is a little bit of a tangent, because, I mean, really, it's just, like, start making things and put it out into the world publicly, and share it with everybody. There has never been, nor do I think there will ever be, a surplus of people creating things and pushing them out in the world. People who are willing and able to do that stuff are rare gems. We've never had enough of them in the world. We never will. And if you demonstrate yourself as that kind of person, there's a spot for you. You just have to, like, raise your hand and say, “Here, I'm out here. I'm doing it. Here I am. Look at all the stuff I'm doing.” And you’ll get scooped up.

MIKE: And I've seen that repeatedly. I’ll give a couple of examples, you know, go and build stuff. So, a woman...I think we may all know her, but I don't want to, you know, like, speak for somebody without, you know, tell their story without talking to them. Home school --

WILL: Is it bad [laughs]?

MIKE: No, it's not bad at all.

WILL: I’m kidding. I’m kidding.

MIKE: It's actually really good. So, she was a homeschooler. She had been teaching her kids for years and found that she needed to organize her stuff. And she’d wanted to code for a while, so she learned to code to build an app to organize her stuff. She was not out working professionally. She just built something because she knew she needed it. And she learned so much about how to get something done by solving this need that she joined an apprenticeship, quickly got to speed, and became an effective, professional developer.

Another woman I knew was working at the front desk at a company, started working with the engineers and writing some stuff. Like, 10 years later, she's running an engineering shop as a professional, you know, she became an engineer, ended up running her own shop, like, very much so. That process of building something, and, you know, from whatever background you're coming from, learning that lore, building, building, building, absolutely, I’ve seen to be the case. And the people who are really effective tend to be people who've done it for a while.

And you're talking about doing it as a kid. I started doing some coding in...well, I had done a little bit, like, as a really little kid, but then I didn't have a computer [laughs], so I didn't do any for a while. But then, when I was in, like, junior high, I got a computer, pulled out the big manual, GW-BASIC, and started writing some stuff, starting from those things, and building it, and then I didn't do it for a while.

But then I came back to coding in college. It went much more quickly. It went much more quickly because, man, I'd done this. I had practiced. I knew what I was dealing with, and it just pushes it forward so much. I also did some...talking about fixing computers, I did tech support for Windows 98 and Me [laughs]. And you learn some things about how a computer works and how to solve things that serve you well in the industry.

WILL: Absolutely, yeah. And I remember selling a lot of Windows 95s. I remember like, yeah, I did a lot of weird stuff, a lot of weird stuff, because I predate Windows 95. I was doing, like, Windows for Workgroups, or before or earlier, a lot of like, MS-DOS type stuff.

MIKE: I remember Windows came out thinking, oh, do you need that [laughs]? I've been doing all this on the command line [laughs].

WILL: Yeah, yeah, yeah. I don't know. And there's a big piece of me, I mean, this is a little bit of a tangent. But, like, I do wonder, to a degree, whether...I wonder how, you know, generations coming up right now, where, like, all the really big bad, like, gross stuff has kind of been watered down. It's been, you know, it's been tamed, more or less. I think there are a lot of computers that are effectively appliances, and there's really nothing to worry about.

If you’ve got a Chromebook, if you’ve got an iPad, you don't have to worry about all that. You don't have to worry about it, unless you want to be a professional, in which case, you know, here there'd be dragons. It's like when everybody opens up their car these days. You pop the hood on your car, and it's nothing but plastic shrouds. You can't see any of the weird hoses, and belts, and cables, and stuff like that. It's all in there, but you don't look at it anymore. You have to take the cover off, and it's like, oh yeah, it's the same nuts and bolts. It always [inaudible 22:13]. So, I mean, I just don't know. I don't know --

MIKE: You're touching something. You’re implying that you do know. You're implying that if you really want to learn this, that learning a language isn't necessarily, like, I mean, it's a part of the process. But you're talking about this lore that you had to learn, even more important than language, because the language of today is probably not going to be the language of 10 years from now. It might be. Java is still around. It’s been around for a while.

WILL: Yeah, Java is going strong. It ain’t going nowhere [laughter]. Well, and Kotlin, right?

MIKE: But it's not the number one anymore. There's new kids on the block. You know, there's Python, JavaScript, and I'm sure Python and JavaScript will be here 10 years from now. But there'll also be another new kid on the block, probably, and particularly talking about GenAI, you know, there's new tools, and that's always been the case. There's always been new tools, and those tools are going to change. And that doesn't change the fact that somebody who is trying to build stuff, and has grappled with hard problems and thought about how to address those problems in a systematic way, has developed skills that are universal.

WILL: Absolutely. There’s sort of, like, call it, like, cognitive, simple machines, like, where there's sort of processes that, like, I could explain to you in five minutes, and I will, and it won't even take me five minutes, but, like, foundational ideas. Like, always debug from the known to the unknown. Always debug from the known to the unknown. And if your stuff isn't working, dumb it down until it works.

I have had a professional day, where I contracted for my full rate, which is substantial, and they hate it, and they were happy to do it. And the butter would have melted in my mouth when I cashed the check, where all I did was make the damn thing turn on. That's it. It didn't do nothing. And that was [laughs] a hard day.

And everybody's had days like that, where you're just getting going. You talk about the lore. That's the lore right there, where it's just like, oh no, no no, just make it work, right? That is a dead, simple idea. Go from the known to the unknown, and if it's not working, dumb it down until it's working. Make it stupider. Like, how much stupider? As stupid as it takes, and then you move forward from there.

I could get a tattoo of that [laughter]. I could write it on my body in indelible ink. And you learn, sort of, like, simple cognitive machines like that conceptually, intellectually. It's a lever, a pulley, things like that. How do you break down these problems? I don't know. I don't know. I mean, nobody ever taught me any of this stuff. A lot of it was just, like, I just had a knack for it from day one.

MIKE: Well, so you say you had a knack. I'm sure there's some truth to that. Well, you say you sat around with a bunch of gray beards [laughs]. I think that software engineering has historically been an apprenticeship industry, and I think that's underappreciated. I think I've said it before on this podcast. I think it's true. You know, a few hundred years ago...there's a reason the most common surname in the United States is Smith is because there's a whole lot of people who are Smiths. There was a whole lot of need for metal workers who could, you know, a kind of cottage industry, who could take the metal and make it into important implement.

And you learn that by going and spending time with somebody who did it and working with them for a few years, maybe 10 years, for some extended period of time, looking over their shoulder and learning all the things that you learn by just being there. And I feel like software engineering is very much the same way. We try to short circuit it, and you think, oh yeah, we can teach this in school, hmm, or, you know, bootcamp. And there's some truth to that, but I think a lot of what you're learning in those places is coming from trial and error and watching the people around you.

WILL: I mean, the principal problem of teaching programming in schools is, I can teach you programming if I know it, but if I know it, do I really want to be teaching a bunch of high school kids, ehhh, you know [laughter], I mean, they pay me pretty good to just do it. And that's problematic if you want to teach people.

I think I've said this before. I think I've said this before on this podcast, and I don't really think of programming as a four-year bachelor's degree type of subject. I think if you graduate from school with a bachelor's in computer science, you have proved yourself to be smart enough and dedicated enough to be trainable [laughs]. The industry, historically, has been willing to absorb the sort of finishing costs to take this raw [inaudible 27:40] and turn it into an actual developer. And I feel like that appetite might be waning.

But if you compare analogous professions in terms of like, skill, you know, which I will make up out of whole cloth in my head, I think what you're expected to know is a lot more like somebody with a master's degree than a bachelor's or a high school diploma, or something like that, you know, all this stuff. Because it's an apprenticeship thing, you could do with a high school diploma or not even. You just have to sit with somebody to fill your apprenticeship. And companies, large companies, especially, they can afford to post that sort of a gate on you.

I'll say this, too, if we're worried about apprenticeships, you know, it's like, I'm a junior developer. I need an apprenticeship. You guys just took the graybeard’s advice on faith. And it's like, you got to have an apprenticeship, you know, to really learn the trade. Well, how do I get an apprenticeship then? The three little magic words: sketchy startups [laughter]. Sketchy startups. Sketchy startups. They're going to pay you in Amazon gift cards. Like, it's going to be in somebody's house that smells funny. Like, you're going to get a paycheck that you can't necessarily cash on the same day [laughter], like, all of that stuff, all of that stuff. But you will get your apprenticeship in.

And this is, like, one thing. There's a lot of stuff that you can and probably will have to let go, like a W-2 form, or health insurance, or anything like that. But the one thing you can't afford to let go is you do have to have at least a journeyman to be learning from. Like, there are a lot of sketchy startups who are like, uh, we don't have anything. You, did you get a web page up one time? You're the CTO/co-founder.

And you got to have somebody better than you, like, a couple of steps ahead of you so you can learn from them, or else it's just going to be a disaster, like, a bad experience for everybody. But, in the end, yeah, sketchy startups, man, so many people, so many people have gotten their start on some, like, really, rinky dink fly by night what do we even sell startup.

MIKE: Another wonderful thing about said startups is you wear a lot of hats. You do a lot of things, and you get to work in a lot of aspects of the business that you're not going to get so much in a large enterprise [chuckles].

WILL: Absolutely.

MIKE: You're setting up the infrastructure [laughs]. You don't have a DevOps team. You are the DevOps team and the engineer. And you've got to do both, and you learn it, or else it doesn't get done [laughs]. Having been at multiple places where they laid off the rest of the team [laughter] and I was doing everything, you learn a whole lot that way, when it's you or nothing.

WILL: That's the truth. But, I mean, it is pretty sketchy. It's pretty intense to go in without any sort of support, especially when you're just a baby dev. I think it's difficult, and when you find yourself in situations like that, the fundamental trade off you make is a lot of people...it's a really high risk, high reward situation, where there are a lot of people who would have made it if they just had anything, anything backing them up, anything at all. They would have made it, but they had nothing. And so, the ones that make it through those guys are going to be serious.

MIKE: Sure.

WILL: But you're going to find a lot of people who, ordinarily, they could have made it, you know. They could have done it. They just didn't have anything, you know. I believe in taking risks and stuff like that, but I don't believe it takes stupid risks. And I think you can get yourself in a bad situation like that.

KYLE: That's a good point, where you can’t get discouraged too fast, like, I don't think I took a necessarily traditional approach. I wanted to go straight into dev, and that backing that you're talking about, or that support, I don't think I ever got it. It just was not available to me. And I ended up finding myself in QA. I found myself in QA for seven years before I really found my niche in DevOps. And saying that, I'm saying, you kind of have to keep at it like. Maybe you don't get the opportunity, you know, right away. But if you stick around and it's something that you really want to go for, maybe the opportunity will present itself. Like, not everybody gets it at first try.

WILL: Oh, absolutely, yeah, completely.

MIKE: Looking for work during the dot-com bust was not fun [laughs]. That was not a great time.

WILL: Oh, it was so rough. Yeah, I worked at a gas station. I worked at, like, straight up, I worked at a gas station, like, I was the ninth shift manager at a gas station.

MIKE: [laughs] I ran up my credit card bill pretty high [laughs], and I very nearly went back to doing construction, because I made money doing construction.

WILL: [laughs] It’s the truth, man, yep. Yep.

MIKE: But on the other side, it got better, which, you know, kind of going back to the theme, you notice how we're talking very little about the language. We said, what did I wish somebody had taught me? I think that you should learn a language. If you want to learn to code, yes, go learn a language. You might not code in it in your job; in fact, good chance you won't, but you're going to learn some things. You are going to learn some things about yourself. You're going to learn some things about the computer. You're going to learn some things about the computer you wish you didn't know [laughs]. You're going to learn those simple tools.

You're going to learn, maybe if I cut this problem in half, then I can very quickly do a binary search and get to the root of the problem. You're going to learn how to write a configuration file, which you will do an endless number of times in your career. You're going to learn how to do all of these lore sorts of things that change your perception of the world. You're watching the master blacksmith banging on that metal, and you're learning what the colorless sparks mean. And it's going to change you more than the language.

Although the language is going to be important, you got to learn something. You need to go out and start building something. And I will express that strongly. I think you said it really well in your monologue [laughter], Will, that by building those things, you're going to learn a different set of skills. And, you know, I came into this, this podcast thing because I wanted to express something about you got to learn the language, but that's not the most important thing you're going to learn.

WILL: Like, I almost think...one thing, and this is something that I’ve thought for a while, maybe I'm just out of touch, but I really feel like...how do I put it? In the age of generative AI, ChatGPT, and stuff like that, like, I've never thought all that much about programming languages, and I plan to think so much less in the future, because, like, what’s a language? A language is a way to express a series of procedures, a series of thoughts, very precisely, which human language does not do, will never do, but, you know, sort of even a high-level programming language. It is a language to express logic, procedures, stuff like that, you know, really specific way.

And ChatGPT, generative AI is amazing at translation, the translation of ideas across, you know, from English to Spanish, from Spanish to Japanese, to Korean, to Smalltalk, you know, whatever you want to call it. And so, you know, language Kotlin, Node, Python, Ruby, Swift, whatever, the ideas are important. The language is nearly meaningless at this point. Maybe that's just me, because I understand the concepts really, really well, and I know exactly what I want. I know exactly what I want. And so, how do I say that in Python? I almost don't even care, you know.

MIKE: Will, I'm suggesting that you've got to pay your dues, that it is helpful. It's helpful to take that stack of CD-ROMs and of source code and compile your operating system. Because you know what? CD-ROMs don't matter. And that particular build of Linux you probably used for a little while and forgot about [chuckles], you know, the specific problems you were solving were irrelevant. But you had to do some work grappling with it, and you have to learn something, I think.

You know, we all, as a child, we don't know how to speak, so we watch people around us speak. And I don't think it really matters what natural language we learn. You got to learn one, and then you can communicate. And then, if you need to go learn another language, that's what you do, another natural language. Likewise, here, you've got to learn how to speak. And I do think that you have to go through some work, and, like you say, work for sketchy startups. Go write a bunch of stuff. Go do a bunch of vibe coding, you know, go use a bunch of generative AI and see what you can build, and see how it's broken, fix it, and [laughs] push it out there. Work on something.

And it's that working on it where you learn those skills. So, I do think you need to learn something specific. You've got to drill down enough to be able to understand. So, if you're writing apps using generative AI, they're going to write it in the language. Underneath, there's an interpreter, compiler. There's pieces that you're still going to have to understand because it's going to do it wrong, and you're going to have to go and learn...say it's Python. You're going to have to go and learn Python enough to fix its bugs to know, hey, this is wrong. Where did it go wrong? And go and fix it, or, you know, ask it to rewrite it again, figure out which one looks best.

You're going to have to learn the underlying structure as you go. I think learning one matters. I would recommend that. Learn one thing well before you generalize. Maybe not perfectly, but I do think the learning one thing really well does help you generalize. Going back to what we talked about at the beginning, [inaudible 39:22] one thing further.

WILL: For sure, for sure. Make something, build something, and put it out in the world. Make something and put it out in the world. Vibe coding is great. I am pro vibe coding, you know, like, just making things. I love making things. Like, that's the foundational thing. There's a big piece of this where it's like, if the idea of making things just for fun, to scratch your own itch, because I want to make stuff, because I like making things, and I'm just going to do it and put it out there, put it out there because there are people who desperately want to find you, but you have to give them an opportunity, some avenue to reach you.

If you want to work, you have to find some way so that somebody who's searching for you desperately but they don't know your name could find you in some way. I think that's absolutely foundational. But if the idea of just making stuff just to make it doesn't really sit well with you, where it's just like, well, but who's going to pay me? There is a big question, I think, you need to ask yourself, which is this for you? Because that is the fundamental thing.

If you don't just love making stuff, then is this really a good idea for you to waste your one precious one-time-only life doing? I don’t know, there's lots of wasted, like, just make a bug. And that's fundamentally why, like, you know, ultimately, like, that foundational loop of making things and putting things out in the world it always works if you're just really into making stuff because not everybody is and, like, we could smell our own.

MIKE: [laughs] It is true.

KYLE: I would tack on, too, to go along with making stuff, understand how stuff is made or what is already out there. Because I feel like that's one thing that I missed going through my career is I didn't take the time to stop and understand what it was that I was coding against. Like, what was the underlying operating system like? How is this running on Linux versus Windows versus, you know, the other OSs? You know, just having that fundamental understanding underneath the hood just helps you understand what you're coding as well.

WILL: That is such an important concept. And this is one thing that I think we're getting very, very, very, very, very wrong in terms of the vibe coding, like, generative AI thing. Like, hey, man, I understand that we all want to make things. We want to push these things out in the world. We want to make things, and that's, like, the goal. But one of the other cognitive levers, the cognitive simple machines, is like, when you do something and you're like, I don't know how that worked, read the manual, man. Just read the manual. You did something, and it's like, oh, I'm just sort of trying, like, I’m just trying something. Oh, that worked. Whoa, whoa, whoa, what did you do and why?

Like, taking that step back, that is another one of those intellectual, simple machines that you really need to master. If the other one is a lever, then this one's a pulley, where you integrate these things, and you get a larger understanding of how stuff works. That's why I don't care about languages anymore, because I've done this so many times that I understand the foundational things. Like, oh, we're doing a [inaudible 42:53]? It's like, oh yeah, that's a thread, but it's got rules around it. I'm like, okay, that's cool, you know, because I understand how that works.

And, like, you know, oh, promise. It's like, oh, this is how a promise works. This is how an async function rewrites my JavaScript with all these freaky returns and stuff that it puts in there, where I'm like, okay, these are the rules for taking this async promise away thing and turning it into what appears to be an imperative function, just all that kind of stuff and, like, that's another one of those fundamental things. And it's one of the things we get so, so wrong about generative AI, because generative AI goes both ways. Like, not only can you ask it, “How do I do this?” You can ask it, “Whoa, what the hell did I just do?”

KYLE: Yeah, we need to do that more often.

WILL: Yeah, we need to do that way more often, way more often. You're talking about the lore. That's it, the generative AI it could do that for you, just basic stuff, where it's just like, you know, like, oh, I Googled this thing on Stack Overflow, which is still really, really relevant, although some of those Stack Overflow answers are getting pretty creaky, where it's like, the accepted answer is, like, 10 years old. And I'm like, I don't know whether that's best practices anymore, homie. I don't even think that'll compile. That's from 2009. We're in a brave, new world.

Like, that Stack Overflow answer can drive a car [laughter] because we’ve moved on. But you can read it. You could feed it into generative AI and say, like, “What is this?” And it'll tell you, and, like, it works actually a lot better. Like, what the hell is this? So, it'll be like, it's this, and this, and this, and there will be references. And you should look at the references because sometimes it'll hallucinate [laughter], but the references are usually pretty good. There's all kinds of stuff out there.

I think it’s never been a better time to learn how to make software, learn to code, learn how to make things using software. I don't think there’s ever been a better time of your life. And I think for the people that have the knack and really want to do this, I think the doors are as open as they've ever been, really. I don't think it's that much harder than when I graduated, you know, college at the middle of the dark days of the dot bomb era, you know, and I had to work graveyard shift at a race track, sell my plasma in the hopes of getting into the industry.

MIKE: I did that, too [laughs].

WILL: I don’t get...Pro tip: drink a lot of water. I'll water that down. You're not stealing my vital fluids for nothing. They gave me the side eye. They're just like, “This is looking pretty thin, dude,” and I'm just like, whatever. I don’t care [laughter].

MIKE: Another pro tip: tums.

WILL: Tums.

MIKE: Calcium. They use an anticoagulant that they'll make you feel funny, but if you have lots of calcium, it counteracts it.

WILL: Interesting, interesting, okay.

MIKE: Totally off topic [laughs].

WILL: I mean, you know [laughter] --

MIKE: Make sure you’re giving blood, too.

KYLE: You need to know how to pay for it. You need to know how to pay for your education.

[laughter]

WILL: That’s right. That’s right. It wasn't for free. Yeah, it wasn’t for free. But, anyway, I don't know, I remain optimistic. Nobody is throwing their phone in the trash, you know, nobody's spending less time on the internet. I think I'm probably going to stay fed, and you can, too. There's room for you. If you're listening to this, get in here.

MIKE: Yeah. So, during the pandemic era, interest rates were really low, perhaps artificially low, and so big tech companies invested a whole bunch of money in themselves because they couldn't invest anywhere else and make money, as much money as they could by investing themselves. Interest rates go up, and they look and say, hey, I can take this cash I've got, and I can go make whatever percent off of it really consistently. I don't have to risk hiring a bunch of people and hope they build something that'll make me money, so they let a whole bunch of people go.

WILL: It’s true.

MIKE: So, there is a flood of engineers out on the market right now, and it makes it really hard to find a job. I actually read a data analysis just this week about this topic. And there is --

WILL: Send it to me, would you? Will you email it to me?

MIKE: Yeah. I think it was in The Washington Post, if I remember right. Basically, they found no evidence thus far that generative AI has had a meaningful impact on hiring in software engineering.

WILL: I believe it. I think there's a lot of weird stuff in terms of big tech. This is a little bit of a...no, it's a substantial diversion from what's going on. So, I work with a lot of people out on the West Coast. And what I have found is that big tech let a lot of people go, but it's having a very local impact, because big tech it's all on the coasts, you know, a lot on the West Coast, to a small degree, on the East Coast. Well, all these guys required people in the office, so you have all these people who are unemployed on the West Coast. But they live on the West Coast, and they need West Coast money. And they can't make Midwestern money, Utah money, mm-mm, that won't do because they live in San Francisco, and they have to pay their San Francisco mortgage.

And so, I feel like a lot of the unemployment effects from big tech losing a lot of jobs are very localized, where, you know, like, I’m pretty senior, and I get paid pretty good, but I don't get paid pretty good San Francisco money. Like, nobody's cracking half a mil. For me, that's not the dough I'm making. You can make some, you know, you can get on levels.fyi, and they'll tell you the Bay Area is a very lucrative place to be an engineer. But you don't get that money, you know, telecommuting from Idaho anymore. They kind of shut that down because they've been doing layoffs.

And so, demand is very high, and they can say, like, no, if you want Facebook money, you're coming in Facebook. You're going to have to get on the half-pipe and skate around every day, or whatever they do [laughter], you know. You're going to have to come and eat our sushi if you want that money. So, I don't know, it's one of those things, that very top tier, very expensive market, very, very, very high pressure stuff. Like, yeah, those guys are hurting pretty good. But I actually don't know whether it's so bad if you fly out of the country.

MIKE: For new hires, it's real bad. Yeah, it's real bad. I was interviewing recently for an internship position, and just the number of applicants we got alone.

WILL: Interesting.

MIKE: I think we posted it, and we had...I don't remember the number. I think it was over 300 applicants in one or two days for an internship position, not even, you know, like, a full-time, long-term job, just to get an internship.

WILL: For how many positions? For one?

MIKE: Two, yeah.

WILL: Two. Mercy, that's brutal.

MIKE: Yeah, it's tough out there for people starting.

WILL: So, I should start, you know, exploiting these [laughter] beautiful young minds for my sketchy startup. You’re saying, Will, start a sketchy startup, like, there are people to be trafficked.

[laughter]

KYLE: Oh geez.

MIKE: And on a more positive note here, it’s always --

WILL: That was great. I’m very optimistic [laughter]. I have so many sketchy ideas. I'm going to be a head in the box. You'll never see me. I'll just Venmo you sporadically [laughs].

MIKE: I think Will said it really well, not very long ago: the world needs people who can think clearly about problems, who can break them down and express unambiguously and in a well-structured way about what the problem is that needs to be solved. This is the problem I need to solve and solve it. That need is great today, just like it was every other time. People have been hiring software engineers for the last, what, 70 years [chuckles]?

WILL: Yeah. All the way back to Grace Hopper, man. Like, who's going to fix all this crap [laughter]? Yeah, fascinating. All right, I'll have to cook up some sketchy startup ideas.

MIKE: Oh, any final words?

WILL: No, no, that was it. That’s the final thought. Like, I’m going to start cooking up some sketchy startup ideas [laughter]. There are exploitable minors out there.

MIKE: Expect some job listings in the near future [laughter]. Until next time on the Acima Development Podcast.