[00:00] Peter: So it’s not the middle and the top that gets cut unless the top gets super efficient, right? You still need the middle. It’s the sort of intro that you don’t need anymore. Yeah, and so the new hires, as the efficiency improves, that’s the first wave that we’ll see a pushback.

[00:20] tuwang: Yeah, but the one observation I have is people who have maybe been here for a couple of years are the most nervous people who are trying to learn and adapt, and the new hires are more like, “Hey, I want to do my manual stuff. I don’t want to adapt.” That’s a very strange behavior.

[00:35] Peter: Well, because, again, the new hires—let’s just call them software engineers, right? So a software engineer is trained to never create software, but to have the ideas of how you create software, right? And so in that teaching, a lot of it is sort of—I would call it—I wouldn’t even call it software engineering, it’s low-level API deployment, right, within a stack. And so the AI can do that, though, and so they, you know, if you’ve…

[01:10] Peter: …never built a piece of software or even seen a piece of large software, what do you do? Yeah, and so the middles and the tops realize, “Okay, I know how to build a system. I’ve built a few systems. I’ve seen the inside. I’ve understood some pieces of that system. Now with AI, I know that efficiency is going to improve with this tool. So my question then becomes: how do I become a better designer?” And then the incoming were hoping to have a high-paying job doing very little and just implementing that API at a low level. And so that’s what’s going on. And then the seniors, I would argue, are like—they’re more the system architects, and they’re like, “Can I build the whole system? Do I need all these other people below me? Or can I build the entire system with like a relatively skillful AI, right?”

[01:50] tuwang: Yeah, yeah. So, yeah, the whole dynamics changed. How do you learn about the mindset? That’s exactly how the company is trying to do, basically.

[02:10] Peter: Well, it’s, again, so you just have to—I would argue…

[02:15] Peter: …it’s like an empathy question, right? So put your shoes as the company: what does the company want to do? And then put your shoes as the sort of senior technical lead, put your shoes as the middle. And if you can just think like: what is their daily basis? What are their responsibilities in the job? And then, you know, so like you can start with the first principle, which is: why does a company exist? Especially a public company, right? A public company exists to improve its shareholders’ value, right? And that—I’m not saying that’s good or bad, but that’s the high level. And so what is the CEO’s goal? Well, the CEO’s goal is to help the shareholder. It’s not to make the world better; it’s to make the shareholders’ profit better. Okay, so then the CEO has the C-suite, and then in the C-suite, it’s like—I would argue that the whole C-suite just cares about the shareholders. But now you actually have a product. So now you go to, so let’s say, a VP level or senior programmer, and they have a product…

[03:21] Peter: …that they’re in charge of, and their job is to improve, win in a market, right? So they’re in a different game. So from their perspective, they have to ask the question: so what does AI allow me to do in terms of my product? Yeah. And then you just start dropping. Okay, so now I’m a product lead, I have one sort of feature that is featured in this product, so what are we trying to do? Well, we’re trying to satisfy the next level up such that they—you basically, as you start moving down the ladder, your whole question is: how do I make the person who’s in front of me look good such that they want to keep me around, right?

[03:55] tuwang: Yeah.

[03:57] Peter: And it just trickles down. So it’s just an empathy exercise to like figure out what do they care about and why. And then you add AI in there. And all AI, I will argue, is it’s just a skill supplement, right? So you have this agentic thing that has certain skills, and then the question is: how good is it at what it does?

[04:20] tuwang: Yeah. So this idea of vibe…

[04:29] tuwang: …coding, which, you know, at certain simple apps and stuff, it seems possible. But as you build a more complex system, to test it and verify it works—like, you just… maybe someday the vibe coding will happen, but I think the hard part from the AI’s perspective is big systems aren’t available for it to learn from, right?

[04:50] Peter: Right. Like, so, where are you working currently?

[04:55] tuwang: Instagram.

[04:57] Peter: Instagram. So the Instagram stack, which is huge, is not available to the LLMs to even start to understand the entire system, right?

[05:10] tuwang: Yeah.

[05:12] Peter: So there’s some open source, like, let’s say Apache servers—that’s a pretty big open source project. But LLMs, they don’t sort of have this higher-level understanding that they can focus and they can say, “Okay, I can do this.” But to have a grasp—humans can’t even grasp these systems, right? So to be able to say, “I just want AI—yeah, ChatGPT, give me the next Instagram,” it’s not really doable. Yeah, it might be someday. But you have to have like…

[05:48] Peter: …the context window of the AI has to be different, and I’m sure people are pushing in that direction. But building a system is way different than just making a feature or an API, right? It gets very complex.

[06:02] tuwang: The way I—some of the things I’m building to use it is to look at a specific folder of code and then understand that structure. Yeah, helping people to learn better, maybe as a newcomer, as a new hire, to learn the codebase—that’s useful. Or sometimes we’re trying to use some sort of agentic workflow to learn end-to-end about one specific, let’s say, class or event. That’s useful. But then whenever I try to ask the agent to do broader stuff, that’s usually going everywhere and taking a long time, first of all, and then it doesn’t do a good job. So the best thing I succeeded in is writing up the task as if I’m writing for the intern, step-by-step with examples. That’s actually working pretty well.

[06:48] Peter: Yeah, yeah. Because you’ve designed it, right? You just…

[06:55] Peter: …have a low-level writing of each of the functions or procedures, right? So you’ve—and it’s the same with writing, right? So when you write it and you give it, “I’d like you to tell me this point, and then this point, and this point, and make an argument, a blend of these.” Like, you’ve arguably written it all; you just haven’t put it into fancy sentences that flow well. And it’s very good at that too, right? Yeah, because the human’s done all the thinking, and they’re just asking the sort of—I’m not going to say low-level tool, but the tool to do what it does well, which is, you know, generate from other existing forms that are very similar.

[07:34] tuwang: Also, I’ve heard one statement saying, for example, all the CEOs now will be the last version or last wave of managers, because all the employees will become sort of managing their AI stuff. You no longer need the hierarchy of managers.

[07:50] Peter: It’s a good question. Like, yeah, the whole thing—the thing is we can’t predict it, right? So all we have is…

[07:58] Peter: …the writers that previously were around, the science fiction, what is possible, right? And so they were the thinkers that sort of gave us these possible scenarios. And it—I have no answers, like, it’s completely… I can’t predict. From my perspective, I’m just trying to predict: what does a student need to know to thrive in the modern AI development world, right? And I can’t even answer that. So it’s like—it’s very difficult to imagine what like a modern-day company of, let’s say, a CEO and an AI—that means we all could be CEOs, right, in some sense, if it’s… So for all CEOs, is there enough products out there that could, that would, you know, like… Like, if the AI is so good, why does Google exist, right? Why does it need to exist? Because I could make my own search engine or my own AI. So I don’t—I don’t imagine that in the short term. And well, let’s ask that—like, you can do a lot of…

[09:14] Peter: …experiments, right? So what if, what if the AI could build huge software stacks? How does—what happens? Why would I pay money to buy a product when I could ask the AI to build that product for me? So then the AI company is arguably the only thing that exists, not the software system company. However, just because I can build it, I still need to run it on some sort of hardware, right? And it just doesn’t magically… So, like, there are layers of that service model, but I just—I don’t see it. I just don’t see it yet. But that doesn’t mean it’s not possible. I just don’t see it in the short term.

[09:50] tuwang: Maybe in the future, people become real estate investors or power investors, that’s the main thing remaining. Everything—what is really a software engineer? What are you really doing? You’re just talking to a machine on how to do computation, right?

[10:05] Peter: Yeah, that’s my entire… move things around and do processes. So it’s like, okay, but to design that and figure out…

[10:22] Peter: …how to do something with that, that will always exist, right? So just because the AI knows about linked lists, and then maybe knows about sorting, there’s still: what is the application I want it to be able to do, and how do I break that up? I don’t know if the AI innovates in that space. I have a friend, he—I think it’s a really good question. We haven’t run the experiment, but imagine we could sort of pull an LLM from, let’s say, four years ago, and we know the innovations that have happened, could you ask that LLM what technology will thrive in the next three years, to see could it create what happened? I think that’s a really interesting experiment, right? I don’t know if it—because it arguably has the corpus of our written intelligence, right? And does it know what the next innovation is, or is that a human that comes up with, “Oh, if I put this and this together,” which is always what happens, “then I can solve this problem for person Z”?

[11:22] tuwang: Yeah, so it’s also like…

[11:30] tuwang: …talking to myself maybe five years ago, “What do you want to build?” It’s probably broad, but not going anywhere because I didn’t have the experience I have now, or the things I trained over the past five years.

[11:45] Peter: Yeah, so it’s like, they are interesting, they’re wonderful, interesting tools, but they don’t—they just, they just change the world. And there’s a book I read—like, if we really think down to what AI is, it’s just prediction machines, right? It’s prediction. However, when you say prediction, it’s sort of very simple, right? Because if I write a task and then it predicts how to respond to that, what it’s doing is remarkable, but it is just prediction machines. And the availability of cheap prediction—we’ve really never had this before to say: how does it change what we do? How does it change the world? It has never happened before, right?

[12:28] tuwang: On that note, how has this changed the class dynamics? I know there are questions about how students…

[12:38] tuwang: …learn or treat exams, those aside, but how does that change how basically teacher and students interact?

[12:47] Peter: Yeah, it’s a good question. Again, we don’t know, right? That’s the hard part. So, I think it’s kind of silly to say you can’t use it. I don’t agree with that at all, because I think that that’s just not the right way. So, but then the question is: how do you use it? How do you use it well? And I think for most beginners, the AI is—it’s just blind trust, which is really not the way to use it. So it’s like, okay, fine, get it to write your essay. Sure, I don’t care. But are you going to read your essay afterwards and check and verify that what it wrote is—and I think a lot of people say, “Yeah, I will,” but then they don’t do it. And same with code, like, they just don’t, right? Because I think we’re just naturally lazy. So, that’s like—to get people to look at what the AI generates, that’s like step one in correct usage, right? Human-in-the-loop. But I don’t even—I think…

[13:56] Peter: …even getting a student to do that is very, very hard. I think it’s just like, if you remember back to when you were learning to debug, there’s a lot of like, “Okay, it didn’t work. Let me try this and hope.” Right? It wasn’t “think about it,” it was just, “Okay, let me just change this word. I don’t know why, but I’m going to try it and hope.” And sometimes that worked. And so, “try and hope” is the same thing with AI: “I’ll try it, and I hope it’s right.”

[14:26] tuwang: And so you get nothing out of it.

[14:28] Peter: Exactly. So, an early debugger, that’s our natural sort of… We like try and hope as opposed to think and figure out why it might be failing, and then go and sort of go deep into what line of code is the problem. So that’s one of the problems with the AI. And I’ll be honest, like, even a lot of my courses, I was shifting towards take-home and projects, right? I have to pull back on that because—

[14:50] tuwang: Yeah, because why wouldn’t you use AI, right?

[14:52] Peter: Yeah, why wouldn’t you use it? Yeah. So we did an experiment this last year, and I know…

[15:01] Peter: …for my digital system design course, the LLM gets about 87% in the course, 87 out of 100. So I tell the students, like, “I know it gets 87 out of 100, so then the question has to be: why? Why would I ask you to do this, right?” And it’s not just about getting the answers in the back of the textbook. It’s about learning how to be able to build these things and think about these things—that’s the exercise. It’s not that it’s doable, because we know it’s doable.

[15:30] tuwang: Yeah.

[15:32] Peter: And that’s—it’s hard to communicate with a student, because it’s a young mind, right? Again, sit in their shoes. What are their shoes? Their shoes are: “I need a career.” And there’s nothing wrong with that. But from their shoes, it’s like: “How do I get to the career? Do I just check off the list?” And it seems like that’s the path. And it’s a little scary now because I sort of have to say, it’s like, “I don’t know if we’re teaching you enough to do well in the engineering world, right?”

[16:06] Peter: …I don’t know if it’s enough anymore. I think you have to level up probably a little bit more than the last alumni year. I don’t know, again, I don’t know.

[16:16] tuwang: You taught me multiple classes. Maybe the technical writing is kind of sketchy with AI, but the embedded systems where we had to build hardware stuff, use the chips—would you say that it’s actually better with AI? Because I can now read specs maybe more precisely or faster and enable me to do manual stuff.

[16:38] Peter: Well, like you said, Shawn, I think the value of the AI from a learning perspective is it being able to explain other things to you reasonably. That’s hugely valuable, because otherwise, as a beginner who’s learning about something, you know, it’s really nice to have a little bit of direction: “By the way, this does this, this is why, this and that.” And otherwise, it’s just you, maybe your lab partner, trying to figure out what is this doing, right? And so there’s big benefits there. The overall hardware…

[17:13] Peter: …like embedded systems design. It’s interesting because, so the good news for embedded systems engineers and hardware engineers is the corpus of knowledge that exists in those topics is just less than what exists for software. So it’s not that the LLM can’t do it, it just doesn’t have a huge dataset to sort of look at and see a lot of the possibilities. However, even with Arduino, right, there’s a huge corpus, and so it can do a lot very, very quickly. Is it better? Again, we don’t—it’s still too early. We barely understand. For me, when I talk to other faculty, I’m like, “Are you playing with it?” Because the first question you have to ask is: how does it change the job of being a professor? What can you use it for? Otherwise, we have no clue how it’s going to impact what a student is doing, right?

[18:05] tuwang: Yeah, how is that going? Have professors or your colleagues been accepting about AI, or using AI?

[18:15] Peter: Yeah, you would think we should. We should be…

[18:23] Peter: …the most accepting. I still think it’s like it’s too fast and too furious, and I actually don’t think professors do a lot that AI can help with.

[18:34] tuwang: Oh, what do you mean by that?

[18:36] Peter: Well, so how much—how much am I actually doing engineering? How much building am I actually doing in a, let’s say, semester? It’s not much, right? So, more of what I’m doing is communicating and, you know… Yes, it’s mostly communication, and it can help in that, but I still—I can’t get it to communicate for me. And so because a lot of being a professor is like interactions with humans that are social and communication, it’s—it’s not that amazing for us. It’s really good when an administrator, or an “admetricator” as I call them, asks me to write a report or do something. I love it then, because it’s like, I don’t care about the report, you don’t really care about the report either, you just want to generate. Okay, let’s get in there and just do that for me because I already hated doing that anyway. Whatever…

[19:25] Peter: …you generate, I’m happy with. But for a lot of the other stuff, it doesn’t really help that much. And that’s not completely true, right? There’s all sorts of spaces it’s useful in, but I think what I realize is there’s not a lot AI helps in in the teaching and learning world for what I do, in terms of teaching engineering courses, or just overall, right? My day-to-day work, it doesn’t help a lot, right? Like, half the time for me to use the AI—and I might be… it might be a form factor. So for me to use it right now, let’s just say I need to write you an email, right? And I’m like, “Okay, I could write Claude,” or whatever we call them lately, “Can you write an email to Shawn?” Well, that takes me probably like five minutes to set up, it generates something, I copy and I paste it, or I can spend five minutes and just write it, right? So the number of times… because it’s not integrated yet into my UIs, it’s more painful, and so I have…

[20:39] Peter: …to walk—like all the work I do, I always ask the question: if I’m going to do this multiple times, how much efficiency can I improve by, let’s say, automating it or writing Python scripts? So I always ask that question, and there’s just not a lot of automation in my day-to-day work, so the AI doesn’t help.

[21:00] tuwang: No, I think that speaks to how I use it at work. People are basically my coworkers, right, who build internal tools. They try to make it so seamless you can use it through your workflows. But that’s designed for that. Outside of work, I ask ChatGPT about, “My toilet is broken, look at this picture, what should I buy?” But I think you would ask YouTube, right?

[21:26] Peter: Yeah, yeah, exactly right. Previously YouTube or Google search, now GPT. But that’s about it. It doesn’t really plug into my daily life workflows. It might get—and it’s getting close to, like, say if you’re going on a trip, I think it’s starting to get good at planning. “Could you… I’m going to leave on this day, I want…

[21:39] Peter: …to leave on this day, could you tell me the flights I’m going to need, how I’m going to get to…” Like, it’s getting closer to being that what, which is I think we all want, is a nice, useful personal assistant. If you had a personal assistant who could do all that and do it well, that gets kind of exciting, right? Because I can offload a lot of that thought to something else. But it’s just not there yet, right? And then I have to always check it, like: do I trust this thing? And so, a personal assistant you would trust, right? But the AI, you’re like, “Okay, does this flight actually exist? Did you book it? Am I actually going to get on the flight when I need to?” Like, yeah. And it—it will get there. I think it will get there, that’s for sure.

[22:30] tuwang: What would be the—the most optimistic, what’s your optimism about AI in the future, maybe as a last question on the AI topic?

[22:42] Peter: That’s a good question. Optimism on AI—it’s funny, because I just look at it as another tool…

[22:48] Peter: …and so it’s like we could sort of frame it as: what happens when the hammer got invented, right? Like, so I got a new tool, and then how does it change my—I guess it’s always the question, like: will it make it so I have more leisure, right? That’s pretty well like the optimism. Will I have more leisure and still be satisfied with what I do on a daily basis? That’s the optimist goal, right? So I make enough to live, support the family, and I get more time to play video game X, right?

[23:20] tuwang: Yes, that’s the win-win games. Cool, yeah. Switch topic to what you have been doing. I’ve been loosely following your LinkedIn—LinkedIn posts. So, switching topic to that: what is MSU, the certificate, I think probably hanging right behind you?

[23:44] Peter: Yeah, so there is a keynote—it stands for “Make Shit Up,” MSU. It looks like Michigan State University. Michelle King did a keynote at the Serious Play conference I was just at, and she was giving these out, and I just thought it was funny because I think that that’s a lot of what…

[23:58] Peter: …we do. So I was like, “I got a certificate, an MSU certificate.” My wife asked the same thing, she’s like, “Congratulations!” I’m like, “I don’t know if it’s congratulations, it’s just funny.” And yeah, so, “MSU: Make Shit Up.”

[24:16] tuwang: Oh yeah, I think maybe two years since—two years ago I started seeing your posts, then you had this website, your website, but a page called “Let’s Play.” So, yeah, I looked into it for a bit, and reading the quotes, you said board games would allow educators to experience role reversal and remind themselves what it’s like to be a beginner and learner again. And then when I read through the PDFs you had, maybe last year, I saw a few things. For example, learning to be competitive, learning about patterns, predictions, learning unwritten rules—that part is funny—and then the importance of scaffolding. To me, it’s like a knowledge map, yeah. So, reading all this makes sense. And of course, if I were to teach, I would put myself in that perspective, think about…

[24:59] tuwang: …these. But my question to you is: what’s the actual pattern you maybe observe through your years at university or other professors, maybe doing differently than this ideal behavior of teaching?

[25:12] Peter: So, again, the hard part about being a teacher—and it doesn’t matter what you’re teaching, that’s why I like games, is because it allows you to teach something to someone else, so you’re actually practicing. But it’s very like… a game is a small system that has a very… like the goal is to win, I can tell you what the goal is, and then I can teach you the rules, and then I can observe if you can execute the rules and play the game—and maybe not win, because that’s strategy, but so it’s a really nice subsystem. But the problem in most spaces of teaching is you have two people: you have the expert and you have the learner, and the expert doesn’t see the world the same way the learner does. And so the hard part in…

[26:04] Peter: …all of that is: how do I… I can’t transfer that knowledge, right? So all I can do is offer you, “Here is roughly how the system works, now go try it.” Because if you just listen to me describe the game and never play it, you haven’t learned how to play the game. So, like, the expert can’t see the world like the learner does. So what’s nice about playing games and being taught new games is you actually get to remind yourself as an expert what it’s like to be the learner again. That’s that role reversal. It’s like, “Oh yeah, I forgot. I forgot.” Like, I forget all the time, right? I’ll use vocabulary and I’ll be like, “Oh wait, does anybody know what I’m talking about when I use that word?” Right? Because in my field we’ll use that word, and if I’m talking with someone else who knows about it, we don’t have to—we just know it. But if it’s a beginner coming in, they’ll listen to that word, they’ll be like, “I don’t know what that word means.” And that word is literally a paragraph, right? That paragraph, you know…

[27:09] Peter: …and we can think of all sorts of, like, design patterns. So I could say, “Yeah, you just look at the design patterns for software engineering, and then you could find something for linked lists, and you can just solve this.” And it’s like, well, if they don’t know what the word “design pattern” is already, we’re at a mismatch; they’re just listening to gibberish.

[27:26] tuwang: So yeah, you sort of gave us this same lesson, maybe in a different flavor, at—I don’t remember the class name, maybe it’s just the other half of the embedded systems, where you ask us to present, get a device, take it apart, and explain what that is to other people, and the audience may or may not know what the thing is. Yeah, I think that—do you think that’s just more speaking to having empathy or mindset empathy?

[27:54] Peter: Yeah, empathy is key. And, you know, all communication, the starting point is: who are you communicating to? Right? That’s always the starting point, yeah. And if you can’t start there—and maybe it’s not empathy, it’s like just imagine…

[28:12] Peter: …what it’s like in their shoes. And then, I think the other big thing when we’re talking about presentations and communication is you can’t really transfer a lot. We always hope that everything’s heard, but the reality is very, very little is heard, right? You might remember three things at a presentation if you even pay attention, right? So if I can convince you to at least pay attention, I might be able to transfer one piece of information that I would like to to you. I might.

[28:44] tuwang: Yeah, right. This one specific thing. I attribute the benefit back to you. And can I tell you a story about this?

[28:54] Peter: Sure.

[28:56] tuwang: When I first graduated, I went to the company called Schneider Electric. And six months in, I’m in the position to teach a class. And what we call “class” is basically 20 or 30 people, and they are 20 or 30 years older than me, like gray hairs, all that. The thing I’m teaching is: “Hey, this is a new software system someone built, I’m teaching you how to use it so you can sell to your partners.” But basically, they are the partners, selling to their customers.

[29:18] tuwang: …they are the partners, selling to their customers. Okay. And then I had to really pick the specific thing I want them to remember, because I know the 40 minutes will just be gone and nobody will remember anything. Then, also thinking in their shoes: what’s their daily struggle? They probably don’t care about this 25-year-old guy teaching or saying anything. I tried to make it fun, make memory points, try to give them kind of something they’ll go back to from the class—like, almost doing all tricks, trying to deliver the message, and do it a couple times in a year. So that was kind of resonating with what you said.

[29:56] Peter: Yeah, yeah. Communication—it’s… And I’ll be honest, those—those presentations probably have a big impact on your career, right?

[30:04] tuwang: Yeah, yeah.

[30:05] Peter: To me, more so than what you technically developed, which is sad, but they care more because they’re like, “Oh, Shawn’s a—Shawn can speak, Shawn can convey, right? Shawn can convince,” as opposed to, “We know Shawn can build, that’s why we hired him, and there are many other people who can build…”

[30:17] tuwang: …yeah, there are fairly few who speak well, who can communicate.

[30:22] Peter: Yeah, yeah, yeah.

[30:24] tuwang: And the other thing from that class—this is not the same topic, but I just remember you were saying the best way to get hired is through relationships. That’s still true. But the thing about me is, I never know anyone big enough to introduce me, and never become big enough to help others.

[30:45] Peter: So, it’ll happen. You’re still—you’re still early in your career, yeah. And the reality is your network slowly expands. It’s funny because I hit in lots of spaces, but I don’t know fancy people either. But I don’t really want to, because they have different expectations. So, again, you’re employed, you seem to be enjoying what you’re doing, you’re doing well. You don’t need to know the big-ups. What are they going to—what are you going to get to do that you don’t get to do now?

[31:16] tuwang: Yeah, that’s true. You were asking, maybe not to me directly, but to the class: why do you…

[31:20] tuwang: …need a PhD? And I think you and I maybe chatted about: why do I want to take two majors? And that sits with me forever now. I basically talk to anyone who attempts a PhD: why do you want to do it? If you want a job, don’t ever do it. It’s a waste of your time, your life, five years. Yeah, yeah. So that’s one of the things I kind of took away from the class—maybe for good—a very good lesson.

[31:50] Peter: Well, as an engineer, computer scientist, or whatever, it’s like—yeah, we somehow think that more letters after your name somehow makes you better. And no, yeah, it might—if you want to be a professor, yes, you have to have the PhD to get that job. So that would be the only argument of why you would do it, yeah, right? Exactly. And then everything else is sort of like a university scam. It’s monopoly money. I probably said to you, like: no one cares. Honestly, like a company, what do they care about? They care if you can build what they need you to build, right? They don’t care that…

[32:29] Peter: …you’re an electrical engineer or a computer scientist. Yeah, those signals however are like a good—that’s why they sort of say they want it, because it’s a signal that says you’ve probably hit all the little pieces I need for you to build what I need. And it’s easy, right? If you have a computer science degree from Miami University, that signals there. So I know that someone has said, “Yes, Shawn showed up to class, he did his assignments.” Right? That’s the signal. But after that it doesn’t matter, like: can you build the stuff that we need built?

[33:04] tuwang: Yeah, can you add value to the company? To the company, it’s a very effective filtering system. They wouldn’t care who you are, they just care about if you can do the thing they want.

[33:15] Peter: Exactly, you want it. And it’s easy. It’s hard to hire people—like, it takes a lot of time and it costs a lot of money, so these signals are just… they make the system more efficient, right? If you’re an HR manager and you know nothing about what a software engineer…

[33:29] Peter: …is, and you’ve got like, probably—I don’t even know, I’m guessing thousands of resumes—you need these quick filters just to even say, “Okay, what subset should I look at?”

[33:40] tuwang: The other thing I did for my resume is I used my English name as opposed to my hard-to-pronounce legal Chinese name. Yeah, “Shawn” is much easier to get through, for that matter.

[33:55] Peter: Yeah, your name—yeah, it’s funny.

[33:57] tuwang: No, yeah, I went by Fan the entire time. And then when I started submitting my resume, I realized, “Oh, Shawn is easier for people to communicate.”

[34:07] Peter: Well, there’s probably a bias too, right?

[34:10] tuwang: There is a bit of bias. And I just don’t need to remind them, “Hey, I need an H-1B visa for my job.” That’s for later. I want to convince them I can do this first, then worry about that.

[34:21] Peter: Yeah.

[34:23] tuwang: Yeah, let me go back to one question about the “Let’s Play” topic. So reading through that, maybe for the learners, from the learner’s perspective: what do you think stops people from being a good learner as they grow? I thought about…

[34:36] tuwang: …maybe their priorities change, or maybe my long-term memory is full, I just don’t have enough disk space to put in, or just sort of a curiosity regression as I grow or get older.

[34:50] Peter: Well, learning’s hard, right? It’s just hard. There’s no—we don’t know how, it’s just hard. Some people love learning, and they’re the ones that don’t find it as hard as others because they really enjoy just learning, right? But in general, for everybody, you just—there’s certain ideas out there you just can’t walk in and, no matter how well it’s explained or how nice the YouTube video is, you need to like grapple with, let’s say, the algorithm of solution or the… you need to grapple. So, I’m doing—I’m really fascinated right now with Bayesian statistics.

[35:36] tuwang: What’s Bayesian?

[35:38] Peter: So, there’s—so, you, when you did statistics, it’s probably more frequentist, which is—it’s still probability, it’s all on probability. But do you remember Bayes’ rule?

[35:48] tuwang: I don’t. Okay, I think so…

[35:50] Peter: …so, a robot runs the SLAM algorithm, which is location, right, and figuring out what my world looks like: so where am I, and what does my world look like? And so if I take a measurement of that—there’s a wall there, and I see there’s a wall there, that’s my sensor, it says, “Ah, there’s a wall there.” Well, how much do I trust that there’s a wall there? Okay, let me take another measurement. “Ah, it’s still saying there’s a wall there.” So the probability that that’s a wall moves up, right? So that SLAM is based off of a Bayesian—I have repeated, and it’s kind of, I would argue, how humans exist, right? We have a model of how the world works, we do something, and then if the model stays the same, we reinforce that model. So that’s like a Bayesian versus a frequentist, which would say: if I had a full sample of all possible sensor looks at that wall, I know with the probability of 99% that it is a wall. So it’s just a different way…

[36:50] Peter: …of sort of using the math of probabilities. Yeah, so, and it’s funny because I can’t really explain Bayesian statistics because I don’t understand it well enough to really get it. But it’s just, for whatever reason, I think it’s super interesting, and that’s what I’m learning. But it’s hard to learn, right? It’s hard for me to learn that because I don’t want to do the sample problem at the end of every chapter. I’m like, “Okay, I get it,” and then later on I’m like, “Oh, I didn’t do that. So do I really understand that, or was I just thinking I understood it?” Right?

[37:25] tuwang: Yeah. One regression I have observed on myself, maybe two. One is—I never told you about this, but I love singing or performing, and then I used to be able to memorize the lyrics without too much effort. I just somehow memorized all of them. Now I’m struggling learning a song, learning anything, a song in that matter. And coming back to games, I still play the two games I always play: Minecraft…

[37:53] tuwang: …and Diablo—Diablo IV, yeah. But then I no longer can learn new games. I kind of get sleepy when I try to learn new games.

[38:05] Peter: So that’s—maybe the regression is motivation, right? You’re not motivated.

[38:10] tuwang: Yeah.

[38:12] Peter: And then number two: you’re not practicing like you used to, yeah. That’s the reality, right? Like, our memories don’t get worse, we just don’t practice memorizing stuff. If you don’t practice it… So you probably were listening to lots of different music and then memorizing them, not even noticing you were, right? Right now, you don’t listen to different music; you probably listen to the same what you liked when you were whatever, and so you’re not in that exercise of practicing memorization. And then for the game one, it’s like… you know, yeah. Actually, I just got into Minecraft, it’s pretty cool.

[38:45] tuwang: You just got into Minecraft?

[38:46] Peter: With my son, it’s a lot of fun. However, I’m like, yeah, we just went to the Nether. Is it the Nether?

[38:54] tuwang: Nether, yeah.

[38:55] Peter: Nether, yeah. And I was looking at how to go to the Ender Dragon, and I’m…

[38:57] Peter: …like, “But there’s not—it’s really just a sandbox otherwise, right?” And it’s like, I keep on asking, I was like, “Well, what do you want to do? Do you want to go to the dragon?” He went to the Nether, he’s really excited, and he dies and he loses all his diamond stuff. I’m like, “Yeah, next time you go, let’s learn about it instead of just rushing in there with all our good stuff.” So, but it’s a sandbox. It’s like, do you want to build a farm that, like, generates the food? And then, do you want to build the thing that protects you from the mobs? It’s a really nice sandbox. I—I think it’s remarkable.

[39:35] Peter: Diablo—not… I played like the original Diablo. For me, I’ve gotten into Helldivers—Helldivers 2. I’ve realized what I… I like first-person shooters, but I don’t like playing first-person shooters against other people. I do like playing against an AI. I really enjoy a bunch of us getting together and playing against an AI. So, yeah, I was similar to you, not finding any new games. But then, I—I had this moment. I was like, “Oh, this is a lot of fun, playing with your friends and shooting, yeah, artificial intelligence, right?”

[39:54] tuwang: That, I think, goes back to childhood. I didn’t enjoy going over too many games, but I do enjoy when a crowd of friends, maybe shooting at each other, or doing something together. And that’s the thing I no longer have—couch co-op is dead.

[40:17] Peter: What’s that? What is that?

[40:19] tuwang: Couch co-op, where you are all on the same couch playing against each other.

[40:22] Peter: Yeah, yeah, yeah. It’s sort of—it’s just now, it’s all because of the internet. It’s play at my couch with you at your couch, and it’s just a little bit different.

[40:31] Peter: But PvE—player versus enemy—I’ve—I’ve realized I don’t need PvP ever again. Like Fortnite, it’s just… I don’t want to be destroyed by a 13-year-old, and they will always beat me because they have the time and I do not. So, it’s like, but me with my friends playing against an AI, and we can set the AI…

[40:53] Peter: …to whatever level we’re at. I think it’s wonderful. I really enjoyed it.

[40:57] tuwang: Yeah, Minecraft to me—I got into the game, I think, maybe the second year or third year at Miami. To me, that’s fascinating because daytime I’m learning tools to build stuff—maybe that was the same time with your class—and when I go home or have time to play, it’s the same mindset that I build small tools, then build a bigger, little civilization, right?

[41:20] Peter: Yeah, you’re trying to set up your—your little farm or whatever.

[41:22] tuwang: Exactly. And then there are certain plug-ins that allow you to build bigger machines. You still have to craft the components and do all the—all the switches, the redstone. The redstone switches, those are amazing.

[41:33] Peter: Yeah, we’re just starting to get into the redstone, and I’ve seen full, like, ALUs built with redstone, and I can see how you would do it. Because it has the automatic switching component, which as soon as you can build a transistor, you can arguably build anything, right? It just takes time to get all the redstone and…

[41:53] Peter: …craft it and…

[41:55] tuwang: Exactly, yeah. I think years ago people built a monitor, a computer screen, to do certain things. The whole screen flips for a while and then shows up to the right number. It’s the same thing we’re doing at school.

[42:07] Peter: Yeah, yeah. No, it’s really cool. Yeah, I appreciate it way more than like—yeah, I think it’s a lot of fun, too.

[42:15] tuwang: And one last thing—I don’t know if you ever intended to teach, but the non-apologetic attitude on questioning “why” goes into class. As you’re questioning how we come up to the final answer sometimes, or about why do you need a PhD, I took that and somehow can use it now, and I don’t feel sorry about questioning why, trying to push things. So, I don’t know if that’s intended, but that’s one of the things I learned.

[42:41] Peter: That’s so funny. Like, I will argue university is—it’s not about the technical, it’s not about the engineering. What it really is, is an opportunity for you… like, if you think about how…

[42:56] Peter: …all humans exist. We just exist, right? And we’re not that complex; we’re relatively simple, and it’s hard for us to deal with. But the world is so complex. And so, what sort of formal education gives you is like little pieces of trying to understand the greater world, little, little small pieces. And that’s—that’s what it is, right? Why—why does this happen this way? Why? And so—and then the other thing university allows you to do is, it’s the first sort of moment you realize that you get to write your own story in the world, right? It’s your… it’s like, “Yeah, my parents told me to go here. Yeah, my family said I should go to this school, and this is what would likely happen.” But it’s the first moment where it’s like, “Okay, there’s this separation where it’s like: I get to write what I will do in the world, whatever that is, good or bad. This is my choice, and I get to define it.” And in those two opportunities, I think you have to ask: why am I doing this? Why is this institution doing this? Why…

[44:05] Peter: …you know. So they’re just… If you don’t ask, if you’re not curious, then you just do, and there’s nothing wrong with that. I actually don’t think there’s anything wrong with that, but I think a lot of us have questions. It’s like, “Yeah, does this make sense?” You know, I get why it’s the way it is—like, we wear ties because the Victorians didn’t like buttons. Okay, that seems insane, but that’s why ties exist. And it’s like, these “why” questions are always good, because everything’s sort of expected; you’re expected to do this, so there’s nothing wrong with asking: “Why am I expected to do this, and is it good for my community, is it good for my friends and family?”

[44:48] tuwang: My hypothesis is a lot of people who went to university, or have the driven mindset to do good on grades, they are probably submissive, or in some way really agreeable to the parents or the previous teachers on what they should do. So, there is more or less a golden path defined, so they just do the thing. That’s one thing. The other one is, coming…

[45:11] tuwang: …from the kind of Asian background, everything about it is to agree to the elder or the teacher, right? It’s: “Yeah, yeah, follow the path, and no questions.” So for me, it’s particularly hard, or maybe mind-opening, when you were questioning “why” on something that seems obvious to people. But then, that was good.

[45:34] Peter: Yeah, again, asking questions is good. Again, you know, our families and our friends—there’s certain cultures within them that we probably should respect, right? So, but you have to ask: why? Is it… you know, and that’s a little proposition, right? It doesn’t make sense for me to do what my parents tell me all the time, and that’s—you know, you’ve got to make… it’s a cost-benefit analysis, and it’s like, it might be and it might not be. It just depends on each person. So you’ve just got to ask those questions. Yeah, and it’s complex, right? We sit on the shoulders of people that have defined things for ages, right? And then we have these…

[46:18] Peter: …new technologies that are changing, so it’s like—it’s so complex. And if anybody—anybody says they know what they’re doing, it’s just an absolute… they might not think they know what they’re doing, and I think our brains have to tell us that, “Yeah, you’re—you’re doing good, don’t worry, you’re fine.” But it’s so complex that it’s just insane. Like, complexity is really… that’s probably what… sustainability and complexity are probably the two words that we should just have in our lexicon. Is this sustainable? Is—is the stock market going up infinitely sustainable? Probably not, right? And is it complex? It’s super complex. Don’t think you can understand the market; it’s so complex. And sure, we’ve tried to make laws, but it’s so complex. The universe is complex. And but yet, from our perspective, it’s like, “Oh yeah, gravity exists, that’s the universe, right? I know how it works.”

[47:11] tuwang: Yeah, the sort of controlling environments.

[47:15] Peter: Yeah, and we have to, right? Otherwise, we’d be just a gibbering mess because it’s…

[47:22] Peter: …so complex. So we have to be able to—our brains have to say we’re good and we kind of understand what’s going on.

[47:29] tuwang: Thanks for the time talking with me through different things, different topics. Just in case people want to learn about “Let’s Play” or learning about your work, where should they go?

[47:40] Peter: DrPeterJamieson.com. I think I—I bought that domain name, so that should exist for a while. You know, I exist on the internet, I guess. Search for “Peter Jamieson” on Google, yeah, that’s where all the papers are, and yeah. But that DrPeterJamieson.com has like everything I usually link up there.