Show Notes
In perhaps the final episode of The TTL Podcast, Andy and Mon-Chaio reflect back on their journey, relive favorite episodes, and ponder what the future might hold, for each of them individually as well as the podcast.
References
- n/a
Transcript
Mon-Chaio: A bittersweet episode. Andy, this is perhaps the final episode of the TTL podcast, or at the very least, the final one for a bit as we both go off and reassess what we’re doing and how podcasting and a leadership might fit into that. I think, I mean, I am still well in the muck, right. Of engineering leadership. While you are moving on, but maybe still thinking about it.
Andy: Yeah. I’m still thinking about it and I’m still thinking about how as I move on, it might fit in. So I have, I have ideas about getting people away from their computers and, and working with their hands to, to learn some of these ideas. So I do have thoughts there.
Mon-Chaio: Right. And I may travel to you for that. Uh, take the, uh, woodworking sabbatical, um, one week in the mountains or something. Actually, you’re not gonna be in the mountains,
Andy: I’m not gonna be in the mountains. I’m gonna be on the bay. Probably I’ll, I’ll be on, I’ll, I’ll probably be on Mor Bay.
Mon-Chaio: Okay.
Andy: Yeah, and, and there is actually a sweet part to this as well, Mancha, that I don’t know if you’re aware of this episode that we’re recording right now, is episode 100?
Mon-Chaio: No. Wow. Um, I was talking to Andy about how the universe was trying to maybe have us take a break. Um, I. It started with, you all can’t see my mic, but my mic has, uh, two rubber bands that kinda hold it in its holder. The first one broke a year ago, maybe. Uh, and just two episodes ago, the second one broke. And um, now it’s just kind of sitting on the metal part of the holder.
Andy: So we can’t, can’t touch. They also, that also happened to mine. We have the same mic, uh, and mine, mine actually broke months ago. It’s, I’ve been using hairbands to hold, hold it into the, into the cradle. Uh, they don’t do too well.
Mon-Chaio: No, I don’t imagine they do. And I actually, I travel with my mic a little bit, so I’m sure that doesn’t help. I actually travel with it out of the holder, but I’m sure that doesn’t help either. Um, so.
Andy: all that.
Mon-Chaio: You know, so there was the mic thing, there was Andy’s new gig. Um, my, uh, my computer died, uh, for a long time.
Um, and I was contemplating should I do this? Should I go without a laptop for a while? Because all the tariffs are high. And, um, I actually lived for a number of weeks without a working laptop. Um, I recorded one episode on a, uh, on like a 10-year-old laptop at one point. Um,
Andy: remember that.
Mon-Chaio: and then, um. Yeah. And then our, uh, our editing softwares, licenses coming up, annual licenses coming up, and then
Andy: to renew in just over a week.
Mon-Chaio: just over a week and then the hundredth episode,
Andy: Yeah. It’s all coming together.
Mon-Chaio: it’s all coming together.
Um, alright, so, uh, I think we decided that today we would try to put together an episode fitting of. The end or the pause of the TTL podcast and maybe we start by kinda looking back a little bit, Andy. It’s been quite a journey. I mean, from the, from the calendar timeline. I think we were what May 23 or July 23 or something.
Andy: It was, uh, well, let me just do this because I just put all of our episodes, all of our episode transcripts, the back catalog we had up on our website, which caused me to go through all of those, and it was 27th of June was our first published episode
Mon-Chaio: Of 23,
Andy: of 2023. Yes.
Mon-Chaio: And so we are gonna be at. Just, just before 27th of June and 25. So calendar time, two years, which is, uh, quite a, it’s a good chunk of time,
Andy: Yeah,
Mon-Chaio: but sometimes when I look at other podcasts that, uh, I’m like, oh, this might be interesting. And I see they’ve stopped producing episodes and I see that they went on for two years.
I’m like, ah, those hacks can even make it beyond two years.
Andy: No, it’s, it’s a lot to get it to two years. It’s a, it’s a lot.
Mon-Chaio: yeah. Especially the way that we did it, right? So calendar time, maybe not two years, but 40 minute episodes every week for two years. I think that’s a lot.
Andy: that’s a lot. That’s a lot. I did not add up how much recorded time we have.
Mon-Chaio: Yeah.
Andy: Well, given that it’s, this is a hundred episodes, generally rerecord about an hour to a little bit more and then cut it down to 40 minutes. We’re. We’re at probably around a hundred hours of recorded material that then got cut down to 40 minutes, so yeah.
Mon-Chaio: And then multiply that by 25% for the discussion time that we don’t record. Multiply that by, um, you know, 200, 300% for the reading time and research time. It’s a lot.
Andy: It’s a lot and it’s been fun. I’ve enjoyed it. I’ve, I’ve enjoyed almost all of it. I should say almost all of it, not all of it. There’s been times when I’ve thought, oh my God, do I have to read this paper?
Mon-Chaio: Exactly. Oh. And it’s been, I mean, I don’t know. It’s been something that we might’ve talked about for a little while, even before we started recording. Right? Like bandied around ideas about maybe blogs, maybe podcasting. I know that, um. Even, um, as recently as when I was back at Uber, um, people said, you know, you have to have a podcast.
’cause we would listen Now, I don’t think they listen, but, um,
Andy: say that and then they don’t listen. I’m, I’m not bitter. I’m not bitter.
Mon-Chaio: Um, but it has been fun. I, I, I’ve really enjoyed it. Um, and yeah, that’s why I think that’s why it’s bittersweet, because yeah, the sweet part a hundred episodes and for the most part, I don’t know, um. I would say of the a hundred hours of recording, I’ve enjoyed probably 90% of it.
Andy: Nice.
Mon-Chaio: remainder may be 50% of the other parts.
Andy: So in, in, in that a hundred episodes, in that a hundred hours of recording time or in that. 200 to 600 hours of reading and research time. Do you have any lessons, any, anything that stands out in your memory?
Mon-Chaio: You know, I think the big thing for me is read the source material, like be curious and read the source material. Now, I’m a little bit biased here because I. Andy, neither you and I came into this podcast, like, we don’t wanna read the source material. And then we’re like, oh, you actually should like neither.
None of us came. Neither of us came in that way. Right. But the podcast forced us to read probably more than we’ve ever read in these last two years. And we were readers, but we’ve read more than we’ve ever read. And even still, I think that, you know, you do some reading and, um. You learn like, wow, things are different than you thought, or things are as you thought, but for a different reason than you thought.
Andy: Mm-hmm. Mm-hmm.
Mon-Chaio: and being able to make those connections. And I think that’s the thing that I’ve been getting, that I’ve been seeing more and more of with not just ai, but I think AI plays into this, this social media hype cycle where everyone takes that 1% nugget of information and they wanna project it out. To the a hundred percent.
Um, and the details get lost about the why and the challenges and the struggles about implementing that. And everybody just wants to focus on the successes, right? And so I think that’s it. Like read the source material. It doesn’t have to be the research paper, but like I was talking to somebody about this the other day, like, go, sorry, let me step back. Um, there was an AI symposium where somebody from Microsoft came in and they said, you know, it used to be 1:00 PM to 10 engineers. Now at Microsoft we can do 1:00 PM to one engineer. And what I was talking to somebody about was I was saying, I have yet to find either that one engineer or those nine engineers I haven’t found either. Like, I haven’t found the one that’s like, yet. I’m the one now with the PM and I haven’t found the nine, which are like, oh, I used to be part of a team, but you know, now I’m not. And I know a lot of people at Microsoft and I was talking to a Google person and they didn’t know anybody like that. So like, go to the source.
This is the, um, the gemba walk thing, right.
Andy: Yeah. Yeah,
Mon-Chaio: go to the source and figure out what’s real. Instead of reading your, uh, PWC analysis and, uh, your Microsoft, uh, speaker who built his slides in 45 minutes using AI at the AI Symposium.
Andy: that’s, I, I like that. I, I, it’s not where I thought it was actually going, but I like that I, I. I really like the idea of, yeah, we’re hearing about all these amazing things that people are doing and that all these jobs are gonna be lost and yeah, accept what they’re saying. But then go and see, go and go and check it out.
Go and find is is there evidence? This is, this is the kind of like, if that’s a hypothesis, how can we test it?
Mon-Chaio: Yeah, absolutely. And to your point, like listen to what they’re saying, right? Because being a skeptic, which I have a lot of skepticism in me, and I think that that’s, um, there’s a default way that I respond to a lot of things isn’t super helpful either.
Andy: Yeah.
Mon-Chaio: say, eh, you know, that’s all fake. I know because of like, again, my one data point.
No, like, take that and like treat that as a hypothesis. Right. And really go to the people that are doing the work. ’cause sometimes I see people, uh, I, I don’t mean to rag on ai, but it’s just the most current hyped up thing in the industry.
Andy: It is all that is on LinkedIn right now. Every time I open
Mon-Chaio: everybody talks about like, yeah. Um, but I see people that are like, oh, I heard about this AI thing and I wanted to know if it was real. So I followed the, uh, the YouTube tutorial online and I was able to build this thing. Okay? But that’s not real. Like that’s, can AI help you build a YouTube tutorial That’s not, can AI help you build a product?
So go to people that actually build product, you know, and have customers and ask, right? And get a second opinion. Get a third opinion.
Andy: And, and, and find out not what they claim it’s doing, but what it’s actually doing. And, um, yeah, it’s, it’s, it’s a really useful skill to have. It was useful for us as we read all of these papers, as we had these conversations where we’d say, okay, well let’s find out what is the evidence that was behind this.
And not just read the abstract of the paper, but go in and try to figure out like, how, how did they come to that conclusion?
Mon-Chaio: Mm-hmm.
Andy: Based on how they came to that conclusion, I can generalize it or I can’t. I, I, I might be able to take what they’ve done and say, oh, that, that seems like that could apply everywhere.
Okay, these claims about AI or, or whatever else. And sometimes I find out that no, the thing that they’re claiming is very specialized.
Mon-Chaio: Mm-hmm.
Andy: It’s very specialized, and it’s highly unlikely that that would work the same somewhere else.
Mon-Chaio: Yeah, I think that’s a really great point, Andy. Yeah, because like as humans, we want to generalize. We just want to say, oh, you know, this was a, we we’re good at pattern matching, right? So this is the same. Um, but very often, or most often, I would say it’s not the same. Uh, I think my anecdote would be, uh, when you talk to, uh, consulting companies who want to build you something, um, the structure I’ve started talking to them about is.
I don’t wanna see in your proposal things you’ve built in the past. I almost don’t care. Like, oh, this company was like yours. Okay. But it’s not mine.
Andy: Yeah,
Mon-Chaio: Right. And I hopefully have provided you with some information, so can you tailor it to the information that I’ve provided you? Versus trying to make abstract predictions based on things that are dissimilar.
And to your point, the research papers, if you don’t read the details, if you just read the abstract, sometimes you can come away with, oh, this is so similar. Instead of you’re like, oh, uh, they only were, um, in, you know, Indonesia and they only talked to people that worked part-time and they only measured, uh, when they were on their mobile phones, but not when they were on their laptops.
That starts to give you a sense of, oh, well. How generalizable is this?
Andy: Yeah, and, and that same. Inquisitiveness, and I’m not gonna call it skepticism, I’m gonna call, call it inquisitiveness, I think is, is useful in another situation. And it, a situation I was in today. I had a, a person I was working with have an, had an aha moment. So I’m, I’ve been helping out, uh, a company with hiring and helping them interview people and make decisions about, uh, candidates for hiring.
And one of them is an engineering lead or an engineering manager, and I. I was sitting there with one of their other engineering managers who I’ve been coaching, and afterwards we did our washup and he said, he said to me, noticed at one point that it felt like we were having this very nice, uh, technical conversation about the proposal that the, the interview, uh, set up was, you, you, you’re gonna go do something, you’re gonna give us an architectural and design proposal about how could this be done?
And then we’re gonna talk our way through that and. He said he was really enjoying it as just a technical conversation. Then I asked something where he suddenly snapped and he realized, oh, this is an interview too. Because I wasn’t having just the conversation, I was challenging the person on their thinking, on their reasoning, on the implications of what they were telling me and that I, and he realized, oh, and I’m assessing the person on that. Absolutely. That’s the inquisitiveness. That’s the kind of you you want to go in there? You want to go in saying like. Hey, I think you might be a great engineering leader. Now, let me understand your thinking. Let me understand your reasoning because it’s that reasoning and that thinking. You might have worked at Amazon in the past, but I’m not gonna take that as evidence that you are a good engineering manager.
Mon-Chaio: Mm-hmm. Mm-hmm.
Andy: I’m, I’m going to go off of how do you think, how do you reason, how do you work your way through this? And to get that understanding, I need to inquire, I need to challenge, and I need to ask the questions.
Mon-Chaio: Well, and this is kind of the foundation of the podcast, right? This idea of you worked at Amazon, that’s a small sample size that may or may not be relevant. Sure. There’s some relevant bits to it. Absolutely. But were you successful at Amazon because you had a great team and you were doing pair programming and.
Or you had a great PM or you know, any number of things, or were you successful in a way that’s replicable in a different situation?
Andy: Yeah. Yeah. Are are, are you successful with the skills that are, the gaps that this organization has? Do you, do you fill those gaps or do you accentuate the things that they already do that are causing them to be missing out on these other things? So there’s a whole bunch that’s going on in these kinds of setups, but it’s all about that inquisitiveness to find out is it actually there?
Mon-Chaio: Mm-hmm.
Andy: So, Mancho, did you have a favorite episode?
Mon-Chaio: Uh, I pretty sure I did. Let’s see. Um, I really liked our remote work series. That was one of the earliest ones that we did. Um, but I really liked it. I felt like it was a timely topic. I felt like we did both research as well as anecdotal evidence. Um, I also felt like it was an episode, which challenges people’s thinking.
I know a lot of episodes challenge people’s thinking, but sometimes when you know, it’s a hot button topic. Um, it challenges people’s thinking more. So I would say that, that those three episodes there might have been my favorite.
Andy: Nice. That’s what I’m trying to think of. Like which ones is these? I think so a lot of the episodes were covering things that I kind of already had, like the psychological safety, the Dunbar’s number, a lot of those things. It’s kind of like those were interesting to go into the depth that we went into.
Mon-Chaio: Mm-hmm.
Andy: The one that I remember that I was just like, watch how seriously are we gonna talk about this?
And then in the end I was like, that was actually good to talk about. And I was just scrolling through and it came up and I remembered it. Can you remove punishment?
Mon-Chaio: Oh. Interesting.
Andy: and part of it is mainly, it’s a lot of it’s just because it was that idea of going through something that I hadn’t really questioned.
Mon-Chaio: Mm-hmm.
Andy: You had us question, you had us kind of say like, no, what, let’s talk about this. Where, where could this go? What does it mean? What are the implications of having it?
Mon-Chaio: Mm-hmm.
Andy: And, and so I don’t know. I, I don’t know if the episode turned out to be in anything great, but just the thinking
Mon-Chaio: Mm-hmm.
Andy: it forced. I, I liked that, so, yeah.
Mon-Chaio: Yeah. Nice. I like that. Yeah. Can you remove punishment? And I don’t think that was something that was on either of the tops of our minds. Certainly not yours.
Andy: No,
Mon-Chaio: but,
Andy: punish people all the time.
Mon-Chaio: but um, as we’ve given you a sneak peek previously, we have a Trello board where we have topics. And I don’t even know. I think that one may have not even been in there or it kind of snuck in or whatever. It wasn’t a, uh, part of our backlog of things that we wanted to talk about. So it’d be really interesting.
I might, uh, go back and try to, uh, recontextualize how that came about. I
Andy: I don’t even remember. We might have said in the episode how it came about.
Mon-Chaio: maybe, I’ll have to listen to it again. Well, what about, um. What would you have done differently, Andy, if anything? If we were to start this again, knowing what we know now two years ago.
Andy: what would I do differently? So the things I’m going through in my head, would I have hired in like a producer or a producer or an editor? Would I have, had us have more people on the show?
Mon-Chaio: Mm-hmm.
Andy: Um, would I have changed the frequency to it more often, less often? I don’t think I changed the, I don’t think I’d change the topics. I don’t think I would do that differently. But I think if I was gonna do it again, I think, I think that I might now in retrospect, be willing to put in that bit of money to have a producer who just kind of keeps it all organized
Mon-Chaio: Uhhuh.
Andy: because it would’ve opened up the option of getting more people on the show or, or things like that.
Mon-Chaio: Yeah, absolutely. I don’t know if I would’ve been willing to do that even now, knowing what I know now, but it would’ve been fun. Uh, I mean, I think, um, I know this podcaster that, um, has a producer that finds them guests all the time, and they’re well stocked with guests. Um, so. It would’ve been nice to be able to do that a little bit more, I think.
Andy: Yeah.
Mon-Chaio: But yeah, I don’t know that I would’ve changed the topics either. I think, um, I think for the most part we had a really good mix of topics. Um, things that were a little bit more leadery, things that were a little bit more engineery, um, you know, uh, kind of more academic model thinking. More open questions that we propped up with academic papers.
So yeah, I think, I think I liked that a lot. I don’t know what I would’ve changed. Um, maybe done more to find here’s where I might’ve spent more money to, um, grow the listenership.
Andy: And, and, and that’s part of what I was thinking with the producer is having someone who actually does the, a bit of promotion, a bit of. Because, because that’s the, that dropped off for us at one point. We were doing it for a little while, but then the amount of extra thinking that had to go into, okay, how do I promote this?
How do I do the little clip? How do I get it out here? How do I get it out there? Uh, yeah, that’s, I just didn’t wanna put time into that. Yeah.
Mon-Chaio: Right. Well, and you know, it was kind of the hypothesis experiment and we didn’t see any metrics movement, right? So.
Andy: So people know we did pay someone for a little while to help us with that. Just give us material, give us ideas, give us the branding, the, the, the branding that you see on the podcast came from that. Uh, and yeah, it didn’t, it didn’t change the numbers as far as we could tell it.
Mon-Chaio: No, it did not. And maybe this makes means we should reassess what we did for the topics. All right, well, um, maybe we should, uh, talk about what if we were to, what if we were to pick up from this pause? Like, we, we, we think that this is the right format. We think that this is the right topics. Um, but, you know, we’re not, we’re not gonna probably pick up materially from this pause for at least a year.
I mean, I may, I may do some episodes here or there. But if we were to get back together, what does the future look like here?
Andy: Are you thinking the future for us in this podcast or the future of more of this domain?
Mon-Chaio: I think a little bit of both, because I think the latter, I hope, would inform the former.
Andy: Yeah. Well, I think given that ai, AI is gonna take away everything from us,
Mon-Chaio: Mm-hmm.
Andy: What the future of this is, is that we feed all of these transcripts into an AI system and then give it prompts of, of topics,
Mon-Chaio: Mm-hmm.
Andy: and it generates, uh, our voices having the discussion. So we don’t even need to do anything.
Mon-Chaio: Yeah, but you know, prompting isn’t very agentic Andy. I think, I think what we write is an agent, like you were mentioning, LinkedIn is all ai, right? So we just write an agent that trolls LinkedIn posts, aggregate sort of topic of the moment, and then generates
Andy: Generates an episode. Yeah. Okay. Okay. Got it. Uh, but a bit more seriously, I think I, I think given that the, the future for me will be a. Fair amount, more woodworking future for you will be much more hands-on running an organization. I think what that means is we might have an interesting thing of kind of a bit of an outsider’s view, a bit of an insider’s view, a melding of different ways of thinking about leadership more in general than, than just, than just, um, uh, in tech.
Mon-Chaio: Mm-hmm.
Andy: So we, we could, we could always have it continue as like, well, let’s just talk about more wide ranging, interesting topics.
Mon-Chaio: Mm-hmm.
Andy: Um, bring in lessons from, from, uh, uh, from the trades, from hand work and, and things like that. And, and kind of see is it, is there something that we can learn from these things?
Mon-Chaio: That’s interesting. Yeah, I can see that. I can see us saying, look, it’s not just tactics and T leadership, it’s tactics in leadership. Right. Or tactics in human relations
Andy: Yeah. Tactics for life.
Mon-Chaio: And tactics for life. Yeah.
Andy: Tactics for to life.
Mon-Chaio: Yeah, I mean, I can definitely see that. Um, I think even in the technical realm though, things are moving, things are moving fast. Um, and I don’t know whether that means that things will change or that things are moving fast to stay the same. There’s a a, a famous Simpsons line though, which says, uh. People never change and then he’s like, or people quickly change and quickly change back.
Um.
Andy: so, so, uh, on that. thing that I’ve done recently is I’ve been doing all this interviewing, interviewing someone, uh, about using a one of these AI coding assistance, and I’ve dabbled in it. I did my vacation cast talking about the, the bit that I’ve done and things I’ve, I’ve experienced and, and my thoughts on how you should probably work with these systems to stay in control.
Mon-Chaio: Mm-hmm.
Andy: And I got to watch someone using it was Claude Code. And watch them as they, they tried to stay in control,
Mon-Chaio: Uhhuh.
Andy: uh, and they did it, they did it initially with kind of like a TDD setup. But, so I’m connecting this in because it’s about inquisitiveness. I’m personally, I have to admit, I’m fairly skeptical about all of this stuff, saying that these AI tools are gonna completely change how, how software is written. I think that they will be an interesting tool.
I think that they will end up a lot like IntelliSense or refactoring tools, um, that they’ll, they’ll be helpful, but they won’t do everything. and so I, I watched this with interest as this person went through this and the, the thing I noticed was that I think I was right. I, I think I was right. I mean, I, I don’t have huge amount of data here, but I think I was right that. The proper way of working with these tools is the very small steps guided by you being in control of what are the, what are the acceptance criteria, what are the tests. Now, it might propose the tests, but I think that as a human you should be much more involved in the tests than in the implementation code.
And as I was watching it, I was like, yeah, because every time that it, I could tell the person was taking a misstep, it was because they hadn’t paid close attention to what the tests were.
Mon-Chaio: Mm-hmm.
Andy: And the test that had proposed, I didn’t agree with 50% of them. It was like it produced a whole bunch of code, but most of them I was like, yeah, that one, that one’s nonsensical, that one’s illogical, that one.
And they would all pass, but they didn’t explain anything.
Mon-Chaio: Mm-hmm.
Andy: And that was the key is what I was watching, was that the person was a little bit missing, that the tests really have to explain the behavior.
Mon-Chaio: Mm-hmm.
Andy: So, yeah, everything old is new. Again, your tests are important. Small steps are important, but I think it’s keeping that inquiry going and, and seeing where it is.
I could even connect that to woodworking if you want. When, when, when you’re, uh, when you’re taking your, your, your finished surface down, you take it in as small steps as you can, even if it takes a bit longer, because once you mess up, it’s really hard to undo that mistake.
Mon-Chaio: Mm-hmm.
Andy: And I think, uh, there’s a lie in software engineering that it’s easy to undo our mistakes.
I think it’s actually even harder because that mistake is a chain of reasoning that has gone wrong and everything after that point will start having implications about how it happens, about what that, from that mistake. Once you start trying to unravel that mistake, and everyone’s hit this where it’s like, oh, it’s a simple bug.
It’s the problem is right here, this is wrong. And then you change that. And then you start noticing, oh, now this thing doesn’t behave right and that thing doesn’t behave right and this other thing doesn’t behave right. And eventually you just say, just leave the bug.
Mon-Chaio: Mm-hmm. Right, right. I’m scared to touch it now.
Andy: Yeah, I’m scared to talk, uh, because everything is gonna break if I change this.
Mon-Chaio: Yeah.
Andy: And, and the way you get away from that is just like in the woodworking, you, you take it in very small, simple, easy steps. When it gets complicated, you go wrong.
Mon-Chaio: Yeah. And I think that makes sense to me now. Um, I think what people like to do is they like to project and they say, well, that’s what it is now. But in the future, and throughout software engineering, uh, you mentioned this in your vacation cast episode, there are a dozen cobols of like, well, but in the future, this is what it is now, but in the future, right?
The future is so shiny and great, uh, where every problem is solved. But yeah, I think at least for now, those tiny steps are so important and so, yeah, I can see, you know, I think, uh, Google is saying that 30% of their code is AI generated. I can believe that
Andy: I can believe that. Yeah, actually I know I, I, my eyebrows lifted, but then I was like, no, actually no. I could believe that. Yeah.
Mon-Chaio: Now the question there, there’s a real big difference between AI generated and. AI generated. I, you can’t see my hands or my facial expressions on a podcast, but I think a lot of people, when they think about 30% AI generated, it’s like, oh, engineers are writing 70 or, or spending 70% less time.
Andy: Mm-hmm.
Mon-Chaio: But are they really?
Because if you’re doing it right, AI’s writing the 30% of the code, does it really take less time to accurately validate that code via inspection
Andy: Not from what I’ve seen. In fact, in fact, it takes longer from what I’ve been seeing.
Mon-Chaio: Because you don’t have the context, so you can’t take the shortcuts to trust things, right? When you are writing code, you can say, oh, I can trust this bit because I wrote this bit. And so when I inspect it, as long as I inspect this bit, I know this bit’s probably good because you, to your point, you have that reasoning chain, right?
But when everything is new, you have to read everything. Um. And so, but I think that’s what people think. They think, oh, well, that’s time that engineers no longer have to spend. But no, they do have to spend time. They first had to spend time prompting it. Then they had to spend time reviewing it. And so does it really save time?
I don’t know that it does. I can.
Andy: I can see it saving time in that exploration of a large code base.
Mon-Chaio: Mm-hmm. Mm-hmm.
Andy: But in, in the final, making the change, I don’t see it. I don’t see it. Saving
Mon-Chaio: Yeah. Or in the structural stuff, right? Like, I know I need a switch statement with nine conditions, like, um. Every time I write a switch statement, I always have to look it up. Like is it colon? Is it semi colon? Like do I have to put a break there? Is it like obvious? And um, you know, and maybe AI can help arrange them in the right order too, because sometimes that thinking about order, it’s like, oh, um, oh no, this one actually has to come first.
’cause I actually wanted to evaluate both these two before or whatever. So yeah, I think it can save time there. So I’m not gonna say it doesn’t save any time, but it doesn’t save what people think is 30% of the time. Right.
Andy: Yeah, I was gonna say, when you said it’s writing 30% of the code, uh, it, it, it’s a little bit like saying that IntelliJ is writing, uh, 60% of the code because most of what gets actually put in the final file, most of those bites. Came from, its suggesting the function or the variable,
Mon-Chaio: Right,
Andy: and I’m just accepting that.
Yeah. That suggestion is the one I want.
Mon-Chaio: right tab complete.
Andy: Yeah. Tab complete on it.
Mon-Chaio: Well in some of the most interesting, um, did you read that Google paper recently about, um, them using AI and how successful it was? Um, really great and actually very believable paper. So they used AI to take a huge code base and migrate from 32 bit inch to 64 bit inch.
Andy: Okay.
Mon-Chaio: And so they just had AI agents that would run through a percentage every night, um, do all the contextualization, generate the diffs, and then publish the diff.
And it was interesting. What they ended up doing is they had the AI go through the code base and generate tag sections like this section, a hundred percent confidence. It doesn’t need to be changed. This section feels like it needs to be changed, but needs human input. This section I made the change, and then I have the diff for the input.
Um. And by all accounts it worked super well. It took away like that tedium, right? Because when you talk about tedium, it’s not the coding tedium, like even the switch statement. I would say that’s a design decision, not a tedious work to be automated. But switching from 32 bit inch, the 64 bit inch absolutely is tedium
Andy: Yeah.
Mon-Chaio: and I, um, and I was not surprised at all that they found AI to like be brilliant in that use case.
Andy: Interesting. I wonder if that means that it’ll be really good at doing safer library upgrades.
Mon-Chaio: Mm. Mm-hmm.
Andy: ’cause you sometimes you have that tedium of what are all the changes that went into this library? I. What are the thing, how is this code base using the library? Which of those changes in the library for this version impact the, the things that we’re using?
Okay. And kind of all that.
Mon-Chaio: And Yeah, absolutely. Um, and I was thinking about things like API upgrades as well. I mean, that’s kind of library changes, but, um, you think about, um, a outside dependency that you use, OpenAI changed their API, right? Like I need to upgrade my entire code
Andy: Yeah. Yeah. There’s not really any design decisions going on. There’s, there’s checks against, is there a behavioral difference if we make this change? Okay. Yeah. I could see a system being put together that could really help with
Mon-Chaio: Mm-hmm. And look,
Andy: that That would be amazing. That would be, that would be so nice.
Mon-Chaio: yeah, I mean, it’s not, it’s not a ton of our time, Andy. Right? Like when we were writing code, what like. Not 10% even of our time would we say we’re spending in tedious activities like that. But when we had to, oh man, was it boring and error prone and I mean,
Andy: Yeah. And, and then, and then your brain switches off and that’s when you make mistakes. And I think. It fits. It even fits with my philosophy of what, one aspect of a really good software engineer, which is that you find that time when you are doing that tedious activity and you automate it.
Mon-Chaio: Yep.
Andy: Our job is not to be doing the tedious activity.
Our, our job is to get the computers to do these tedious activities for us, and if we now have a tool that can pick up some of those tedious activities, that is amazing.
Mon-Chaio: Yeah. Well, and then there’s the ex, uh, education of so little of it is tedious. It’s not just your smart developers do the non tedium and there’s so much tedious activity to do. Um, I had this funny conversation where they were like, well, can we just have more doers? Um, and I used the household example. I was like, well, it’s not like laundry is the tedious activity, and cooking is the non tedious activity.
It’s like there’s a little bit of cooking, there’s a little bit of laundry, there’s a little bit of like making the beds. There’s a, so then your tedious guy that you hire, because they don’t have the thinking skills to do the other stuff, they would have to understand every single domain in your entire code base,
Andy: Yeah.
Mon-Chaio: right?
Because it’s not just one domain that has the tedious activities. Um. So they, they seem to understand that, but there’s, there’s a lot of education I think that goes on around that as well. Um, yeah, in any case, I mean, maybe that’s the next version of the podcast, is you bring on an AI skeptic and you bring on an ai, uh, uh, proponent, and you have them pair program together on the problem that they’re trying to solve. Right then I don’t have to do anything. I just like have them record their
Andy: match style.
Mon-Chaio: Oh, man.
Andy: Okay. Yeah. Uh, I, something, something might come out of this, something might come out of this pause and then we’ll pick it up. Um,
Mon-Chaio: But look. We don’t have to have long goodbyes Andy, or long pauses or long goodbyes to the pause. I think we do want to say that one, we really enjoyed doing this, and two, I don’t think we would’ve enjoyed it as much if like there were three listeners, right? If it was like my mother-in-law and somebody at your coworking space. Um, I really think it was the listeners, the questions that we got. Um, the people that came to us and said, look, I find I found value. That was an interesting topic you talked about.
Andy: Yeah. The feedback that I’ve gotten is that they enjoyed listening because. Even if they thought that they knew the topic as we were talking, they would be like, oh, I hadn’t thought of that before. And I hadn’t thought of that. And every time I heard that, I was just like, oh, that is, that is heartwarming to hear that we are, even if we’re talking about something that they already know about, that just hearing a different perspective, hearing us go through it, hash it out, um, uh, debate it was valuable that they enjoyed, they, they, they, they found that.
Useful to them.
Mon-Chaio: Yeah, absolutely. I mean, so much of content out there is simply reinforcing things that you already know or already believe in, right?
Andy: That’s most of the content I consume.
Mon-Chaio: Yeah, I mean, even the content that I consume, I mean, that’s what algorithms are designed to do. Um, so to hear people say, look, um, even if I already believed it, I still found new ways to think about the problem.
Or I didn’t believe it, now I do. Or the vice versa. Um, in other case, I think, look listeners, people that engage, people that enjoy us, people that supported us, those, um, those are really what made it fun because doing a hundred episodes of like no one listening and no one writing and no one engaging, I think we would’ve made it 20 probably.
Um, so. At least for me, and I’m sure Andy, like, thank you, all of you, everyone who listened, whether it was, you know, 10 minutes, one episode, or all a hundred. Um, thank you. I mean, it, it, it, it really makes a big difference for us to know that like people enjoy the content. I.
Andy: Absolutely. Uh. I’m terrible at goodbyes, so I will just agree with you wholeheartedly. Mancho, I, I don’t think I would’ve made it very far if every time I looked at our listener numbers, it was like two.
Mon-Chaio: That’s right when it’s difficult to do. Goodbye Andy, when I can see you crying and like, you know, you’re just
Andy: are not supposed to tell them that.
Mon-Chaio: alright, well, um, like I said, this is either goodbye or pause. Um, you know, uh, continue to write us though. Um, we still love hearing from everyone. Um, there are a hundred episodes out there. We would love for new listeners to listen, so feel
Andy: Still pass it around. Everything is still relevant. I don’t think there’s anything we talked about that was very time limited.
Mon-Chaio: No, absolutely. And I mean, I fall into this of like, I have to listen to the most recent episode of a podcast, but the content is all relevant, right? So, um. It’s like the, uh, when we were kids, NBC had the, if you haven’t seen it, it’s new to you for the reruns, right? Like,
Andy: I don’t remember that.
Mon-Chaio: yeah. Uh, I’ll reuse that. Like
if someone hasn’t heard it, it’s new to them. So please like. Recommend it. Um, we still, even if we’re not doing new episodes, would love to hear what people think because these, a lot of them are gonna be pretty, at least in the next five years, I think. Pretty timeless episodes of information. Um, yeah, so recommend us get back in touch with us.
Um, we still love to hear from people. Uh, and then I think maybe we’ll just sign off here, hopefully coming back to you at some point.
Andy: With our standard sign off.
Mon-Chaio: With our standard sign off, so until next time, which may be a bit longer than previously, but until next time, be kind and stay curious.
Andy: Absolutely Mancho.
Leave a Reply