Algorithms + Data Structures = Programs - Episode 265: 🇦🇺 YOW! Live 🇦🇺 Kevlin Henney & Damian Maclennan

Episode Date: December 19, 2025

In this episode, Conor chats with Kevlin Henney about the past, present and future of programming languages and with Damian Maclennan about YOW! 2025!Link to Episode 265 on WebsiteDiscuss this episode, leave a comment, or ask a question (on GitHub)SocialsADSP: The Podcast: TwitterConor Hoekstra: LinkTree / BioAbout the Guests:Kevlin Henney is an independent consultant, speaker, writer and trainer. His software development interests are in programming, practice and people. He has been a columnist for various magazines and websites. He is the co-author of A Pattern Language for Distributed Computing and On Patterns and Pattern Languages, two volumes in the Pattern-Oriented Software Architecture series, and editor of 97 Things Every Programmer Should Know and co-editor of 97 Things Every Java Programmer Should Know.Damian Maclennan is a technologist, software architect, trainer, developer, cyclist, and musician in Brisbane, Australia. With over twenty five years experience building software and leading teams across many industries he has worked as a developer, software architect, consultant, troubleshooter, trainer and educator, and senior leader. Damian is the Technical Director of YOW! Conferences. Full bio here.Show NotesDate Recorded: 2025-12-11Date Released: 2025-12-19YOW Conferences!ADSP Episode 190: C++, Python and More with Kevlin Henney97 Things Every Programmer Should KnowThe Past, Present and Future of Programming Languages - Kevlin Henney - ACCU 2025The Past, Present & Future of Programming Languages • Kevlin Henney • GOTO 2024TIOBE Language RankingsRedMonk Language RankingsProgramming Language RankingsContext Free YouTubeYOW! 2025 - Beyond Sonic Pi: Tau5 and the Art of Coding with AI - Sam AaronYOW! 2025 - Conceptualisation - Michael FeathersIntro Song InfoMiss You by Sarah Jansen https://soundcloud.com/sarahjansenmusicCreative Commons — Attribution 3.0 Unported — CC BY 3.0Free Download / Stream: http://bit.ly/l-miss-youMusic promoted by Audio Library https://youtu.be/iYYxnasvfx8

Now, this is the interesting question. What about AI? Well, in theory, I've just said that you can now do anything you want? Yeah, but do people do that? But it turns out they don't. They tend to follow the path that's already been set. And it turns out that AI does this as well. It actually reinforces it. Because all the training data is based on stuff that's already out there. So guess what? It's really good at the stuff that's already out there. Listen, folks, it's hot garbage, you know, whatever their ranking is, like, Fortran and Cobol are like the two fastest growing languages. We're still needed. Our insight is still needed. The robots aren't going to replace us just yet. And the fundamentals still matter. Welcome to ADSP, the podcast, episode 265, recorded on. December 11th, 2025. My name is Connor, and today I interview both Kevlin Henney and Damien with Kevlin and I chat about the past, present, and future of programming languages. And Damien is the technical director of the Yao conferences, and this is his first year in this position, and he gives us his review of Yao 2025.

All right, I mean, we've got 20 minutes. We might not record for the whole time. Maybe we'll cut this into the first part, and maybe we'll talk. to Damien later, but we haven't talked yet about your Yao talk. I mean, technically it was your Yao talk, your NDC Oslo talk, and then I saw it on YouTube at one point, so you probably gave it at ACU or some other conference, and fantastic talk as well. It changes, so depending on the version you see online, I mean, we'll probably link the ACCU one, maybe the NDC one, that one is, I don't, not sure if the recordings are out for that yet, they might be. And changes every time. And even right before we started recording this, I was like, I got to get set up for a sec. And

you were editing some slides and you're injecting photos all the time. So tell us a bit about that talk. And then maybe you can tell us about predictions. Because we didn't get to talk about that. The last time we interviewed you, one of the questions I had, and admittedly this was like a year plus ago now, it was kind of like, what is your, you know, you've given tons of talks in the past that cover different languages and paradigms. And, you know, how does the age of AI and LLM's impact, you know, in your opinion, programming languages and, you know, the languages that we're learning even if we're not typing in them. Anyway, so you can cover your talk and then future predictions. Okay, so the first thing is that, you know, the future impossible to predict it is,

so I'm going to borrow from Yoda there. So obviously there's certain things that can be wrong here. The title of the talk is one of the most prosaic and literal that I've ever done. It is called the past, present and future programming languages. I couldn't come up. with anything snap here but i thought you know here's something it it does it does what it says on the tin so there's a couple of aspects of that i look at the um i guess there's three three general parts to it one is i look at a number of different language rankings and without you know and i don't push i don't lean on the numbers too hard i'm just saying here is here's a language here's language rankings and they are they are made from different criteria so you know um

So, Tyobi, which is Conner's least favorite. But what's interesting about that is it is susceptible to web search and SEO. So that gives us kind of like an indicator of its potential volatility, but also its biases. I look at Red Monk, which is based on GitHub and Stack Overflow. Stack Overflow is not looking so healthy these days as a source of useful long-term data. And I also look at I-Triplea spectrum. And the point here is that I've looked at others. they each have their areas of focus and I don't push too hard on the numbers what I'm actually interested in

and one of the reasons by the way folks why the talk does actually update is that I will always try and include the latest version in that and I did have to do the version I have just done in Sydney uses different rankings to the ones I did earlier in the week and last week because you know you know when when a rank new ranking is updated I will incorporate it so the point there is to look initially at the top 10, the top 5, and say, well, have they moved much? And then look at the top 20. It's like, are they moving much? And are they, and then this is the interesting, first of all, they're not moving much. There is movement. It's not static, but it is gentle. The movement within them is gentle. The stuff in the top 10 is surprisingly stable. There is movement up and down. It's not, again, it's not static. But the really interesting bit comes when you look at the age of the languages and I divide them

initially I sort of say okay here's the languages from the 20th century and here are the languages from the 21st century we are a quarter of the way through the 21st century and 21st century languages are only just making a dent in the top 20

by any ranking okay and it is it's just like well that's kind of interesting and that did not use to be the case I then break down the languages and I say, okay, well, let's look at their overall age. And the fun bit is not just to look, you know, 1950s, 1960s, 1970s and whatever,

when was the language created? Look at it in terms of social generations. And the fun one is that there is a huge millennial bump. Most, the language space in terms of the mainstream of what people are using is dominated by languages that will be considered millennials. It's kind of an interesting sort of oblique observation, but the key thing here is that the languages are biased towards the old. So if you think about, so we've got Python, let's think about Java,

we've got C++, we've got, these are all languages that are quite old. See, yep, that's still there in anybody's top 10. And then we've got JavaScript, well, that's mid-90s. We've got a few clearly coming in, TypeScript is making, you know, that is relatively modern. And then I hear people talking about Rust and Go, and they're saying, yeah, but these are new languages. these are young languages. They're not.

You know, Go is 2009, it was introduced, okay? Rust is over 10 years old at this point. These are making some kind of dent, but the couple of things to be aware of is, first of all, when we look at any language ranking, as a human's we have a bias

towards any form of ranking. We think it's linear. We think of them in linear terms, but they're not. Number one is normally way more adopted than whatever number two is, and so on. By the time you hear, 10, it's just that it's a skewed curve. Not necessarily a power law, but some of them do look like that, but it's highly skewed. So it turns out that anybody's top five or top 10, whatever

ranking is highly biased towards that and has a very long tail. And that's really interesting because there is a lot of activity happening in the long tail. The tail is getting longer. There's lots of new languages to be create and new language is easier these days than it used to be. You can actually just make up a spec, shove it at an LLM and say, hey, you know, generate me something. or pretend you are the interpreter for this. But what's happening is that the core is consolidating. And there's a number of reasons for that. And it is interesting to look at the rate of change.

So if we go back to the 90s, Java was introduced mid-90s. Within five years, it was dominant. It was in the top 10 languages. That's not the case now. Well, why? 30 years ago, if you were around in the business, 30 years ago, you might have thought the world had a lot of code. Compared to what it has now, merely a fraction.

most of the code that exists is the stuff people are getting jobs in. In other words, it is to extend existing systems. It is to maintain existing systems. It is to work in organizations where that is the default culture. And so these languages have a foothold. They have the foothold of the incumbent. It's just that that's where the jobs are. That's where the training is.

That's where the companies are. This is not to say that you're not going to get movement elsewhere, but the movement is being slowed down just by the sheer presence of all the other stuff. So there's a deceleration. Now this is the interesting question. What about AI? Well, in theory, I've just said that you can now do anything you want. Yeah, but do people do that? But it turns out they don't. They tend to follow the path that's already been set. And it turns out that AI does this as well. It actually reinforces it because all the training data is based on stuff that's already out there. So guess what? It's really good at the stuff

that's already out there. And this is an observation, Anders Halesberg, inventor, you know, designer of TypeScript and C-sharp and Teva-Pascal, he made this observation as well. It's just like there's all of this data out there. The amount of Python code out there, the amount of blogs about Python, the amount of training material is just staggering.

So automatically you have a bias in an already biased landscape. It reinforces that. It just puts, you know, touches the scales a bit, puts the thumb on the scales even more. But what is interesting is the more recent trend that I've noticed is the way that LLMs are using languages. So, you know, how many R's are there in strawberry? If you haven't come across this one, this is quite a fun one. A couple of years ago, if you asked ChachyPT, how many hours are in strawberry, it would tell you two.

You know, typically the LLMs, because they don't think about words the way that we think about. They don't have that, they don't have that kind of relationship to the letter, the visualization and all the rest of it. It feels really odd. I can see that there are three R's on the screen. How come you can't? Well, that's not how LLMs work. But what is interesting is that their answers are getting better. And they're not getting better just because people are putting if statements in around.

Oh, we're going to ask the strawberry question. They're getting better because the way that LLMs are being couched and surrounded by other software is like, wait a minute, this is a problem that is already solved. How is it solved? It's solved by programming. And what is interesting is a, I guess about six months ago, I asked the R, I asked the R's in strawberry question to chat GPT and Gemini, and they both gave me the right answer. And I asked each of them, how are you doing this? And they both said, yeah, we wrote Python code. So, you know, show me the code. And they showed me the code that did literally count the number of ours. Here is the word, and here is how you write the code. What is funny is that the latest Gem and I, so I checked this out a couple of weeks ago when I was updating the slides, I don't even have to ask. it. When I ask, when I don't even have to ask it, how did you do this? When I ask it, how many hours are in strawberry? It actually has a tag saying show code. In other words, the whole point

is LLMs are now consumers of programming languages. They look at something and go, yeah, I could kind of like wave my hands at this problem and do it, you know, LLMly, if that's an adverb. You know, I could take advantage of this kind of like pattern matching ability or I could recognize that using my pattern matching ability, this is a really classic programming thing. I shall write some code and that I think is interesting. So LLMs are becoming consumers of our languages and by default it chose Python. So there's a reinforcement again that we are seeing. There's all of these things that are effectively decelerating what's happening in the mainstream of languages, which is not to say that there isn't interesting stuff happening. I just want to say that's out

in the long tail. Whether it breaks through to the mainstream, well, the road has got a lot steeper. That's kind the issue. And that's part of the thesis of the talk. I also have a bit of fun in the middle of the talk talking about where how old some of the features we think of actually are and how they are rolling into still rolling into languages. So maybe you got excited about the fact that your language is adopted or now allows you to do multiple assignment. A comma B equals C comma D. That's a 1963 idea right there. And there were languages in the 70s and it's slowly been coming into languages. Maybe your language has a yield keyword or equivalent to allow you to do kind of routine or iterator like stuff, yeah, that's a mid-1970s idea. The thing is that these

innovations of the 60s and 70s are still working their way into languages now, and there's not a lot of novelty in the mainstream space. And that's kind of reinforcing itself. I think what we're going to see is that in the next decade or two is that languages adopt these features, so Python in 10 years is probably going to have a whole raft more features. C++ is going to, well, C++. Every three years, there's something radically different about C++. It's going to have a whole load of more stuff, but it's still called C++. So the languages are changing in place, but the set of languages is changing ever more slowly. Yeah, what to add and what to ask.

I mean, so TIOBE, I can't have it mentioned without just these piling on a little bit more. Listen, folks, it's hot garbage, you know, whatever their ranking is, like, Fortran and Cobol are like the two fastest growing languages. I think Fortran. Not this week? Not this week? All right, well, at some point there was like three arrows next to Fortran or Cobol or one of them. What are he talking about? Just because a couple banks were hiring a couple people. You're smoother.

And then another thing is interesting is that stack overflow, it's on its last legs. Linguish, one of the sites that I linked to in plrank.com, it hasn't actually updated its latest. They usually do it on a quarterly basis, and it should have been updated in October, I know the guy, Tom Palmer, that does the site Context-free on YouTube, link in the show notes. Stack Overflow can't even run its queries anymore that he uses to generate the data. And so I think he just threw his hands up after a couple attempts. So anyways, you mentioned Stack Overflow and it not doing too well.

And it's probably a website that gives hints that may or may not be out of date when we now have perfect tools. Anyway, so yeah, languish, Stack Overflow, T-I-O-O-B-E. I totally agree. I mean, if you look at the rankings, you basically are to show in the facts. I mean, technically, according to some sites, Go, I think, is ranked 9th or 8th. So I think it's in the 10. It has actually hit the top 10.

But it's funny because I usually every year in January, I make three programming language predictions. And I usually go one for three or two for three. But one of my predictions since I started doing it two or three years ago was that Rust, which I think is ranked like 13th or 12th. It's going to break into the top 10. and every year it doesn't, it just stays where it is. I can't remember if I predicted it last year. Anyway, so you always think, oh, there's a lot of momentum,

but you forget how many decades and millions, if not billions of lines of C++ code there are out there, and that's what you're competing against. And I guess I did see that in the Octover's survey this year by GitHub, that I don't know if it was TypeScript that switched with Python, but they had something in their theory, because they kind of pontificate as to why it's that for LLM's generating code,

they use TypeScript way more often than JavaScript because it gives them another feedback mechanism to see it, you know, did I get the typing of this stuff, right? Anyway, so I guess the question is, if you had to, I guess maybe not predict the future, but how are these, you know, what people are reaching for? Is it going to be, you know, business as usual, and for the next five, ten years,

we're not going to see some big shift, or is there going to be more stuff like the TypeScript taking over JavaScript? Yeah, I think the TypeScript, that GitHub one is interesting. Don't forget the TypeScript is Microsoft. Hey, and so is GitHub. So, you know, it's kind of, there's an interesting thing there. That Anders-Halesburg interview, maybe I'll send Conner the link,

because that's an interesting one, because he's obviously talking about TypeScript there. But there is a case that in some of these cases, every single ranking is going to have some kind of spotlight bias on what it focuses on. The one that I talk about in the talk is I always say, well, when looking at GitHub, one has to be, I don't actually talk about the Microsoft buyers there because that's not actually representative of people that use it. But what I always point out is that some cultures and groups of people are freer at sharing code than others. So you're not going to find COBOL programmers

committing to GitHub. But interesting, you'll find a lot of JavaScripters will quite happily put a single function up there, but C programmers are less likely to put small elements up there. There's no, it's not, C is not the kind of culture that is the, the NPM culture of JavaScript, the level, although there's open... They don't even have a package manager. Exactly. So the point there is there is a bias. So Rust gets very well represented as a language that has grown up in the open source

space and has, you know, kind of end-to-end sort of support for packaging distribution to work. So there's always going to be that kind of nudge. And in particular, new software is going to be much more clear, new code is going to be much more clearly represented than old code locked away behind corporate doors and the closed source space. So we must always be sensitive to these. But the general sense I'm getting is that, yeah, there's movement in rust, there's movement and go, but always remembering that kind of the fact that whatever is at number 10 in whatever

ranking is probably an order of magnitude lower in adoption than whatever is at number one. So rust and go, yeah, having done variations of this talk for over a year, but also look back over some previous years, they are moving slowly up, and they're about to kind of break in on some rankings, not others. Rust has benefited hugely from, basically, you know, Linux has said, yeah, we'll allow some in. It's a case of you've got that acknowledged by that community. Microsoft have been saying for years that many of the issues, security and other bugs,

it's just like, oh, when we do this in Rust, there's a lot better. So you've got somebody like Microsoft, you've got Microsoft and Linux effectively on the same side. which is like bizarre. Then you've got the US government going like, yeah, you want to be careful about certain memory safety language, which has inspired things happening in the C++ space, but more importantly, has again shot on the spotlight on other languages that have a slightly more rigorous approach

to how they think about memory. So these are all nudges. That's going to change something, but there's nothing radical happening overnight. So my prediction for the next five years is those languages are probably going to be continuing there, They're kind of upward trend, but they are just outside the space where there is the killer Apple, the reason why everybody suddenly wants to do it.

And, you know, that's the thing. JavaScript, you know, kids are learning JavaScript because there's this kind of like unholy trinity of JavaScript, HTML and CSS. If you want to get into programming and do stuff at home, then that's one of the ways. You're out on the web with those three. Alternatively, you use Python because they teach it at school, but also it gets used for build systems. Oh, and also it gets used by data scientists. And in other words, it's very well represented in quite distinct communities.

There's not one reason Python is there. There's actually half a dozen to a dozen. And that's going to be a hard one to dislodge, even without the question of incumbent code. It's just represented so well. So you kind of need a language to have either a really strong, specific appeal that suddenly becomes hugely popular. Objective C and the Apple ecosystem, or from Next originally, but that took a language that was dead. I saw Objective C in the early 90s, and it was already on its way out.

That was Stepstone Corporation, and it was only because of Next Step and then, well, no, not even Next Step. It was Apple buying Next Step and saying, hey, we're going to use this for Mac OS and then iOS. That made that. And now it is dropping like a stone because of Swift. It's not dead yet, but it is dropping like a stone

because it relied on one source of support. You need a language that has broad support. And that's the key. I think that's going to be the key to any languages uptake, a killer app, broad support, representation across a lot of disparate spaces. And that's going to drive these things. Everything else is just going to move up and down, a little nudge at a time, various reasons.

Systems programmers kind of say, I've got enough performance. Maybe I'll go for an interpreted language or a VM-based language. It's okay. Stuff is fast enough. maybe I didn't need to worry about the speed I thought I needed a decade ago. So, you know, I think that's one of the things. I think if you dropped into a time machine and went back to 2005 and said,

guess what we used to do a lot of, you know, a lot of our work. Oh, what do you do? Number crunching. Oh, okay. What are you using? You're using C++? Oh, we're using Python. They would just laugh in your face.

We've got to a point where the landscape has changed in terms of performance and library relationships. You know, so I think that I'm not going to make any specifically wrong predictions, but these are all kind of vague suggestions. Obviously, my role is futurologists, you know, I've got to say something's safe. But, yeah, it's going to move in gentle ways. All right, well, I just heard the applause of the Sam Aaron closing keynote. I don't actually know if it's the last talk of day one. So we're going to be flooded by a couple people.

But last two questions that we can do rapid fire. if you would point people at one talk, or if two, if you have to, from the Yale conference, that were your personal favorites that you saw? And then two, it's the end of 2025, so maybe 2026 hasn't happened your scheduling of conferences. But if you're going to be anywhere, you know, let the people know where they can find you. And we've got like 60 seconds before the noise here is going to quadruple in a decibel level. Okay, Sam Aaron's talk, absolutely brilliant. Sam Aaron created Sonic Pie.

There's some really good stuff there. Captain John Will's talk on basically kind of hacking AI is really also worth catching. And likewise, Michael Feathers talk on conceptualization. I would recommend these. And where, do you have a conference schedule for 2026 yet? Oh, good question. Yeah, there's a lot of maybes. I'm hoping to be at Kraft in Budapest.

I'm expecting to be I'm doing Voxed at CERN and then into Chino and I think the rest are just guesswork and hope thanks so much Kevlin

we'll do this again sometime we might take this on to the end of episode 265 we were just chatting with Kevlin I'm okay I've just been offered a vegetarian something something we're talking to Damien

technical director and Damien just stepped into this role six months ago, took over for, I believe, Stefan, am I pronouncing that correctly? So I, Stefan reached out to me, I think, a year ago. Asking if I wanted to speak at the 2024 edition, I was already had commitments, but then I said I'd love to next year, and then you swooped in halfway through. So this is your first, I mean, not your first Yao, but first Yao is technical director. You've been, I think you said you've been coming since basically the Genesis.

15 years. 15 years. So tell us, how has this? this edition, being technical director, gone. What were your favorite talks? What were the highlights? And what can people expect? Because I imagine you're going to be organizing this edition next year.

So, yeah, just tell us all the things and what people can expect in 2026. It's too early to tell for 2026. We need to debrief this one. Look, I've really enjoyed this year. It's been a really nice program, a good mix of speakers. I think the last few years, everyone is very scared of AI. I think what we're seeing with this edition is, we still have jobs. There's still a future. We're still needed. Our insight is still needed. The robots aren't going to replace us just yet. And the fundamentals still matter.

I mean, we saw in a couple different talks, the same or the same slide, not exactly, but they would say that, you know, in this age of, you know, the age of AI and LLMs, it's actually more developers that are needed, not less. And this idea that that junior developers are no longer is the complete opposite. Is that the case? It's like, we're in a world now where, it's the on-ramped to coding has gotten even easier. How is that going to lead to less jobs? So I think that's a fantastic sentiment. I mean, you've seen a number of the talks. I mean, it's hard to, it's like picking favorites with kids and stuff. But if you had to recommend one or two or three, what are the first ones you would point people at once,

because these are going to be up on YouTube. Which ones would you point people at? That's a tough thing to ask me. I can't answer that one. They're all good. Look, I think the opening keynote, I think Kent's talk gives us. a lot of hope for people. And again, it is that we're still needed. The junior thing was

especially good. A lot of people have looked at AI as a replacement for juniors because it's more productive. And Kent's insight that we don't get juniors for their productivity. We're essentially taking an option on their future productivity. And so get the juniors, teach them, use the AI, use the AI to teach them, and then, you know, it's an investment into the future. So I think that was a really nice hopeful message for everybody. Completely agree. And we've also had a, conversation earlier in the week that was about how in this age of remote work, because I work entirely remote, that these conferences honestly have, for me, become even more important because

the socializing that I, it's not socializing is the wrong word, but it's a form of socializing and networking. I miss out on that entirely as I work remote. And you can get basically a concentrated version of that, not just from employees at your company, but across the industry and hear different stories. Anyways, and you were saying basically that we have to do a better job. I mean, you're a conference technical director of putting that message out there that, you know, this has really even become a more valuable thing, which, anyways, I'll let you add to that. Yeah, look, absolutely. In the last few years, we've seen, like, a lot of the content that people

come to conferences for, you can find that online. There's so much of it on YouTube. What you cannot get on YouTube is what we're doing right now, the hallway track, standing around, meeting people, discussing the talks, you know, exchanging of ideas, and we've exchanged a lot of ideas this week. There's been some robust exchanging of ideas. This hallway track cannot be replaced, and yes, it is even more important. Like you, I've worked remote for many, many years. I've long said, yeah, it was my Christmas party.

This is where I come to hang out with my people. So, yeah, I think we need to get that message out there that conferences are more important than ever because you don't get this online. Yeah, I think it was, I hadn't. attended my first conference this year was in September and you don't realize what you have until you're reminded and then when I went I was like oh my God like I've forgotten what it's like to give a talk in person where people immediately come up afterwards and they tell you like you can give an online talk you maybe get one or two questions in the chat and then it's just poof the screen

goes black and you're kind of there's like a little bit of an emptiness inside it's exhausting it's exhausting as well I've done remote training just with to 30 people with their screens off and it's no fun for anybody um so yeah something i will say is most years at yow when i've come i've walked away just super inspired like i want to go on a bill i want to go away and build something and i've felt that again this this time around with yow even though i'm working here and you know i'm running around doing a bunch of things i sit and know what you're talking i'm like i need to build i need to build i'm going to pick up that project that i'd abandon so um yeah i i always get very inspired at Yao and come away wanting to do stuff. So definitely feeling that.

100%. I feel the exact same way. Anyways, thank you for talking. Hopefully this is Audible listeners. They close the doors and the decibel value. So if it's not good, we'll get you, we'll just have to come back next year and do another interview. But thank you so much. Thank you. Thank you for coming. It's been great. It's been awesome. Be sure to check these show notes, either in your podcast app or at ADSP thepodcast.com for links to anything we mentioned in today's episode, as well as a link to a get-up discussion where you can leave thoughts, comments, and questions. Thanks for listening. We hope you enjoyed and have a great day. Low quality, high quantity. That is the tagline of our podcast. It's not the tagline. Our tagline is

chaos with sprinkles of information.