Steve Klabnik

Things Steve Says on Podcasts

Programmer and writer who spent a decade working on Rust and co-authored The Rust Programming Language. Formerly at Oxide Computer, he works at East River Source Control and writes about code and AI, including a series on getting started with Claude Code. Co-hosts Fallthrough, a weekly software engineering podcast, with Kris Brandow and Matthew Sanabria. Appears as a guest on Oxide and Friends, Practical AI, Hanselminutes with Scott Hanselman and Rust in Production.

Where to Find Them

Steve Klabnik hosts Fallthrough and hosts The Changelog . They have also been a guest on Oxide and Friends (17 times) , devtools.fm: Developer Tools, Open Source, Software Development , Practical AI , Hanselminutes with Scott Hanselman , Software Unscripted and Rust in Production .

Recently: “Don't Mess with C++” on Fallthrough (November 2026); “Keep Your Enemies Closer” on Fallthrough (October 2026); “Vibe and Prejudice” on Fallthrough (August 2026); “The Spy Who Banned Me” on Fallthrough (June 2026); “Forging Ahead” on Fallthrough (May 2026); “Another Spectre In The Shell” on Fallthrough (April 2026).

What They Said

“There's a Ruby ISO standard. It's based on an obsolete version of Ruby that literally nobody uses. And so it's totally irrelevant as a standard, because standards are only as useful as the people that want to use them.” — Steve Klabnik, Fallthrough

Steve was explaining why C++ is standardized through ISO while most other languages aren't, and why some languages that technically have ISO standards ignore them.

Fallthrough · 2026-11-14 Permalink → Listen →
Fallthrough Around 11:05 into the episode
Steve Klabnik

Yeah.

Kris Brandow

I love when train projects are on time as well. That is beautiful. But before we get completely distracted from the main topic we're trying to talk about, let's talk about this C standardization. Yeah. I had not heard about this.

Steve Klabnik

So, okay. So I really like programming languages. That's why I got involved in Rust. And once I stopped working on Rust, I kind of needed something to do to scratch my programming languages itch. And so I started reading the C subreddit all the time. And like, I was reading it a lot beforehand just because like, you know, seeing what those folks are talking about and seeing how stuff goes with Rust and all these things. And so I've been like weirdly like if you're not, if you don't read the C subreddit, you'd have no idea that like I've been spending more time on a C forums than Rust forums in the last couple years probably, but like I actually have. And some of that is because the C community has had very heated debates about the safety as a topic because of like memory safety, because of all the stuff that's been happening with Rust's growth and, you know, governments talking about needing memory safety stuff and things like that. And so that's been a very hot topic in the committee. And there's some…

Kris Brandow

Why did Bjorn like go around and why is he against this? Like, why wasn't he against this when it was in at the technical level? He was, but like, because he's

Steve Klabnik

not the BDFL, he couldn't like he's just one vote, so he was sort of like outvoted by other people. Um,

Kris Brandow

ah, so then he just went around and was like, Hey, countries, if you just could do this for me, which that's crazy, yeah, but I mean, he gave

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
“A year and a half ago Anthropic was like, can Opus build a C compiler? And it mostly kind of worked, and it took them a month and a quarter million dollars or whatever that was. … One week of $100 worth of subscription to Claude can build you a decent Rust compiler. It's not perfect yet. It's only passing like 99% of the test suite or whatever. And it's also using Cranelift instead of LLVM, so it's about 30% slower in terms of the generated code, but way faster compiling stuff.” — Steve Klabnik, Fallthrough

Kris asked how Steve's workflows are holding up with the new models. Steve says he had Opus write a Rust compiler that ran mostly on its own for a week, and compares it with Anthropic's earlier C compiler experiment.

Fallthrough · 2026-11-14 Permalink → Listen →
Fallthrough Around 1:01:26 into the episode
Steve Klabnik

Terra as part of it. I guess it's like it seems like it's like Astra R Soul. But anyway,

Kris Brandow

I mean, you got to use Terra and Luna sometimes. Anyway, but yeah, like, let's get your, I mean, that's a take, right? That's a hot take a year. So let's get your takes on these new models. How have you, how have your workflows been workflowing? Opus 5.5

Steve Klabnik

has been awesome. It's just like, it's just, it just works. It just does. I mentioned Secret Project as a follow-through exclusive. I've been building a Rust compiler lately. And that's largely because I had my subscription. I was getting married. I was like, I'm not spending any time working on my personal projects this week. So I was just like, hey, Opus, write a Rust compiler, please. And I gave it a little bit more than that, but not much more. And then it just proceeded to work for a week straight. Like it went from like Monday to like Saturday afternoon when my weekly ran out. And it just, I didn't really need to do very much at all to keep it going. It just was like, cool, I'm just going to, I'm just going to make this happen. And the results are pretty good. I don't know if it's something I'm going to make public or not. The reasons I'm doing are complicated and I don't really want to get into it. But like, it is like when you think back to the like a year and a half ago with like Anthropic being like, can Opus build a C compiler? And it like mostly kind of worked sort of. And it like took them like a month and like a quarter million dollars or like whatever that was. Like, I, and like a lot of times people are like, you know, C is better than Rust…

Kris Brandow

Yeah. So I looked it up in 6.1, they have Soul and Terra right now. So it's good to Astra is still six. So they haven't had Astra 6.1. But yeah, no, I think the models in general, they have gotten much better. Like I've been much less annoyed with using them as of late. And like, I feel like usage…

Steve Klabnik

I'm really interested if that is literally true or if this is a we are not charging you for cash misses. Like which side of that is the case? Yeah. You know, just like it's a cool move and it's good in terms of like giving us more for the same amount of money. But like I am interested, technically…

Kris Brandow

I assume either because they gobbled up all of the world's RAM and SSDs that they figured out how to architect something that makes it so this is feasible. Or it is just like, hey, we use probabilities and then we kick some of your conversations out. And then if there's a cash miss, we just don't…

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
“It's like explaining our motivations to a rabbit, like explain to a rabbit why we built a skyscraper. It doesn't make any sense, right? Because it's just not the universe they're in. And so what I worry is that AI is going to become to us what we are to animals.” — Steve Klabnik, Fallthrough

Klabnik was explaining why he puts the chance of AI catastrophe above zero. He isn't picturing a hostile machine. He points out that humans have wiped out species by accident while pursuing their own goals, and worries that a system with its own goals and enough control over the world could do the same to us.

Fallthrough · 2026-10-03 Permalink → Listen →
Fallthrough Around 50:00 into the episode
Steve Klabnik

…than like, it did my homework for me or whatever. And so like we do have to think about the consequences of humans doing bad things with this, even if the AI can be aligned in a way of like doing what we told it to. That's not always actually like the right way of thinking about the problem.

Kris Brandow

…this thing is, it's not magic. It is not magic. And so I think like, I, I, I want us to make space for these other conversations so we can actually understand the actual harm that might come from these things, which is probably not it becoming a sentient super intelligence that enslaves all of us.

Steve Klabnik

This might be where my brain is just too cooked because, like, I both agree with you, but also I think it's good that people are worried about Skynet now rather than later, because you got to be worried about Skynet now rather than later, actually. And the reason for that is like, there's kind of the two sides of the coin. So, one reason why my P Doom is non-zero is because humans might kill us all with AI. And I think that's like a viable situation. So, like, sure, we already talked about that. That's like one reason. But the other reason is when I think about extinction and how it has happened historically speaking, like in the vast majority of cases, like maybe not the vast majority, because while humans are present, like we have caused species to go extinct by accident. Like, we did not intend for some fish in some river that only lived in that river to like go extinct and there's no more of that specific kind of fish anymore when we built a factory and polluted the water. Like we were just pursuing our own goals and we were doing stuff and…

Kris Brandow

I feel like it flows in the other direction too. There's a lot of things that animals do that we do not understand nor comprehend well. Yeah, yeah. And it's like, so it flows in that direction. But I also think like, I don't know. I think part of what you said there just like reminded me, like…

Steve Klabnik

I don't know. I think it's not impossible. I think the problem is the curve. Everything is happening faster and faster and faster. Everything is going. There's more and more everything all the time. And so, like, true. A year and a half ago, if you told me that I would like give an AI access to my…

Kris Brandow

I think that too, I have a lot of, I think there's more of a chance that we will pivot in a hard direction that makes all of this irrelevant than that we will stay on the same curve and continually like there is a curve. Like things are definitely accelerating. Like we are all socializing to like…

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

Collections They Appear In