I’m writing a book about epistemology. It’s about The Problem of the Criterion, why it’s important, and what it has to tell us about how we approach knowing the truth.
I’ve also written a lot about AI safety. Some of the more interesting stuff can be found at the site of my currently-dormant AI safety org, PAISRI.
Typescript
325k LoC
Pretty familiar, but not like world leading expert familiar
With rare exception we limit PRs to under 800 LoC added, but it’s common to have a stack of 4-5 PRs each with 400-500 LoC added (harder to stay on deletes, but there’s usually plenty depending on the PR, and we don’t place a limit on them in a PR so I don’t think about it too hard)
Not that thorough. Rely a lot on manual testing + type system
Maybe 60% new features, 20% refactoring, and 20% bugs
I think it’s pretty interesting! In my mind we’re a pretty typical B2B SaaS product. So there’s a wide mix of stuff, but most of the work is shaping code to do what users want, not pushing the edges of computer science.
Yes. This is something I frequently try to emphasize when someone is meditation curious but not already committed to doing it. I say that for most people it’s great, but some people have trouble, and if you’re in the category of people who might have trouble (especially people with high risk of schizophrenia), then you should avoid doing it.
Sure. I mostly wrote this post because I wanted to share my process, not because I wanted to defend the headline claim of a 5x productivity improvement, so I skipped over that. I could have written a version of this post where I didn’t make a specific claim about how much productivity gains I got; maybe I should have rather than stating my belief, since that’s the thing people want to object to the most!
There probably is some argument I could make and it would probably produce a more accurate number than my 5x estimate. I’m just not very interested in making it given that I and approxmiately everyone has high confidence that AI is making programmers more productive.
It’s also likely that claude code work better for you because you’re more experienced and can basically tell claude exactly what to do when it’s stuck.
I strongly suspect this is a lot of what makes my workflow work well for me. My problem is rarely figuring out what broadly needs to be done or how I want it done, and mostly just actually making the changes I want, which is far more tedious if I have to do all the typing.
Where in Europe, if you don’t mind me asking? There’s a lot of variation between countries in Europe so I’m curious to have a more specific data point.
Ah, this is really interesting and helps explain why low fragrence would become important in contra dance but not so important elsewhere!
My theory is that, in the US, The Sort really took hold in the early 1980s when the highest marginal tax rates were radically reduced, thus making income a more efficient means of rewarding employees. The places where The Sort has the least hold are those where income taxes serve the place a soft ceiling on upper middle class incomes and cultural and political forces make either emigration to more lucrative states hard or otherwise limit economic mobility.
I’ve heard a theory that cheap labor is also why Japan is so nice. Not that Japan is a low income country, but rather that for complex structural reasons Japanese workers are underutilized, so everyone in low productivity jobs is overqualified, and it makes everything nice.
Or in short, Japan has isolated itself from The Sort.
Claude Code doesn’t work that well if you’re not an experienced programmer. I mean, it works okay, but it has no taste, so it just produces syntactically correct stuff that is only randomly useful by default. It takes active steering to get good code out of it.
Also, I’m rarely trying to solve algorithmic problems, I’m trying to build production software, which is mostly about managing abstractions and plumbing different systems together to produce something useful to customers. All the hard algorithm work happens somewhere else by someone else because, while it’s essential, it adds very little marginal value for our customers. If algorithms to do something don’t exist, the solution is often to just wait for someone to figure it out, then build a feature on top of them. When I do get to work on algorithms, it’s usually solving complex concurrent execution problems, which Claude is okay at helping with but not great on its own. Luckily, I mostly design smartly to rely on systems that manage these details for me so I don’t need to work them out all the time.
Thanks! I’ve yet to see much value in this approach. I find the time to run the agent to generate the code is pretty short, and what it produces in these unmonitored runs takes more work for me to clean up than just iterating with Claude directly. But, I do expect that the tech will keep improving and that eventually this will be the superior workflow!
Oh sure over time expectations will change and the free lunch will end. Right now I get the benefit of Product just being happy that Engineering finally does something that looks like hitting timelines at all. I’m sure the situation is already different at some other companies. I also have a lot of latitude to work on tech debt because of my high degree of seniority. In some sense my job is to pay down tech debt and otherwise improve the engineering organization to make the company machine better; the feature shipping is incidental.
I don’t use an IDE. My setup is tmux with claude running in one pane, nvim in another, a bash shell in a third, and another bash shell in a fourth where I run our local development environment (it prints logs to console I sometimes need to see).
This is roughly my expectation as well.
I’m not keeping a quantified record, if that’s what you mean, but yes I have a strong sense of how much debt there is in the code base and how it’s been trending.
I’m consistently confused why contra dance people are so concerned about events being “fragrance free”, however the policy is implemented. This is something I encounter no where in my life, I go out and do plenty of things, and only hear about it from people who attend contra events. Is this just a random local norm that reached fixation? How do these chemical sensitive people go shopping or otherwise live their lives in places that aren’t fragrance free? I’m not trying to be rude here, but I’m confused why there’s such a strong norm at such a niche event that I never see even attempted anywhere else.
My 5x number is because I can do roughly as much work as I used to do in a week in a day, if I hold the quality bar steady. In reality, I wouldn’t normally end up being able to do that, and would trade off quality for speed to impact and just live with tech debt to be paid down later. With Claude, I essentially never accumulate tech debt (other than the kind of debt that creeps up because I don’t notice it) and actively pay it down with each PR.
In terms of deliverables, though, it’s as I say more like a 1.5x improvement. The real productivity gains are coming from putting more effort into quality than I’d otherwise get permission to.
(Of course, I could just be wrong because I don’t have great ways to measure counterfactual code quality.)
Hard for me to evaluate this. It’s a lot of speculation that might be right, but doesn’t really supply a model by which I can evaluate it.
Sure, dropping it seems fine. I’m a bit hesitant to address your points too directly because I anticipate the conversation would require something like explaining our entire worldviews and me trying to convince you of mine, so I’ve been doing something more like just explaining my worldview as I see it as relevant to the points here in the hopes that it gets you to see what I’m pointing at, but if that’s not working no reason to continue since I’m not excited right now about making time to explain the whole thing.
I like this. I notice you don’t mention religion in this post, but I think one of the things religions do really well is try to provide access to all three of a scene, a clique, and a team at the same time (though I wouldn’t have known to put it this way before reading your post!).
Why I say this:
A “church” (a term I’m using loosely here to mean both the place in which people practice a religion, the people who practice that religion in that place, and the organizational entity which organizes these two and interfaces with the secular legal system) is at its most basic level a team: some people come together to make religion practice happen at a specific place and time.
It’s also a source of cliques. A church if often too large for all its members to be in a single clique, but it creates an environment with shared experiences upon which cliques can form (and formation is often aided by participation in the team aspect of a church).
A church is a scene in that it’s part of the larger goal of the religion. People often first join a church because they want to be part of the scene, and know nothing about the clique or team aspect. They then get to be part of the collective process of figuring out how to practice the religion in the modern world (which may or may not include things like evangelizing).
I’ve previously made a case that rationalists should be more religious, and being able to talk in more detailed terms about what the community benefits offered by religions are is helpful!