kaashif

My only conclusion from this is that "faster" will be interpreted in different ways by different people. This is actually just a fact and can't be disputed, the evidence is too strong.

It would be even better, in your example, to report "4x ops/sec" or "0.9x ops/sec" rather than faster or slower.

Interesting how people interpret the word faster differently.

To you, 2x faster means 2x the work done in the same time, but 25% faster means 1.33x the work done in the same time.

I see this a lot - in software people interpret 50% faster as a 50% reduction in time taken.

If a car is 50% faster, I hope people would universally interpret that as e.g. 20 to 30 mph. But if I asked people how long would a 1 hour journey take if it were 50% faster, I bet a lot would say 30 mins, because it kind of seems like that's what I mean.

That is correct, I am not trying to have fun while driving to work or driving my kids to school.

You're not sure how the system's code relates to incident debugging for that system?

Isn't the alternative to government being slow government being fast? It's just that it can be fast by disregarding democracy, which is bad.

You actually have a slow government and are complaining about it. You have not experienced any alternative to slow government.

I don't really understand what the post is about.

I don't think it's natural for a software engineer to want to build a "software factory" because I have never heard of anyone doing that. I'm sure it's a thing.

What I see everyone doing is just building software, incorporating their feedback into agents.md, skills, etc, to make their harness better. Is that a factory? Should people not do that?

I still don't really understand.

I doubt the guy is malicious, but it should really be obvious why people are nervous about this stuff. If I put a password into a home automation tool, is my default expectation that it's going to phone home and send any information (even if it's just a bit of information about a hash) about my passwords?

Which option will minimize the feelings of betrayal and violation: making this feature opt-in or opt-out?

Any level of understanding of humans will produce the right answer.

I used to keep a dream journal because I thought it might be interesting to recall all of my dreams.

Eventually I did, but it was hours and hours of effectively random stuff happening that I had to write down or dictate with speech to text.

It was a waste of time and I'm now glad I forget almost all of my dreams. My point is even my own dreams are uninteresting.

I'm not that interested in my own agent setup either - usually the tools become better on their own after a few months and render it all obsolete.

The point is that a car flying for 10 seconds is nowhere near the Jetsons car.

We actually do already have Jetsons style flying cars - those things that look kind of like big drones are basically that, and those are actually practical.

Forget the comparison to the Jetsons if it upsets you, let's just compare to actual flying vehicles.

C++/C#/Python absolutely do have ergonomic ways to do this (and have for decades by this point...) but programmers continue to use sentinel values. It is a bit puzzling.

Yeah, this is the tough part. In order to write code that worked, you had to have some kind of mental model of it. Now that's not true.

Now, when someone sends a working PR in, even high quality and well tested, they may actually have no idea how it works.

Spore had that but I kind of liked it. I guess that was 20 years ago almost, maybe my standards were lower. It felt charming seeing others' animals, some of them looked pretty cool.

AGI would produce novel treatments for diseases at rates equivalent to what a human can do today.

Which is to say, not that fast.

Yeah, workflows like ruff then ruff --fix are ideal, or just outright banning certain libraries, or having policies like dependency cool downs.

It's funny because we're once again speed running what the best practices already are!

There is a step missing. Impolitely list what's wrong and try to incorporate any persistent themes into a markdown file so that the agent stops making the same mistakes. Put any rules into linters and static analysis and sanitizers and ...

I sometimes see people reviewing LLM code leaving the same review comments multiple times, or worse, after some time they just give up reviewing.

But this seems to be in agreement with the comment you're replying to.

Tourist visas are nonimmigrant visas. So are H1B visas.

Was the comment edited?

Pointless images are bad. Like those images at the top of blog posts, I didn't like those pre AI.

The way to use AI is to make sure it has clear, verifiable success criteria, test suites, etc. Make sure any output has citations, reduce the need for trust to zero, etc.

I see people one shot stuff and it makes no sense, is completely fake half the time, just like you point out.

It should be the case that Codex and Claude Code should incorporate this kind of thing automatically at some point.

At some point we'll have enough RAM, surely. The incentives to produce more are huge and all the fabs are booked out.

Maybe it'll take 10 years or 20 years. <5 years is not long enough for manufacturing to catch up.

Not much of a comment on the phone stuff but I'd caution against suggesting technology will never be good enough to do X. Maybe it'll be horrendously wasteful but it might happen.

Which is to say, they'll be banned in the US.

It's a pun on AI winter - if there's not going to be a winter, then it's always summer. But things won't be static.

It is in fact normal for new technologies to be adopted, and for there to be no cycle where the technology just disappears. There was no Internet winter or railroad winter, not in the same way there was an AI winter where AI just went away.

Those technologies had valuation bubbles but the technologies themselves stuck around and changed the world beyond recognition. If someone sees the rate of change and thinks it'll stop for...some reason, then they're the ones who believe things are different this time.

Can you give an example?

Being able to efficiently allocate labor with a more mobile workforce does increase productivity, this is well studied.

Here's a source, although it isn't that detailed and is more of a descriptive literature review: https://www.urban.org/sites/default/files/publication/103581...

This too: https://www.oecd.org/en/publications/2025/07/oecd-employment... - but I didn't have time to read it in detail.

I don't think anyone believes labor mobility reduces productivity.

Why does it have to be a choice? Raise the median, raise the minimum, use wealth created through higher productivity and a healthier labour market to redistribute to some degree, everyone wins?

The frictions we're talking about, like health insurance being tied to an employer, make things worse for families anyway - getting rid of the distortionary regulations that cause that can only be a good thing.

I'm not American, but most people I know who are not in tech just use ChatGPT like they'd use Google but better e.g. what bird is this, what's a good word for X or Y, what's the weather, what was that movie with Z. Edit this meme for me, make this photo not have the guy in the back, how do I grow chillies, etc.

Now if you ask them if AI is making their lives better, they might say they don't know.

I literally don't know anyone who "hates" AI.

Why would you restrict to the US workforce? ChatGPT has a billion users.

Why does being a compressed corpus of knowledge mean it can't discover things? What if it has knowledge of techniques to discover proofs?

It does and that's what's happening.

Damn, looks like we have lower productivity than cavemen!

(1) Yes.

(2) Are you saying that you think we're at the limits of computing in general, or that specific technology?

We know, for example, that a human brain level intelligence is possible to run on a human brain. We are nowhere near that. And actually that's not even a physical limit necessarily.

But that is...not a low hanging fruit.

What if you forget to use that and just use let or const? It is better than plain try-finally or defer because you don't have to remember how to dispose of the resources but in terms of remember to dispose of the resource at all, I don't think that is all that different from Python's with or Java's try-with-resources. You can just forget to use using, with, try-with-resources.

There isn't any way to forget to drop a resource if you have RAII.