allknowingfrog

It's a long-standing tradition. Our forefathers famously said "NIMBY" to taxation without representation.

Is this something that the authors of SQLite actually claim? I don't think anyone else can decide what something is meant for.

Isn't Wright sort of famous for designing structures that leak when it rains? My understanding is that Falling Water House is a beautiful money pit.

There's a whole spectrum of employment between faceless corporations and personal side projects. AI will replace humans because giant business believe they can do the same work, not because they will actually be able to.

The correctness of an application is limited by your ability to understand and describe what you need. We have a word for an application specification tool so detailed it eliminates all ambiguity. It's called a "programming language".

The mistakes are always in the transfer from human to machine. I still find a high-level programming language to be the best way to express my intent. Humans will make mistakes in the hand-off to AI just like they make mistakes in the hand-off to code, but at least code is deterministic.

This was a discussion about LLM usage patterns. I'm not opposed to lunch breaks. I'm opposed to being required to take the equivalent of 12 lunch breaks a day while I wait for slow responses.

This thread started with me saying "we clearly have different goals" and then being told that I just need to hold the screwdriver differently...

How do you maintain a flow state during a lunch break? I'm looping with Claude on a scale of minutes. While you're waiting, I'm iterating.

We clearly have different goals. I want an LLM to review my code, not the other way around.

Have you heard of TigerBeetle? Being smart enough doesn't seem to be the primary factor. It's about having a strategy and the discipline to follow it. No type system will ever free you from the burden of doing the actual engineering.

If you're confident that Linux machines aren't going anywhere, the pass utility is a nice option.

I really appreciate the "it's okay to be weird" sentiment. It has never been easier to try out a crazy idea. We may as well embrace it and try to learn something.

Sure, just like AI enthusiasts seem to be unfamiliar with the concept of local maxima...

I've hiked in the mountains and I've swam in the ocean, and I'm perfectly content to live amongst the hills and streams of the Midwest. I suppose it's relative. I live on the east end of the state, and I find the west end pretty flat and boring. :)

If you're reviewing the code yourself, then I don't think this article is about you.

I feel like I'm missing something here. Was this supposed to be a game? I just kept clicking and reading and clicking and reading. I finally gave up. If the goal is to tell a non-interactive story, I'd rather just scroll...

It's awfully literary. It reads like James Joyce attempting to convey advice about effective leadership for technical teams. In my opinion, it's an obnoxious, pretentious approach to writing about practical subjects, but I may be in the minority.

I think "because they are helpful" is OP's point of contention. Tobacco isn't helpful, but it's a product that people spend money on.

SO has always had a pretty strong stance against opinion-based questions, but this is maybe the niche they should be exploring now. Humans still have a lot to say about the "best" solution to a given problem. The whole idea of an "accepted" answer could be removed, for example, since that's what AI will already generate.

Claude is just a tool. My team members are each free to choose the text editor or IDE that they are happiest with. In the near future, I hope to be able to say the same for coding agents. I really like Claude, but I don't track Claude resources in our repos. If something better comes along, I'm betting it will be perfectly happy to parse the markdown of my existing memory files, and nothing in the repo itself will force anyone else to know that I switched.

It kind of blows my mind that the majority of Claude users have just accepted that CLAUDE.md is a tracked file that the whole team has to standardize on and share. Coding agents are the ultimate API. They conform to however you prefer to interact. Is anyone really expecting to enforce standard operating procedures with this non-deterministic black box of magic?

Yeah, this was yesterday in the web version. I only did the one run, and only for a couple of levels, so I don't think I can give you any help reproducing it.

I did a quick run on the web version. I was able to sneak attack everything on the first two levels, which felt like a bug, but I'm honestly not sure.

When I found a spear, I kind of expected to be able to throw it, but I didn't find a throw option anywhere. I think that makes the short sword better in every case, but maybe I missed something.

Overall, I love the execution. Quality retro fun with a really nice interface.

Because the goal isn't "keep this exact version of the app alive and running". The prototype is never the whole application. If your only metric is incidents, then yeah, don't ever touch the code again.

You can get a few feet closer to the moon by building a treehouse, but you still can't turn it into a spaceship.

This problem definitely predates AI coding agents, though it may be exacerbated by them. The article essentially concludes with the ancient advice of "plan to throw one away". Well sure, I also read Mythical Man Month, but how do I convince the decision-makers?

I got excited about TUIs when I was exposed to the Bubble Tea framework for Go. I'm sure that Claude has accelerated the trend, but interesting things were already happening years ago.

Isn't it all just information? The lock code and the location of the safe are both just data. I think it's possible that all security is obscurity.

I think both can be true. Ideally, we should do more to address the social issues that cause people to drink and drive, but revoking licenses is still a good short-term move. We could outlaw AI policing while we work on deeper issues with law enforcement.

Does anyone know how this fits into the larger picture of recent Ruby drama?

I've used foreign keys and unique indexes to enforce validity on even the smallest, most disposable toy applications I've ever written. These benchmarks are really interesting, but the idea that performance is the only consideration is kind of silly.

I think this is too soon to call. No one questions whether AI can build things. We question whether they can build stable things that work as expected and stay online in the long run.

I think "make a product" is the important point of disagreement here. AI can generate code that users are willing to pay for, but for how long? The debate is around the long-term impact of these short-term gains. Code _is_ a means to an end, but well-engineered code is a more reliable means than what AI currently generates. These are ends of a spectrum and we're all on it somewhere.

You ever notice how everyone who drives slower than you is a moron and everyone who drives faster than you is a maniac? Your two camps have a similar bias.