AstroBen

"I saw her duck" - how much would you be willing to bet on the intended meaning of this sentence?

English isn't a good programming language if you want to trust that the output will be as you want it to be. High level code has no room for ambiguity - that's why we use it. It's a concise, very explicit language to tell the compiler exactly what to do.

Dealing with an LLM is a fundamentally different thing. The reason it's able to speed us up is because it makes assumptions for us and reads between the lines. The only world where we trust the output is one where it can literally read our minds. That's not coming any time soon

This is just pure vibe coding and trying to convince yourself it's quality work. AI can't be trusted to go from an English written spec to tests

Here[0] we have benchmarks for AI writing tests for code that already has the correct behavior. Best case scenario is that they're at 67% with Fable, and that's with the highest quality specification we could possibly have (code). English is far more ambiguous

[0]: https://labs.scale.com/leaderboard/sweatlas-tw

I thought about doing that with Todoist.. and then just decided I'd rather pay the $4-5/mo or whatever to not have to use unpolished apps and sync server

And engineers have a lot of leeway to just make shit up and the PM would have no idea until it blows up.

So you just click around the UI, assuming there is one?

Can you break down for me exactly how the unit and integration tests are made? Do you ask the AI what to assert?

How do you test code that you don't understand?

Black-box e2e testing isn't suitable, or practical, for a lot of applications. Unit and integration tests generally require you to understand the code you want tests for, to know what to actually assert

Are you just trusting the AI to do it all?

Makes a lot of sense to me. If you're young and things aren't going so well then at least you can have hope that you'll be able to turn it around. You have time on your side. That all goes away when you start to get to middle age. You're locked in by your earlier choices.

Now I'm here staring at AI with the reality that if it makes my skills irrelevant I'll have no fallback, and I'll be at an age where reskilling is extremely hard and I'm still nowhere close to retiring.

Yeah, fair point. They're certainly trying to push it in that direction but so far there are still alternatives. I've seen age verification get a hell of a lot of pushback so that's encouraging.

Yeah, but you're not beholden to them. There are 100 different hosts you can use if you own your own domain. If a host changes in a way you don't like, just move your domain elsewhere. If you're using Gmail, you're stuck with Google. Being independent of any one host is the important part to me.

Personally I have my own mail server and use smtp2go for sending which handles the deliverability issue. I'm not sure it's worth it going this way but I found it fun and its been 0 maintenance

Why is it an issue? I don't care what other people do with their email.

If it's not deterministic you can never fully trust it. In a deterministic abstraction I don't need to audit the lower levels.

If you don't feel that's worth it you can use Gmail, yeah.

It's not just work meetings. This is being taken to healthcare settings, also.

Register your own domain and use that for your email, and you'll no longer be held hostage by Google. Takes almost no effort and will cost you a few dollars a month.

They're different goals though. Someone selling software is doing it to put a roof over their heads and money in their retirement accounts. If you straight up replace that with working on passion projects and giving them away you get.. homeless people.

I'm not sure the starving artist is an ideal to strive for, either. Surely there's a middle ground?

Obsidian's model seems fair: https://obsidian.md/pricing

JetBrains' lifetime "subscription" which gets locked at the version you paid for seems fair to me

I don't see why you can't work on something you're passionate about and make money from it. For those of us not retired, the money is essential to make it sustainable.

I wonder if you'd also be arguing for libre software to reject available funding? What's the difference?

The Ruby on Rails foundation brings in a million per year: https://rubyonrails.org/foundation

The Linux foundation is funded at around $250m / year from a quick Google search

> That's not "AI tries to design something". That's an AI that has been trained, by the prompt stack, to behave like a senior designer with a working filesystem, a deterministic palette library, and a checklist culture

What, you don't want your senior designer to have a working filesystem and checklist culture? No deterministic palettes?

The only reasonable thing to do if you care about the longevity of your workflow is to build it around open-weight models.

If you choose to not be able to get work done without Claude you're at the mercy of whatever they want.

Not really. A person will eventually drink dirty water if it was the only thing available in a desert.

There's very little competition for SOTA models. The models themselves also weren't built by Claude. The current revenue has almost nothing to do with what Claude built.

Hell if it was so far ahead then they wouldn't be desperately trying to block OpenCode.

Even if this is right, by responding to the public issue here he's taking on some level of customer support. A simple "I forwarded the refund request to the relevant team and you should hear back from them" would be a million times better than ignoring it and closing the issue.

> bad customer service seems firmly in the "better to stay and advocate doing better from the inside" category

How about Anthropic agreeing to a $1.5 billion settlement for perhaps the biggest theft in history?

Weird how people forgot about that.

A real employee (bcherny) read the issue, responded that the bug was fixed, and then completely ignored the request for a refund.

So much of Apple's growth is anti-consumer.

You can't repair your device.

They're intently focused on locking you in as much as possible, making it hard to leave, and not by making such a good product.

They try their best to force app developers to pay them their 30% tax, even when the devs brought the customers in from elsewhere.

They, for so long, refused to support RCS and downgraded the messaging experience with android.

They were trying to intentionally downgrade SPAs so people again need to go through their app store.

Anything I missed?

They make good hardware, yes, but I can't support them as a company.

> What’s not talked about is the long-term consequences of falling into these groups where excessive drug self-experimentation is normalized.

Lots of people from the 2010-ish era of "aesthetics" and steroids are having heart issues now in their 30s (or earlier). Pretty sad to see.

To me it's fairly clear where this comes from: ambitious people convinced they've figured out some secret cheat code that no-one else has. I'm yet to see that path end well for anyone.

> You should recognize that they are a bad person

Maybe I'm giving them too much credit but I don't really think they're bad people. Young, arrogant, stupid, unaware of the consequences of what they're doing sure... but I don't think it comes from a malicious place where they're intentionally trying to hurt others.

I only see this being the case for throwaway code and prototypes. For production code you want to keep long term it's not so clear cut.

> skimming through an alien looking codebase, scratching your head trying to figure what crazy abstraction the last person who touched this code had in mind. Oh shit it was me? That made so much more sense back then

This is exactly how you learn to create better abstractions and write clear code that future you will understand.

I wish more was being invested in AI autocomplete workflows. That was a nice middle-ground.

But yeah my hunch is "the old way" - although not sure we can even call it that - is likely still on par with an "agentic" workflow if you view it through a wider lens. You retain much better knowledge of the codebase. You improve your understanding over coding concepts (active recall is far stronger than passive recognition).

I'd be very surprised if they're not already doing this.

Design is very hard to verbally describe, and AI doesn't have good judgement on what is easy to use or attractive.

Yeah I'm not a huge fan of it, either. Well organized CSS is much nicer to work with. On the other hand, I'd prefer Tailwind to badly organized CSS.