rjh29

You can still enjoy Reddit and it's still a useful source of information. Just avoid the popular subs (including those you mention) or anything with 100k users or more. It's not difficult.

It's just how he is. Prior to LLMs he was cramming a datasette link into every thread. Downvote and move on.

I think it's 2. I frequently get told there's no capacity for Pro and the query is answered by Flash with extended thinking. And tbh it's hard to tell the difference between the two, especially if you're not coding with it.

It's easier to split USB 3 into two stages:

1) USB 2.0 - 480 Mbps, 0.5A (2.5W / 5V)

2) USB 3.0 on Type-A - increases data to 5/10 Gbps (e.g. for SSDs) and power to 0.9A (4.5W / 5V).

3) USB 3.0 on Type-C - adds pins to enable most of the new stuff: base 3A (15W / 5V) and up to 240W with power delivery; extra data lanes for 20 Gbps, DisplayPort for external monitors.

4) USB 4.0 on Type-C - changes protocol completely (using packets instead of fixed lanes per pin) and requires minimum speeds and the use of type-C port.

A-to-C cables act like A-to-A as they don't have extra pins - they're just easier to use.

Pre-Power Delivery lots of phone chargers used A-to-C cables with a custom protocol to negotiate power draw beyond 0.9A. And some devices pull 3A anyway just assuming there's a power brick that supports 3A attached. But technically none of that is compliant.

Choice of tone can communicate the same thing as 3-4 paragraphs of calmer discourse. Should we ban the use of tone? Personally I appreciated it and the frustration was palpable.

I agree there is some technique involved. It's just so trivial compared to hand coding that I find the comparison meaningless.

"practice techniques for AI-assisted development"

I guess people don't like to admit that AI is doing 95% of their job so they pretend they are "practicing techniques" or "experimenting with new models", or "heavily planning the spec"

Writing what you want in English and having the AI spit it out does not take any skill anymore, it's not 2022.

Pretty good summary. It can be confusing since type C is "just a connector" but it's pretty heavily coupled with the protocol now.

Usb isn't the issue, you bought an extremely budget mechanical keyboard.

It's complicated. They are underpaid relative to inflation, but there's also a massive queue of graduates who want to replace them. If the govt aquiesces to their demands then they may need to do the same for other unions.

"In the UK, most people simply get Bupa dental insurance."

Respectfully that is completely wrong, do you even live here? Nobody wants to pay for private healthcare here (takeup is ~10%) and people don't care about teeth as much as they probably should.

Finding an NHS dentist is nearly impossible in most of the UK, though.

Perl is another example. It has a dedicated linter to detect the myriad ways of using it wrong.

The author was talking about eliminating memory being copied between processes. That's a performance improvement. I don't know how the 300x improvement was made or whether that's even real - you should really be arguing with the author about that. I didn't seriously put stock into it because you know, they used AI to rewrite it. I doubt their testing methodology is sound.

I appreciate the condescension though. Thanks for teaching stupid old me how a database works.

Ladybird's Rust port of the JS engine was a good example. Compare the output byte-for-byte, run both in production (with the new code disabled but checked) before releasing. It was LLM-translated but done carefully.

I'm not interested in a technical conversation with you; I was answering someone else. The top voted comment thread of this page, which dominates discussion, already covers everything you'd want to know.

There are hundreds of comment chains about this already, go troll somewhere else.

It is the opposite for me. AI loves terms like "you've just hit upon ..." or "it's the holy grail" or "The X Trap" so much that I internally cringe when I see them, and would avoid using them.

Yes, see top comment.

In this case it's justified because Rust allows safe implementation of threaded code. Current Postgres is per-process. Switching to threading yields performance improvements.

Windows versions running on Linux over Proton tend to be more stable than any native Linux version would be.

AI is a great use for this kind of boring, rote translation where precision is important. Humans are quite bad at it and tend to make mistakes. In either case the focus should be on improving testing, not trying to manually verify if the translation was correct by eye.

Surely rented GPUs would be used for inference, which runs anywhere and needs to scale much larger than a training run.

It's also sensible for them to provide a cheap, intelligent model to users if they have capacity, then once they built a user base, tighten the screws. All the other AI providers have done that.

Stop publishing them! Vibe coded projects are totally fine for your own consumption. But if you're not going to do the real hard work - build a community, fix bugs and add flexibility for all types of users - just keep it to yourself. Your slop project has no value to anybody.

They already invested in the massive datacentres of GPUs sitting idle. They have fewer users so they can deliver more inference per user - more thinking, larger models.

It's not easy. I learned Japanese at 28 and am now proficient. But I didn't want to spend my whole life asking other people to translate things for me, being permanently ostracised from conversations, or causing problems for others by forcing them to speak English. It's basic decency imo.

Some of it sure. But "30mins" has no space, "Where it was really weak was UI design.", "reasoning were top-notch", etc. prosidy in general has Eastern European / ESL feel. LLM-assisted yes, 100% LLM-generated? No.

I think we should be more forgiving of AI use by ESL speakers. They generally don't know how cheesy it sounds.

They're definitely original human thoughts. It's just written like a sales pitch.

Author is CEO for an AI company it seems, so perhaps they live in a LinkedIn bubble where everybody talks about AI like this.

Isn't it already ridiculous that we can get to the other side of the world in half a day? It used to be straight up impossible, and then it took months by boat. We're now at diminishing returns, saving a couple of hours.

How long are we going to keep running up this hamster wheel instead of being satisfied with what we have?