Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Taste Is All That's Left

notashelf.dev

Post reply on HN
Print view View original

201–210 of 575 posts

Re: Taste Is All That's Left

#201

Post by habinero »

aprdm wrote: post #176
deathanatos wrote: post #167

Earlier quoted context omitted.

Short term can, IME, be very short. I've seen people generate, say, a bash script with an LLM. It's generated: short term, the problem is "solved": we've generated a bash script. … but does it work? Someone comes along, reviews it, "this is garbage, and does not do what it says it purports to do". Perhaps it even gave an output: the script computed … something, but it's just GIGO. But that "check if this works" frict…

You can have another agent write the tests and verify the former agent ? This is pretty basic stuff. Makes me question if people are actually trying to use AI

Now you have two problems lol, in that you don't know if the tests are any good or actually test the thing in question.

Sooner or later you run out of turtles to put on the stack.

Re: Taste Is All That's Left

#203

Post by soundworlds »

eudamoniac wrote: post #106
DiscourseFan wrote: post #25

I keep seeing people trying to define judgement and taste and then they don’t actually like cite Kant’s Critique of Judgement where this is basically the entire concern.

Hume also has an essay about this. People don't read philosophy anymore. They think blog posts are just as good, or at least more "relevant", somehow.

I'm not disagreeing with you at all, but I have personally found most philosophy "classics" painfully hard to get into. That doesn't mean they shouldn't be respected, but the truth is that sections of society (like myself) do rely on the people specializing in philosophy parsing it into current language and metaphor.

Re: Taste Is All That's Left

#204

Post by AdieuToLogic »

pornel wrote: post #173
notashelf wrote: post #146

Earlier quoted context omitted.

I'll risk sounding antagonistic and ask you this: if LLMs are not "good enough", why are they still around? Obviously "good enough" was a poor pick of words of words on my behalf---and I'll gladly own it---but they should be "good enough" for something if a significant amount of resources keep getting allocated to them. Yes on a very personal level you look at LLM-generated code and think to yourself "wow this is gar…

They're here, because even "bad" code has uses. We can have quantity over quality. We can generate throw-away programs (where the lack of maintainability doesn't matter). We can solve some problems by brute force (keep trying until tests pass). For people who can't program, even bad code can do more than no code. LLMs are also useful as a natural language interface. They're pretty useful for search (as RAG and autore…

> They're here, because even "bad" code has uses. We can have quantity over quality. We can generate throw-away programs (where the lack of maintainability doesn't matter). We can solve some problems by brute force (keep trying until tests pass). For people who can't program, even bad code can do more than no code.

If this is an acceptable definition of software engineering productivity, where "quantity over quality" is prized, then hire me.

Because I guarantee I can produce dozens of PRs daily, each having tens of thousands of LoC deltas (pick any language you desire), none of which having research and/or understanding underpinning them, and all easily quantified as being "brute force."

Re: Taste Is All That's Left

#205

Post by soundworlds »

Also just a shout out to the author - you're getting a lot of flak here, but your article hit me at the right time in the right way. I trust you're telling the truth about writing it yourself, and thank you.

Re: Taste Is All That's Left

#206

Post by yulker »

ricardobeat wrote: post #131
vor_ wrote: post #107

Earlier quoted context omitted.

They've updated the article to state that no LLM was used.

The update itself has strong AI tells: > I do write like this sometimes. The short sentences, the reversals, the one-word lines—all of it. It’s just the way it is. > That says something about my writing. It says nothing about who wrote it. > So let me be plain about it: Claude was not here. No LLM wrote this—not a sentence of it, nor was it outlined, drafted, reviewed, checked, etc. by one Either the author is doing…

The whole post has that immediate LLM "mouthfeel"

Re: Taste Is All That's Left

#207

Post by j16sdiz »

swat535 wrote: post #111
mdwelsh wrote: post #35

This article resonated tremendously with me, speaking as someone who has been coding since the 1980s and had to develop my own taste the hard way, through lots and lots of mistakes. Whenever I see a software demo built by someone mostly using agents to do the work, I have to question whether it's any good on the inside - or whether any real intuition or judgment went into its design. On the flip side, maybe none of t…

For majority of people, including business executives it's the latter. No one cares about artisan code. Software is just a means to an end for them and not an art form. People treat their computers like any other appliance. They don't care if it's riddled with bugs, they'll gladly restart their computer or refresh the page at infinitum to get it to work. I think it's also why marketing and designers get more of a say…

> .. why marketing and designers get more of a say in the software than engineers.

I hate it when visual designer do a product design or ux design job.

I hate it when marketing just imagine some fake use case in their head.

Market Research is real research. UX study is a real study.

now everybody just vibe their job

Re: Taste Is All That's Left

#208

Post by dimgl »

socketcluster wrote: post #86

In software development, good taste translates to readability, succinctness and maintainability of the code. It also translates to intuitiveness, reliability, security, performance, scalability and utility of the software. All this doesn't necessarily translate to sales though. Because, to evaluate those things requires attention, trust, time and effort. This is a significant barrier because a lot of software will ap…

> I think AI slop code is going to be a much bigger problem than people anticipate.

I'm seeing the same thing. Of course, it's always possible that AI capabilities grow significantly from here and this worry is moot (and there will be other things to worry about, like whether I'll have a job). But seriously, after a while AI generated codebases really fall apart.

It's a bit of cope from me that this problem won't be solved in the future, but I'm also really disappointed with how many people really do not care about writing high-quality software that is easy to understand, performant, and resilient to bugs. Like, why did you decide to work on software in the first place?

I think we're going to see some pretty careless mistakes in the future that will result in a lot of pain...

Re: Taste Is All That's Left

#209

Post by esikich »

boron1006 wrote: post #128

I don’t like the term taste, but the problem that I have is that LLMs don’t seem to work “good enough”. They seem to be able to solve the immediate problem, but stacking this on the scale of 3-4 devs over 6 months or so doesn’t seem to produce anything. One thing that I’m particularly frustrated with is the writing quality of LLMs. Like this is the thing that they should be able to do, but I would say almost everythi…

This is just a problem of articulating what you want. Most people can't write for shit, I'm sorry. Because they maybe took one writing class in college. It's a skill in it's own. LLMs write better than anyone I've ever worked with and the reasons are obvious. If it rambles or is too "purple", that's on the person prompting it, because they can't write for shit to begin with.

Re: Taste Is All That's Left

#210

Post by red75prime »

KerrAvon wrote: post #140
asdfman123 wrote: post #122

Earlier quoted context omitted.

Right now we're all training AI models on our taste. All of our chat logs are being uploaded, and the AI companies are training the models on them to get them right the first time. It will soon no longer require taste. Or at the very least be 95% of the way there on the first run.

This was the basic theory behind Tesla eventually being able to stop killing Tesla drivers with FSD. It has not worked out.

I think it's obvious that repeatedly noting that the absolute perfection isn't achieved yet is an absolutely useless way of tracking progress. No?

BTW, when was the last time Tesla driver died due to FSD fault?

Post reply on HN