[sudonanohome]

A lazy wall of AI slop:

> I was looking for something simple. Something right-sized for my scale.

> Clear boxes don't have this problem. They scale.

> That's not a failure. That's the system working.

I wonder if there's a simple regex that could detect these. Perhaps I should ask Claude

The entirety of this post could be explained in 20 tokens: 1) use transparent boxes and bags for organizing 2) track the usage with stickers 3) remove rarely accessed boxes

We need a sponsorblock-style crowdsourced solution against such slop. Meanwhile I'm just blocking offenders' domains on all of my networks

[gyomu]

Seriously.

It's such a shame too because the author clearly cares about sharing their system, they went to the lengths of taking nice photos etc. - but then it's this low quality, meandering, repetitive, predictable AI writing.

It's also surprising to see that most people in the comments here just do not seem to be affected one bit. Not sure if it's because it's become so standard now that everyone's completely accepted that this is what content is going to be from now on... or if most people just don't care about poor writing.

The second hypothesis seems less likely given how central to the ethos of this board pg's old essays about good writing seem to be, but maybe that's just a bygone era.

[Karuma]

Yeah, I've been just slowly blocking all these domains, users, etc. But nowadays it's just unbearable. We have already lost this war.

And seeing every day this kind of crap at the top of the front page of the websites I used to love, with hundreds of comments of intelligent people not even noticing all this useless AI slop... Very sad future ahead.

[foltik]

> The habit takes seconds. No database. No server. No app.

> It wasn't the specialized components. It wasn't the sensors I had so many of.

> These aren't the exciting parts. They're the infrastructure that every project shares.

[eichin]

I was just disappointed that they didn't include a https://en.wikipedia.org/wiki/Survivorship_bias#/media/File:... joke (or maybe that's also a tell...)

[refulgentis]

It's against the rules as of 2 weeks ago, in the short term, you're more likely to get downvoted for complaining than see less of it.

And it's astounding. Because this is awful writing.

Author, one year ago: "you replied to an LLM generated comment. if you look at the posting history you can confirm it"

Now they can't be bothered to take an edit pass on the most rote slop.

[sixhobbits]

Technically it's against the rules to use AI for comments. I don't think you have to be 100% sure that AI wasn't used for a submission before submitting or upvoting

[wwweston]

> Because this is awful writing.

Can you describe what you think is awful (beyond rhetorical choices that have become LLM markers)?

[refulgentis]

TL;DR: 2400 words ~= 18 minutes reading time for "stickers on boxes" (which I am v interested in!) x entry-level LLM markers = worst of both worlds. You're throwing a lot at the reader & it's nakedly obvious some of its computer-generated spam, and they can't tell what they can skim until its too late. I 100% empathize because I've done this too, generally, needs more time with the LLM and also metrics you're shooting for.

I'm assuming you're the original author: the problem is the LLM markers.

But its bigger than you're thinking, I think, otherwise you wouldn't have done what you did.

No matter the discipline, painting, code, etc., people appreciate things that are unique and require unique effort.

You can see this over and over again in art history, and the products / code we appreciate are things share that property.

LLMs are great tools, you should use them. Problem was when it turned into (I'm guessing, because this is what I do) "I wrote too much...god I hate my writing...LLM, do your magic...LGTM! So much shorter and cleaner!"

The failure mode I'm seeing with people posting obvious slop who otherwise wouldn't want to is, the editing makes things shorter but it also tends to make everything repetitive, both locally and across across the essay.

Locally, to the person operating the LLM, it looks like "its so much more approachable!", to a reader, it reads as "the author things I am very dumb and takes 3 sentences to saying something that was 4 words"

Please don't take it personally, btw. We're between a rock & a hard place right now and it's important to tease things like this out over time. I'd rather have an extended dialogue about it, but...in practice, people absorb microsignals about what's appropriate over time. Better to say "It's awful writing" then write an extended essay empathizing via guessing what was in your head and what you actually did.

I enjoyed the essay and it's an interesting idea and right up my alley. It's just a pitch perfect example of something going bad, "I had problem X, it was solved via boxes and stickers" has to be 600 words, tops. And you probably sensed that when you were done, I'm assuming. Just needed to go farther rather than lean into sense of relief.

[wwweston]

Thanks. Not the author (the person who's participated elsewhere in thread with an account of the same name as the site probably is).

But I appreciate the elaboration. I don't generally use LLMs for writing, but I'm interested in what people perceive as good and the gap between that LLM output among other things. I actually liked this piece, and could outline some things that I thought were effective about it as is, but being curious about what other people dislike seemed more likely to be educational.

I especially appreciate the criticism of repetition in this piece and LLM output in general. That's a great one to think about because repetition has a place in rhetoric, especially with some audiences, but probably less so with other audiences (perhaps especially an engineering minded audience). And for any audience there will be a point of diminishing returns. All things LLMs may be poorly positioned to dial in.

> Better to say "It's awful writing" then write an extended essay empathizing via guessing what was in your head and what you actually did.

But you did! Thank you. And FWIW the extended criticism that you eventually provided including attempting to guess at the author's perspective boosts signal for me. When someone does that, it gives me particulars to learn from and makes it seem less likely they're just cranky or grinding an unknown axe. Though of course no one has time for essay responses all the time (and the pay usually isn't great).

[fgfarben]

Learn how to skim?

[sudonanohome]

Skimming thru shit doesn't make the end result taste any better. All this terrible "writing" deserves is to be fed to another LLM and summarized into 3 sentences. Because that's all there is to the whole post. Why would the author choose to sloppost in a personal blog is beyond me. Personal blogs used to be a place for posting cool stuff you did in your basement. Now it's just another "personal brand" promotion engine. Hacker news is turning into linkedin, and people are turning into algorithmically entertainable slop-fed cattle. The end is near, I hope

[twoodfin]

This is the real apocalypse scenario. Not Skynet, but a model trained so well to trigger dopamine hits across the bulk of the human population that our creative muscles totally atrophy. Then we’re the stochastic parrots.