This is cool, and clearly a popular post, yet also... doesn't the 'Show HN' guidelines specifically exclude lists and reading material?
With the massive drop in quality that AI has brought to Show HNs, this is a great example of something that truly is cool... but would be equally cool without 'Show HN' in front of the title. We're failing to moderate the site well unless we recommend OP does what the guidelines say and "make a regular submission instead."
I'm not sure how this would "ruled on", but my take is that the website itself has enough functionality, enough ways to view and slice and dice the info, so to speak, that it can be tried out, and is not merely a "list".
HN rules are slightly bendy and not blindly applied. Things are generally okay if it's within the realm of good taste and judgement.
For instance, I've seen plenty of submissions which were retitled, like a news story which had a vague headline written for "normal people" which ends up being an article about a specific tech issue relevant to this community.
Looks a lot alike Web UIs SOTA models produce, and you know what I'm really glad.
No more people doing crazy stuff with UX just because.
No more frontend dark magic only the most dedicated zealots can make sense of.
Democratised UI for all. And this might not even have had an AI assisted development, I'm just abusing this comment and going slightly off topic to say I'm happy for this tech, as someone that only dealt with backend and DevOps/SRE to be able to confidently create useful graphical applications for regular users and learning a ton about frontend architecture in the meantime.
Btw, a hall of fame for comments would also be interesting!
Great work! I'm doing something similar, but using upvoted submissions I did so I can rank posts better, filter out posts about AI but most importantly flag potentially interesting submissions in New that couldn't have made frontpage.
I find you're perspective quite baffling. It is interesting in that you like the standardization and lack of certain web design crimes and I should not be overly negative to the OP showing their work^1. But man, I truly cannot understand how anyone can see that slop design and feel anything positive. It makes me want to instinctively close the tab immediately as soon as I see it.
[1] It's mostly not their work and just generated though that's not as relevant. But still I think we should be polite when someone shows something that has the chance of having some significant amount of human effort in it.
Separating emotion from logic without one dominating the other is a skill. That's Simone Weil's attention. You need to look at the world with curiosity and openness in order to filter out and assimilate only what you wish. See past what is for what in can be. I'm not saying the project is good or if it's aesthetic, but the functionality works and without LLMs this would never have seen the light of day. All these problems you say are easily solvable with something like Stitch, Impeccable and a tad of good taste from the part of the author. But for a MVP or a PoC? That's really more than enough.
And yet the search on mobile opens half cut, the dark mode icon isn't centered like the search icon and in the resurfaced section are overlapping texts. And thats just from glancing through the page on my phone.
Maybe I'm too picky but it gives a sloppy, unpersonal feeling.
The fact that things like that get into a production app at all is the bigger point. AI slop doesn't even hit a base level of "it works on my machine". When some piece of an app just doesn't work, regardless of whether it was LLM-generated or not, it means there was no QA done on the app. Or at least, not enough.
That is why the term "AI slop" exists. Not because AI cannot produce working code - we all know it can - but because it breeds a low-effort process where people just slap code into production with no regard to quality.
I'm surprised how many of those I've never seen before, even after spending more than a decade on HN and arguably way too much every day. Such a good surprise :)
I wish the page used the CSS :visited selector and provided a link on the index page, so we could easily see which one we've visited before or not.
> For example, Chrome only holds your history locally for 90 days.
What, that would be silly no? I daily drive Firefox, just Chrome for testing stuff, and Firefox remembers websites I visited years ago, I think I'd almost consider the browser useless if it forgets addresses after just 90 days.
This concept has annoyed me so much that I decided to give it a name and create a bunch of shortcuts in my AI tools to stop it.
I call these things "sprinkles" Because I feel like AI can't stop sprinkling dumb shit all over its designs. Small text that annotates things, little callouts are examples but to me, the canonical thing that AI does when you ask it to build a UI that makes no sense is to create a bunch of stuff on the page that couldn't possibly come from any kind of deterministic database.
For example, if you ask it to sort and filter a library UI for a book finder, it'll add some element on a SERP page that's like "22 first time novelists" Even though that piece of information doesn't exist anywhere and couldn't possibly wire into a dynamic website.
Thank you for sharing this extensive work. I’ve bookmarked it; I’ll likely analyze each category in detail—covering the topics, ideas, and statistics—whenever I find the time.
This was also my first instinct. Wikipedia says the US Baseball Hall of Fame has 351 inductees currently, for a sport that has existed in professional form since 1876. That feels like a more appropriate rate for this kind of enshrinement. Over 3000 in the span of less than 20 years is far too fast. The entire English language literary canon doesn't have that many entries.
on the other hand, there are much more links than baseball players. also there are only around 1k hall of fame links, the rest are candidates (i.e. have not met all requirements - 5+ submissions · 3+ distinct submitters · 3+ active years · 2 qualifying breakout events from different submitters at least 720 days apart)
i do accept thought that i have exaggerated in the title)
I was initially skeptical, but after looking through it I see some articles that I distinctly remember (eg Gervais Principle) and blogs like SSC which led me down rabbitholes.
Lots of good things come from the orange site. I’m sure myself and others will dig through these to find those hidden gems that we missed :)
This is cool, and clearly a popular post, yet also... doesn't the 'Show HN' guidelines specifically exclude lists and reading material?
With the massive drop in quality that AI has brought to Show HNs, this is a great example of something that truly is cool... but would be equally cool without 'Show HN' in front of the title. We're failing to moderate the site well unless we recommend OP does what the guidelines say and "make a regular submission instead."
https://news.ycombinator.com/showhn.html
I'm not sure how this would "ruled on", but my take is that the website itself has enough functionality, enough ways to view and slice and dice the info, so to speak, that it can be tried out, and is not merely a "list".
HN rules are slightly bendy and not blindly applied. Things are generally okay if it's within the realm of good taste and judgement.
For instance, I've seen plenty of submissions which were retitled, like a news story which had a vague headline written for "normal people" which ends up being an article about a specific tech issue relevant to this community.
Looks a lot alike Web UIs SOTA models produce, and you know what I'm really glad.
No more people doing crazy stuff with UX just because.
No more frontend dark magic only the most dedicated zealots can make sense of.
Democratised UI for all. And this might not even have had an AI assisted development, I'm just abusing this comment and going slightly off topic to say I'm happy for this tech, as someone that only dealt with backend and DevOps/SRE to be able to confidently create useful graphical applications for regular users and learning a ton about frontend architecture in the meantime.
Btw, a hall of fame for comments would also be interesting!
This was 95% made with gpt 5.6 sol high
I actually did the comments internally and will probably post later in the week.
Great work! I'm doing something similar, but using upvoted submissions I did so I can rank posts better, filter out posts about AI but most importantly flag potentially interesting submissions in New that couldn't have made frontpage.
You might have forgot to tell it 'no mistakes' in the prompt?
The 4 boxes at the top are somehow double loading items on refresh (didn't look at the code, just the output) and generating an awful flicker.
I find you're perspective quite baffling. It is interesting in that you like the standardization and lack of certain web design crimes and I should not be overly negative to the OP showing their work^1. But man, I truly cannot understand how anyone can see that slop design and feel anything positive. It makes me want to instinctively close the tab immediately as soon as I see it.
[1] It's mostly not their work and just generated though that's not as relevant. But still I think we should be polite when someone shows something that has the chance of having some significant amount of human effort in it.
Separating emotion from logic without one dominating the other is a skill. That's Simone Weil's attention. You need to look at the world with curiosity and openness in order to filter out and assimilate only what you wish. See past what is for what in can be. I'm not saying the project is good or if it's aesthetic, but the functionality works and without LLMs this would never have seen the light of day. All these problems you say are easily solvable with something like Stitch, Impeccable and a tad of good taste from the part of the author. But for a MVP or a PoC? That's really more than enough.
And yet the search on mobile opens half cut, the dark mode icon isn't centered like the search icon and in the resurfaced section are overlapping texts. And thats just from glancing through the page on my phone.
Maybe I'm too picky but it gives a sloppy, unpersonal feeling.
I could take your comment and use it as a prompt and depending on the model quality it will have been an ex-bug
The fact that things like that get into a production app at all is the bigger point. AI slop doesn't even hit a base level of "it works on my machine". When some piece of an app just doesn't work, regardless of whether it was LLM-generated or not, it means there was no QA done on the app. Or at least, not enough.
That is why the term "AI slop" exists. Not because AI cannot produce working code - we all know it can - but because it breeds a low-effort process where people just slap code into production with no regard to quality.
I'm surprised how many of those I've never seen before, even after spending more than a decade on HN and arguably way too much every day. Such a good surprise :)
I wish the page used the CSS :visited selector and provided a link on the index page, so we could easily see which one we've visited before or not.
18 year HN vet here. I've forgotten most of these, even though I was around for nearly every single one of them!
I wish the page used the CSS :visited selector and provided a link on the index page, so we could easily see which one we've visited before or not.
That only lasts as long as your browser history. For example, Chrome only holds your history locally for 90 days.
> For example, Chrome only holds your history locally for 90 days.
What, that would be silly no? I daily drive Firefox, just Chrome for testing stuff, and Firefox remembers websites I visited years ago, I think I'd almost consider the browser useless if it forgets addresses after just 90 days.
some of the older ones are pure gold
Also notable, the internal list that this may or may not derive from: https://news.ycombinator.com/highlights
One thing that most LLMs do in their UIs are way too small fonts and too much information density.
For example, the "max pts" text has no reason to be this small, there is easily enough space to the left.
This concept has annoyed me so much that I decided to give it a name and create a bunch of shortcuts in my AI tools to stop it.
I call these things "sprinkles" Because I feel like AI can't stop sprinkling dumb shit all over its designs. Small text that annotates things, little callouts are examples but to me, the canonical thing that AI does when you ask it to build a UI that makes no sense is to create a bunch of stuff on the page that couldn't possibly come from any kind of deterministic database.
For example, if you ask it to sort and filter a library UI for a book finder, it'll add some element on a SERP page that's like "22 first time novelists" Even though that piece of information doesn't exist anywhere and couldn't possibly wire into a dynamic website.
Drives me crazy. Thought I'd share.
bookmarked. will never open again
There is a code freeze coming up at may job, I'll probably browse through some of the links in the spare time. So thanks for curating this.
There's no type for People which I would have thought has provided some of the all time high scoring submissions.
Very cool. Do you have any plans to aggregate book recommendations and papers (via doi perhaps or other link hints) as well?
i think books already exist https://hackernewsbooks.com/
i've previously done wikipedia and youtube links popular on hacker news: https://www.orangecrumbs.com/ and https://www.orangecrumbs.com/v
This is cool! Had to scroll way too far to find "wat" but it's a candidate.
>How Hall induction works
Well there's a thing called a Hall Effect sensor . . .
Looks cool! Some of the listed posts are such a blast from the past
Thank you for sharing this extensive work. I’ve bookmarked it; I’ll likely analyze each category in detail—covering the topics, ideas, and statistics—whenever I find the time.
Love these websites! Design is very cool as well
The ole legendary blog post. Get these submitted to the National Register of Historic Blog Posts.
Can 3100 links be legendary? I feel like there can be at most 310 legendary links
This was also my first instinct. Wikipedia says the US Baseball Hall of Fame has 351 inductees currently, for a sport that has existed in professional form since 1876. That feels like a more appropriate rate for this kind of enshrinement. Over 3000 in the span of less than 20 years is far too fast. The entire English language literary canon doesn't have that many entries.
on the other hand, there are much more links than baseball players. also there are only around 1k hall of fame links, the rest are candidates (i.e. have not met all requirements - 5+ submissions · 3+ distinct submitters · 3+ active years · 2 qualifying breakout events from different submitters at least 720 days apart)
i do accept thought that i have exaggerated in the title)
Please elaborate. What issue do you have with the stated methodology?
fair comment, i've probably exaggerated somewhat, but i think there are quite a few links outside of top 300 that are very interesting
I was initially skeptical, but after looking through it I see some articles that I distinctly remember (eg Gervais Principle) and blogs like SSC which led me down rabbitholes.
Lots of good things come from the orange site. I’m sure myself and others will dig through these to find those hidden gems that we missed :)
Good initiative