For desktop users, it seems intended to be navigated as a slide deck (with arrow keys or space bar to advance). Nevertheless, I agree that this scroll-jacking is a crime against human decency.
I found it amusing being told what humans cared about, while this human was caring about waiting for the page to finish animating the text I wanted to read
I actually normally don’t hate scrolljacking if it’s done right but this particular website’s scroll handler is so laggy as to basically be unusable.
Edit: I also would like to be given paragraphs rather than having to read one word at a time and sit through a transition scene each time I try to read the next.
Edit 2: I gave up reading it halfway through because scrolling too fast ran out of memory and crashed my (mobile) browser.
This website is really begging to just be a video with some voiceover. There are segments that complete automatically, and others where you must scroll to continue the animation without any nudge. I think a better example would be something like https://animejs.com/ where the entire page is animated, but your scrolling (which works as expected without any smoothing) just influences the speed of the "animation".
The way the website over-animates my scrolling, to the point of going multiple seconds behind the input; I just ended up scrolling a bunch, and watching the website as a silent video.
A video would have been been simpler and easier to navigate. A video would have been faster as well; even when scrolling ahead i would have liked a 150% option.
It's not true that LLMs can only perform as well as the average of their training data. They can learn to perform at an elite level in domains like programming by extracting quality signals from code reviews, bug fixes, tutorials, etc, and the labs are doing post-training to teach them what excellent code looks like.
Maybe right now that's only true in a few domains like math, programming, language translation, maybe some writing task? But soon it will be true in many more domains.
Given that we can't hope to out-execute the LLMs on the things they are elite at, what is left for us? This site wants to reserve "taste" for the humans, but I'm not sure LLMs won't learn "taste" too.
The best I've got right now is that humans live and interact with the real world in ways that disembodied LLMs can't (yet). Maybe they will be the theoreticians and we the experimentalists?
The world is pretty diverse. There’s not an objective correct answer to a lot of things. Leaving that decision up to an LLM is acceptable sometimes, and not in others.
I’m not sure how you train for that. Though, maybe the world just converges on homogeneity out of convenience. Certainly happens in other areas.
> It's not true that LLMs can only perform as well as the average of their training data. They can learn to perform at an elite level in domains like programming by extracting quality signals from code reviews, bug fixes, tutorials, etc, and the labs are doing post-training to teach them what excellent code looks like.
That's just adding more training data, out of which it will take the average
I think we should not take part in an arms’ race with the LLMs. The thing that LLMs structurally cannot do is care: care is choosing to invest attention in s preferred object, even if this object can change you. But current LLMs cannot choose (they follow a prompt), they do not prefer (averaged priors), and they cannot change (immutable weights). If they were able to do so those they would become entities, which are much harder to control and sell.
Care is why we choose to remain relevant even though we don’t need to to produce.
Frankly, I give up reading a piece as soon as I recognize that tired, punchy marketing-speak. Looks visually appealing though. Might be cooler if scrolling completely controlled the animation playback. Scrolling is currently a nuisance.
Same. No idea what % I made it through, but had to tap out eventually. When are people going to learn that scroll driven animations are almost always a crime?
I quickly gave up scrolling and clicked “watch” at the top. I still didn’t make it all the way through. It’s pretty, I like the art, but it was just too much.
Someone else pointed out you can just read the file at the button of the index dropdown
the `.md` in the name made me think i'd be able to _read_ something. i'm a reader. watching videos or animated images or scrolling to get spoonfed little bits of animated text is not for me.
Love it! I actually like the scrolljacking as it forces you to take your time and actually take in the sentiment. So easy to skim everything we read on the web.
Sorry - can't dedicate that much time to simply read a few dozen words of text. The irony is thick here - as a .md it purports to be a plain text file but nooooo, it's ~50MB of janky, slow, unskippable animations which each reveal 3-5 words at a time.
I think if you're asking what humans "bring to the table" you've already lapsed into misantrhopy. I don't want to live in a world where my art is evaluated by non-humans, I want a separate internet where it's not even visible to them.
This is a beautiful expression, and the author would do well to refine their notion of taste further.
Anyone familiar with Dr. John Vervaeke's work will find another notion, relevance realization (RR), fits nicely.
The thing I like about RR over "taste" when talking about AI is that it captures what the taste metaphor does not - that there is a missing _relevant_ dimension of consideration from the LLMs output.
"Hey you got the code working but this 4000 line PR will never be approved."
"You spent tokens hardening the API, when this entire project is just a local test that will never hit production". Etc, etc, ad nauseum.
Vervaeke reveals why in his lecture series "Awakening from the Meaning Crisis".
Fundamentally, RR cannot be achieved through reason alone. Full stop. Vervaeke uses this fact to point out the irrationality of the "rationalist" movement in order to highlight its dangers (which many would agree the effective altruism crowd has thoroughly proven him right about), but the same framing and argument can be used to explain why your AI has bad taste.
There is no way to enumerate the set of dimensions that are or are not relevant in a given real world context. The detail of the universe, combined with the set of actions an agent does (and does NOT!) take leads to a combinatorial explosion of complexity.
AI agents need to be told what to do, and even then, they will only stochastically sample what you tell them.
Anthropic tells Claude lots of things about Rust, for example, and things to look out for in security context, but they fundamentally cannot put enough information in to grant the agent the ability to wisely choose what is and is not relevant. You can only ever consider _some_ relevant things.
I don't personally think it is impossible for computers to perform RR, it's just that today's agents do not, which is why its taste and yours differ.
> For me, the webpage just looked like a markdown file.
Thanks. It was irritating trying to read the jsGUI version and so I quit. I came across your comment and tried 'lynx'. It really is just an MD file: short, easy to read, straight to the point.
The comment section… cmon people, did you even read it! The whole point is to ship stuff with a human touch - not be correct. Fine, the scrolling is wonky. Who cares. That’s what makes it human. I, for one, thought it was creative and artistic, with a message that resonated with me. Bravo.
I didn't because it was painful to read. It would help if they had a skip to text. Also the text I _did_ read certainly did seem to have been not actually human written.
when combined in one paragraph they do sound AI-ish but that's not how the text was presented, it's in chunks. i suggest not blindly trusting those AI checkers although i'm not sure if it's AI either. but the creator is the only one who can give a correct claim
This seems to apply to emotional work like design. It's a bit far from the docs and coding I usually do, and for ideation AI definitely seems to spit out average stuff. It's good for searching the information that becomes the material for it, but it definitely seems like the work only goes well when you set the keywords well.
What an obnoxious website. If you want to read the entire text, without a pointless need to scroll and see 100000 animations: https://pastebin.com/p07iBVYg
This is AI-generated, and I seriously mistrust the judgement of anybody who can’t tell so at a glance. Set aside that reliable tools can confidently ID it as AI, set aside the blatant AI writing style, the important tell is fundamental: guys, this is extremely shallow, meaningless content.
It’s “in plain text” and also the most whimsical excess you’ve seen today. Both of these things are “very human”. No further thought, your honor.
It’s a manifesto and notes-to-self, not a thinkpiece. No further thought, your honor.
It’s not a thread or a deck. You’re reading this on a website. No further thought, your honor.
“Symmetry is a rule. Break it. Three-act structure is a rule. Break it. "Don't shoot into the sun" is a rule. Break it.” This is itself a bizarre symmetrical trifecta. No further thought, your honor.
Near the beginning it’s “last edited at the kitchen table, coffee #2”, near the end “last touched, kitchen table, late”. No further thought, your honor.
I don’t care to sacrifice more of my life to the rest of this, I’m sure there’s worse to dig for.
Propaganda works not by stating the message, which makes it only something a person said, but by assuming it, which makes it something you’ve figured out about the world. Of course, this is very shallow, and accidentally says the weaponized assumption plainly, too: “the machines are very good and getting better every day”.
If they have to resort to this, and if this is the best they can do at it, I’m really not worried about “figuring out what’s left for humans”.
“Taste” means you are still just a sensor for the machine, noticing what it yet can’t, and telling it so it can reward you for being a good sensor, and next time replace you successfully.
abahlo | 21 hours ago
mihaaly | 21 hours ago
Scrolljacking I hate too.
staticshock | 21 hours ago
Slow_Hand | 21 hours ago
what | 21 hours ago
computomatic | 21 hours ago
daemonologist | 21 hours ago
shearerp | 20 hours ago
Expected markdown document, got scrolljacked content payload from hell.
broodbucket | 20 hours ago
odo1242 | 20 hours ago
Edit: I also would like to be given paragraphs rather than having to read one word at a time and sit through a transition scene each time I try to read the next.
Edit 2: I gave up reading it halfway through because scrolling too fast ran out of memory and crashed my (mobile) browser.
0gs | 19 hours ago
noumenon1111 | 19 hours ago
miguel-muniz | 18 hours ago
aDyslecticCrow | 21 hours ago
A video would have been been simpler and easier to navigate. A video would have been faster as well; even when scrolling ahead i would have liked a 150% option.
mahboi | 12 hours ago
what | 21 hours ago
xydac | 21 hours ago
bpev | 21 hours ago
SwellJoe | 21 hours ago
It's also a high bullshit design...a lot of random shit happening for no reason other than to make it hard to read and weird to navigate.
Maybe good intentions, but I can't get past the surface level stuff that's very off-putting.
hamuraijack | 21 hours ago
jrdnfrd | 21 hours ago
Maybe right now that's only true in a few domains like math, programming, language translation, maybe some writing task? But soon it will be true in many more domains.
Given that we can't hope to out-execute the LLMs on the things they are elite at, what is left for us? This site wants to reserve "taste" for the humans, but I'm not sure LLMs won't learn "taste" too.
The best I've got right now is that humans live and interact with the real world in ways that disembodied LLMs can't (yet). Maybe they will be the theoreticians and we the experimentalists?
popkkksidhd | 21 hours ago
I’m not sure how you train for that. Though, maybe the world just converges on homogeneity out of convenience. Certainly happens in other areas.
FrameworkFred | 20 hours ago
875323897533 | 18 hours ago
BrainInAJar | 14 hours ago
That's just adding more training data, out of which it will take the average
deimos_28 | 11 hours ago
meshweaver | 21 hours ago
vunderba | 21 hours ago
https://imgur.com/a/3YuSVrD
Seriously - these are so bad I almost want to believe this was done using Stable Diffusion 1.5 on purpose. Or maybe they used a Simpson LoRA.
epgui | 21 hours ago
QuercusMax | 21 hours ago
nchmy | 15 hours ago
40four | 15 hours ago
Someone else pointed out you can just read the file at the button of the index dropdown
justusthane | 4 hours ago
tiborsaas | 21 hours ago
OuterVale | 20 hours ago
redanddead | 20 hours ago
Luke2642 | 20 hours ago
FrameworkFred | 20 hours ago
achairapart | 20 hours ago
[0]: https://humanstxt.org
achairapart | 12 hours ago
https://www.google.com/humans.txt
https://www.theguardian.com/humans.txt
https://medium.com/humans.txt
https://www.tumblr.com/humans.txt
https://www.nytimes.com/humans.txt
https://www.netflix.com/humans.txt
https://flickr.com/humans.txt
and... https://disqus.com/humans.txt
account42 | 4 hours ago
penguin_booze | 11 hours ago
Whom.
MetaMalone | 20 hours ago
0x4e | 20 hours ago
dsjoerg | 20 hours ago
dsjoerg | 20 hours ago
fg137 | 20 hours ago
gregrobson | 20 hours ago
a3w | 20 hours ago
smusamashah | 20 hours ago
happyopossum | 20 hours ago
xnx | 20 hours ago
gcanyon | 20 hours ago
That page is anything but plain text...
decidu0us9034 | 19 hours ago
mjdiloreto | 19 hours ago
Anyone familiar with Dr. John Vervaeke's work will find another notion, relevance realization (RR), fits nicely.
The thing I like about RR over "taste" when talking about AI is that it captures what the taste metaphor does not - that there is a missing _relevant_ dimension of consideration from the LLMs output.
"Hey you got the code working but this 4000 line PR will never be approved." "You spent tokens hardening the API, when this entire project is just a local test that will never hit production". Etc, etc, ad nauseum.
Vervaeke reveals why in his lecture series "Awakening from the Meaning Crisis".
Fundamentally, RR cannot be achieved through reason alone. Full stop. Vervaeke uses this fact to point out the irrationality of the "rationalist" movement in order to highlight its dangers (which many would agree the effective altruism crowd has thoroughly proven him right about), but the same framing and argument can be used to explain why your AI has bad taste.
There is no way to enumerate the set of dimensions that are or are not relevant in a given real world context. The detail of the universe, combined with the set of actions an agent does (and does NOT!) take leads to a combinatorial explosion of complexity.
AI agents need to be told what to do, and even then, they will only stochastically sample what you tell them.
Anthropic tells Claude lots of things about Rust, for example, and things to look out for in security context, but they fundamentally cannot put enough information in to grant the agent the ability to wisely choose what is and is not relevant. You can only ever consider _some_ relevant things.
I don't personally think it is impossible for computers to perform RR, it's just that today's agents do not, which is why its taste and yours differ.
spieke | 17 hours ago
Turns out, if you have Animations disabled in your Windows settings (aka prefers-reduce-motion), you see a simple text web page.
account42 | 4 hours ago
O1111OOO | 3 hours ago
Thanks. It was irritating trying to read the jsGUI version and so I quit. I came across your comment and tried 'lynx'. It really is just an MD file: short, easy to read, straight to the point.
holtkam2 | 17 hours ago
bikezen | 15 hours ago
tsunitsuni | 17 hours ago
oabragh | 4 hours ago
withinrafael | 15 hours ago
aalazz | 14 hours ago
lhk931122 | 14 hours ago
dmitrygr | 12 hours ago
needfish | 12 hours ago
makerofthings | 12 hours ago
mahboi | 12 hours ago
kqp | 11 hours ago
It’s “in plain text” and also the most whimsical excess you’ve seen today. Both of these things are “very human”. No further thought, your honor.
It’s a manifesto and notes-to-self, not a thinkpiece. No further thought, your honor.
It’s not a thread or a deck. You’re reading this on a website. No further thought, your honor.
“Symmetry is a rule. Break it. Three-act structure is a rule. Break it. "Don't shoot into the sun" is a rule. Break it.” This is itself a bizarre symmetrical trifecta. No further thought, your honor.
Near the beginning it’s “last edited at the kitchen table, coffee #2”, near the end “last touched, kitchen table, late”. No further thought, your honor.
I don’t care to sacrifice more of my life to the rest of this, I’m sure there’s worse to dig for.
Propaganda works not by stating the message, which makes it only something a person said, but by assuming it, which makes it something you’ve figured out about the world. Of course, this is very shallow, and accidentally says the weaponized assumption plainly, too: “the machines are very good and getting better every day”.
If they have to resort to this, and if this is the best they can do at it, I’m really not worried about “figuring out what’s left for humans”.
workfromspace | 6 hours ago
mkgiga | 6 hours ago
dreamcompiler | 6 hours ago
(Too cute; didn't read)
queenkjuul | 4 hours ago
It's very pretty but after 30 seconds i gave up so i have no idea what the point was
tomaskafka | 3 hours ago