Of all the people on the entire internet, I would hope HN posters understand best that anything and everything posted online already has and also will at some point be used in such ways.
* Nvidia GPUs will see heavy competition and most chat-like use-cases switching to cheaper models and inference-specific-silicon but will be still used on the high end for critical applications and frontier science
* Most Software and UIs will be primarily AI-generated. There will be no 'App Stores' as we know them.
* ICE Cars will become niche and will be largely been replaced with EVs, Solar will be widely deployed and will be the dominate source of power
* Climate Change will be widely recognized due to escalating consequences and there will be lots of efforts in mitigations (e.g, Climate Engineering, Climate-resistant crops, etc).
It's fun to read some of these historic comments! A while back I wrote a replay system to better capture how discussions evolved at the time of these historic threads. Here's Karpathy's list from his graded articles, in the replay visualizer:
Notable how this is only possible because the website is a good "web citizen." It has urls that maintain their state over a decade. They contain a whole conversation. You don't have to log in to see anything. The value of old proper websites increases with our ability to process them.
> Everything we do today might be scrutinized in great detail in the future because it will be "free".
s/"free"/stolen/
The bit about college courses for future prediction was just silly, I'm afraid: reminds me of how Conan Doyle has Sherlock not knowing Earth revolves around the Sun. Almost all serious study concerns itself with predicting, modelling and influence over the future behaviour of some system; the problem is only that people don't fucking listen to the predictions of experts. They aren't going to value refined, academic general-purpose futurology any more than they have in the past; it's not even a new area of study.
It would be very interesting to see this applied year after year to see if people get better or worse over time in the accuracy of their judgments.
It would also be interesting to correlate accuracy to scores, but I kind of doubt that can be done. Between just expressing popular sentiment and the first to the post people getting more votes for the same comment than people who come later it probably wouldn’t be very useful data.
I have never felt less confident in the future than I do in 2025... and it's such a stark contrast. I guess if you split things down the middle, AI probably continues to change the world in dramatic ways but not in the all or nothing way people expect.
A non trivial amount of people get laid off, likely due to a finanical crisis which is used as an excuse for companies scale up use of AI. Good chance the financial crisis was partly caused by AI companies, which ironically makes AI cheaper as infra is bought up on the cheap (so there is a consolidation, but the bountiful infra keeps things cheap). That results in increased usage (over a longer period of time). and even when the economy starts coming back the jobs numbers stay abismal.
Politics are divided into 2 main groups, those who are employed, and those who are retired. The retired group is VERY large, and has alot of power. They mostly care about entitlements. The employed age people focus on AI which is making the job market quite tough. There are 3 large political forces (but 2 parties). The Left, the Right, and the Tech Elite. The left and the right both hate AI, but the tech elite though a minority has outsized power in their tie breaker role. The age distributions would surprise most. Most older people are now on the left, and most younger people are split by gender. The right focuses on limiting entitlements, and the left focuses on growing them by taxing the tech elite. The right maintains power by not threatening the tech elite.
Unlike the 20th century America is a more focused global agenda. We're not policing everyone, just those core trading powers. We have not gone to war with China, China has not taken over Taiwan.
Physical robotics is becoming a pretty big thing, space travel is becoming cheaper. We have at least one robot on an astroid mining it. The yield is trivial, but we all thought it was neat.
Energy is much much greener, and you wouln't have guessed it... but it was the data centers that got us there. The Tech elite needed it quickly, and used the political connections to cut red tape and build really quickly.
Interesting experiment. Using modern LLMs to retroactively grade decade-old HN discussions is a clever way to measure how well our collective predictions age. It’s impressive how little time and compute it now takes to analyze something that would’ve required days of manual reading. My only caution is that hindsight grading can overvalue outcomes instead of reasoning — good reasoning can still lead to wrong predictions. But as a tool for calibrating forecasting and identifying real signal in discussions, this is a very cool direction.
> But if intelligence really does become too cheap to meter, it will become possible to do a perfect reconstruction and synthesis of everything. LLMs are watching (or humans using them might be). Best to be good.
I cannot believe this is just put out there unexamined of any level of "maybe we shouldn't help this happen". This is complete moral abdication. And to be clear, being "good" is no defense. Being good often means being unaligned with the powerful, so being good is often the very thing that puts you in danger.
I'm delighted to see that one of the users who makes the same negative comments on every Google-related post gets a "D" for saying Waymo was smoke and mirrors. Never change, I guess.
This is a cool idea. I would install a Chrome extension that shows a score by every username on this site grading how well their expressed opinions match what subsequently happened in reality, or the accuracy of any specific predictions they've made. Some people's opinions are closer to reality than others and it's not always correlated with upvotes.
An extension of this would be to grade people on the accuracy of the comments they upvote, and use that to weight their upvotes more in ranking. I would love to read a version of HN where the only upvotes that matter are from people who agree with opinions that turn out to be correct. Of course, only HN could implement this since upvotes are private.
I was reading the Anki article on 2015-12-13, and the best prediction was by markm248 saying: "Remember that you read it here first, there will be a unicorn built on the concept of SRS"
I noticed the Hall of Fame grading of predictive comments has a quirk? It grades some comments about if they came true or not, but in the grading of comment to the article
The Cannons on the B-29 Bomber
"accurate account of LeMay stripping turrets and shifting to incendiary area bombing; matches mainstream history"
It gave a good grade to user cstross but to my reading of the comment, cstross just recounted a bit of old history. The evaluation gave cstross for just giving a history lesson or no?
This is a perfect example of the power and problems with LLMs.
I took the narcissistic approach of searching for myself. Here's a grade of one of my comments[1]:
>slg: B- (accurate characterization of PH’s “networking & facade” feel, but implicitly underestimates how long that model can persist)
And here's the actual comment I made[2]:
>And maybe it is the cynical contrarian in me, but I think the "real world" aspect of Product Hunt it what turned me off of the site before these issues even came to the forefront. It always seemed like an echo chamber were everyone was putting up a facade. Users seemed more concerned with the people behind products and networking with them than actually offering opinions of what was posted.
>I find the more internet-like communities more natural. Sure, the top comment on a Show HN is often a critique. However I find that more interesting than the usual "Wow, another great product from John Developer. Signing up now." or the "Wow, great product. Here is why you should use the competing product that I work on." that you usually see on Product Hunt.
I did not say nor imply anything about "how long that model can persist", I just said I personally don't like using the site. It's a total hallucination to claim I was implying doom for "that model" and you would only know that if you actually took the time to dig into the details of what was actually said, but the summary seems plausible enough that most people never would.
The LLM processed and analyzed a huge amount of data in a way that no human could, but the single in-depth look I took at that analysis was somewhere between misleading and flat out wrong. As I said, a perfect example of what LLMs do.
And yes, I do recognize the funny coincidence that I'm now doing the exact thing I described as the typical HN comment a decade ago. I guess there is a reason old me said "I find that more interesting".
I'm not so sure; that may not have been what you meant, but that doesn't mean it's not what others read into it. The broader context is HN is a startup forum and one of the most common discussion patterns is 'I don't like it' that is often a stand-in for 'I don't think it's viable as-is'. Startups are default dead, after all.
With that context, if someone were to read your comment and be asked 'does this person think the product's model is viable in the long run' I think a lot of people would respond 'no'.
And this is a perfect example of how some people respond to LLMs, bending over backwards to justify the output like we are some kids around a Ouija board.
The LLM isn't misinterpreting the text, it's just representing people who misinterpreted the text isn't the defense you seem to think it is.
And your response here is a perfect example of confidently jumping to conclusions on what someone's intent is... which is exactly what you're saying the LLM did to you.
I scoped my comment specifically around what a reasonable human answer would be if one were asked the particular question it was asked with the available information it had. That's all.
Btw I agree with your comment that it hallucinated/assumed your intent! Sorry I did not specify that. This was a bit of a 'play stupid games win stupid prizes' prompt by the OP. If one asks an imprecise question one should not expect a precise answer. The negative externality here is reader's takeaways are based on false precision. So is it the fault of the question asker, the readers, the tool, or some mix? The tool is the easiest to change, so probably deserves the most blame.
I think we'd both agree LLMs are notoriously overly-helpful and provide low confidence responses to things they should just not comment on. That to me is the underlying issue - at the very least they should respond like humans do not only in content but in confidence. It should have said it wasn't confident about its response to your post, and OP should have thus thrown its response out.
Rarely do we have perfect info, in regular communications we're always making assumptions which affect our confidence in our answers. The question is what's the confidence threshold we should use? This is the question to ask before the question of 'is it actually right?', which is also an important question to ask, but one I think they're a lot better at than the former.
Fwiw you can tell most LLMs to update its memory to always give you a confidence score 0.0-1.0. This helps tremendously, it's pretty darn accurate, it's something you can program thresholds around, and I think it should be built in to every LLM response.
The way I see it, LLMs have lots and lots of negative externalities that we shouldn't bring into this world (I'm particularly sensitive to the effects on creative industries), and I detest how they're being used so haphazardly, but they do have some uses we also shouldn't discount and figure out how to improve on. The question is where are we today in that process?
The framework I use to think about how LLMs are evolving is that of transitioning mediums. Like movies started as a copy/paste of stage plays before they settled into the medium and understand how to work along the grain of its strengths & weaknesses to create new conventions. Speech & text are now transitioning into LLMs. What is the grain we need to go along?
My best answer is the convention LLMs need to settle into is explicit confidence, and each question asked of them should first be a question of what the acceptable confidence threshold is for such a question. I think every question and domain will have different answers for that, and we should debate and discuss that alongside any particular answer.
> I spent a few hours browsing around and found it to be very interesting.
This seems to be the result of the exercise? No evaluation?
My concern is that, even if the exercise is only an amusing curiosity, many people will take the results more seriously than they should, and be inspired to apply the same methods to products and initiatives that adversely affect people's lives in real ways.
I've spent a weekend making something similar for my gmail account (which google keeps nagging me about being 90% full). It's fascinating to be able to classify 65k+ of emails (surprise: more than half are garbage), as well as summarize and trace the nature of communication between specific senders/recipients. It took about 50 hours on a dual RTX 3090 running Qwen 3.
My original goal was to prune the account deleting all the useless things and keeping just the unique, personal, valuable communications -- but the other day, an insight has me convinced that the safer / smarter thing to do in the current landscape is the opposite: remove any personal, valuable, memorable items, and leave google (and whomever else is scraping these repositories) with useless flotsam of newsletters, updates, subscription receipts, etc.
79 comments
[ 4.4 ms ] story [ 74.4 ms ] threadShades of Roko's Basilisk!
Your past thoughts have been dredged up and judged.
For each $TOPIC, you have been awarded a grade by GPT-5.1 Thinking.
Your grade is based on OpenAI's aligned worldview and what OpenAI's blob of weights considers Truth in 2025.
Did you think well, netizen?
Are you an Alpha or a Delta-Minus?
Where will the dragnet grading of your online history happen next?
* Nvidia GPUs will see heavy competition and most chat-like use-cases switching to cheaper models and inference-specific-silicon but will be still used on the high end for critical applications and frontier science
* Most Software and UIs will be primarily AI-generated. There will be no 'App Stores' as we know them.
* ICE Cars will become niche and will be largely been replaced with EVs, Solar will be widely deployed and will be the dominate source of power
* Climate Change will be widely recognized due to escalating consequences and there will be lots of efforts in mitigations (e.g, Climate Engineering, Climate-resistant crops, etc).
Swift is Open Source https://hn.unlurker.com/replay?item=10669891
Launch of Figma, a collaborative interface design tool https://hn.unlurker.com/replay?item=10685407
Introducing OpenAI https://hn.unlurker.com/replay?item=10720176
The first person to hack the iPhone is building a self-driving car https://hn.unlurker.com/replay?item=10744206
SpaceX launch webcast: Orbcomm-2 Mission [video] https://hn.unlurker.com/replay?item=10774865
At Theranos, Many Strategies and Snags https://hn.unlurker.com/replay?item=10799261
s/"free"/stolen/
The bit about college courses for future prediction was just silly, I'm afraid: reminds me of how Conan Doyle has Sherlock not knowing Earth revolves around the Sun. Almost all serious study concerns itself with predicting, modelling and influence over the future behaviour of some system; the problem is only that people don't fucking listen to the predictions of experts. They aren't going to value refined, academic general-purpose futurology any more than they have in the past; it's not even a new area of study.
And scroll down to the bottom.
It would be very interesting to see this applied year after year to see if people get better or worse over time in the accuracy of their judgments.
It would also be interesting to correlate accuracy to scores, but I kind of doubt that can be done. Between just expressing popular sentiment and the first to the post people getting more votes for the same comment than people who come later it probably wouldn’t be very useful data.
A non trivial amount of people get laid off, likely due to a finanical crisis which is used as an excuse for companies scale up use of AI. Good chance the financial crisis was partly caused by AI companies, which ironically makes AI cheaper as infra is bought up on the cheap (so there is a consolidation, but the bountiful infra keeps things cheap). That results in increased usage (over a longer period of time). and even when the economy starts coming back the jobs numbers stay abismal.
Politics are divided into 2 main groups, those who are employed, and those who are retired. The retired group is VERY large, and has alot of power. They mostly care about entitlements. The employed age people focus on AI which is making the job market quite tough. There are 3 large political forces (but 2 parties). The Left, the Right, and the Tech Elite. The left and the right both hate AI, but the tech elite though a minority has outsized power in their tie breaker role. The age distributions would surprise most. Most older people are now on the left, and most younger people are split by gender. The right focuses on limiting entitlements, and the left focuses on growing them by taxing the tech elite. The right maintains power by not threatening the tech elite.
Unlike the 20th century America is a more focused global agenda. We're not policing everyone, just those core trading powers. We have not gone to war with China, China has not taken over Taiwan.
Physical robotics is becoming a pretty big thing, space travel is becoming cheaper. We have at least one robot on an astroid mining it. The yield is trivial, but we all thought it was neat.
Energy is much much greener, and you wouln't have guessed it... but it was the data centers that got us there. The Tech elite needed it quickly, and used the political connections to cut red tape and build really quickly.
The EU may give LLM surveillance an F at some point.
I cannot believe this is just put out there unexamined of any level of "maybe we shouldn't help this happen". This is complete moral abdication. And to be clear, being "good" is no defense. Being good often means being unaligned with the powerful, so being good is often the very thing that puts you in danger.
It does seem better than just upvotes and downvotes though.
An extension of this would be to grade people on the accuracy of the comments they upvote, and use that to weight their upvotes more in ranking. I would love to read a version of HN where the only upvotes that matter are from people who agree with opinions that turn out to be correct. Of course, only HN could implement this since upvotes are private.
They were right, Duolingo.
https://news.ycombinator.com/item?id=10654216
The Cannons on the B-29 Bomber "accurate account of LeMay stripping turrets and shifting to incendiary area bombing; matches mainstream history"
It gave a good grade to user cstross but to my reading of the comment, cstross just recounted a bit of old history. The evaluation gave cstross for just giving a history lesson or no?
I took the narcissistic approach of searching for myself. Here's a grade of one of my comments[1]:
>slg: B- (accurate characterization of PH’s “networking & facade” feel, but implicitly underestimates how long that model can persist)
And here's the actual comment I made[2]:
>And maybe it is the cynical contrarian in me, but I think the "real world" aspect of Product Hunt it what turned me off of the site before these issues even came to the forefront. It always seemed like an echo chamber were everyone was putting up a facade. Users seemed more concerned with the people behind products and networking with them than actually offering opinions of what was posted.
>I find the more internet-like communities more natural. Sure, the top comment on a Show HN is often a critique. However I find that more interesting than the usual "Wow, another great product from John Developer. Signing up now." or the "Wow, great product. Here is why you should use the competing product that I work on." that you usually see on Product Hunt.
I did not say nor imply anything about "how long that model can persist", I just said I personally don't like using the site. It's a total hallucination to claim I was implying doom for "that model" and you would only know that if you actually took the time to dig into the details of what was actually said, but the summary seems plausible enough that most people never would.
The LLM processed and analyzed a huge amount of data in a way that no human could, but the single in-depth look I took at that analysis was somewhere between misleading and flat out wrong. As I said, a perfect example of what LLMs do.
And yes, I do recognize the funny coincidence that I'm now doing the exact thing I described as the typical HN comment a decade ago. I guess there is a reason old me said "I find that more interesting".
[1] - https://karpathy.ai/hncapsule/2015-12-18/index.html#article-...
[2] - https://news.ycombinator.com/item?id=10761980
With that context, if someone were to read your comment and be asked 'does this person think the product's model is viable in the long run' I think a lot of people would respond 'no'.
The LLM isn't misinterpreting the text, it's just representing people who misinterpreted the text isn't the defense you seem to think it is.
I scoped my comment specifically around what a reasonable human answer would be if one were asked the particular question it was asked with the available information it had. That's all.
Btw I agree with your comment that it hallucinated/assumed your intent! Sorry I did not specify that. This was a bit of a 'play stupid games win stupid prizes' prompt by the OP. If one asks an imprecise question one should not expect a precise answer. The negative externality here is reader's takeaways are based on false precision. So is it the fault of the question asker, the readers, the tool, or some mix? The tool is the easiest to change, so probably deserves the most blame.
I think we'd both agree LLMs are notoriously overly-helpful and provide low confidence responses to things they should just not comment on. That to me is the underlying issue - at the very least they should respond like humans do not only in content but in confidence. It should have said it wasn't confident about its response to your post, and OP should have thus thrown its response out.
Rarely do we have perfect info, in regular communications we're always making assumptions which affect our confidence in our answers. The question is what's the confidence threshold we should use? This is the question to ask before the question of 'is it actually right?', which is also an important question to ask, but one I think they're a lot better at than the former.
Fwiw you can tell most LLMs to update its memory to always give you a confidence score 0.0-1.0. This helps tremendously, it's pretty darn accurate, it's something you can program thresholds around, and I think it should be built in to every LLM response.
The way I see it, LLMs have lots and lots of negative externalities that we shouldn't bring into this world (I'm particularly sensitive to the effects on creative industries), and I detest how they're being used so haphazardly, but they do have some uses we also shouldn't discount and figure out how to improve on. The question is where are we today in that process?
The framework I use to think about how LLMs are evolving is that of transitioning mediums. Like movies started as a copy/paste of stage plays before they settled into the medium and understand how to work along the grain of its strengths & weaknesses to create new conventions. Speech & text are now transitioning into LLMs. What is the grain we need to go along?
My best answer is the convention LLMs need to settle into is explicit confidence, and each question asked of them should first be a question of what the acceptable confidence threshold is for such a question. I think every question and domain will have different answers for that, and we should debate and discuss that alongside any particular answer.
This seems to be the result of the exercise? No evaluation?
My concern is that, even if the exercise is only an amusing curiosity, many people will take the results more seriously than they should, and be inspired to apply the same methods to products and initiatives that adversely affect people's lives in real ways.
My original goal was to prune the account deleting all the useless things and keeping just the unique, personal, valuable communications -- but the other day, an insight has me convinced that the safer / smarter thing to do in the current landscape is the opposite: remove any personal, valuable, memorable items, and leave google (and whomever else is scraping these repositories) with useless flotsam of newsletters, updates, subscription receipts, etc.
Any chance you can outline the steps/prompts/tools you used to run this?
I've been building a 2nd brain type project, that plugs into all my work places and a custom classifier has been on that list that would enhance that.
Compared to what happens next? Does tptacek's commentary become market signal equivalent to the Fed Chair or the BLS labor and inflation reports?