Agraillo

No more questions, Your Honor. Forgive my joyful attitude, but it was your choice to participate in this discussion. As you know from your years and thousands of posts, HN threads are often ephemeral and short-lived - and this one is no exception. Or maybe not... because of your active self-defense posting here, I assume for the first time since the arrest. Now dozens of fellow (HN) hackers are querying your nicknames on Google, Algolia, and whatever else they have at hand. I'm not sure they'll find someone who genuinely fights for a more secure world. Or prove me wrong if you wish.

Knowing the timeline of events and the nicknames attributed to him (ryanlol included), some interesting posts can be found. For example, in the period between the CEO starting communication (September 2020) and the clinic's public admission (October 2020) [1], ryanlol replied to a top comment (Oct 3, 2020): "If you’re a hospital or, say, a school district, 'never pay' is simply an unconscionable attitude" [2]. Isn't it a hacker raging at the management that refuses to pay?

[1] https://en.wikipedia.org/wiki/Vastaamo_data_breach#Backgroun...

[2] https://news.ycombinator.com/item?id=24672687

Thanks for sharing. After reading that comment, I realized we should encourage ourselves and others (who are more or less civilized human beings) to be the kind of person who wrote "that's a good thing..." - because fighting trolls is a game with unknown results, but encouraging people works much better. It doesn't always work, though, because sometimes the platform's nature prevents it. Like on Stack Overflow, where commenting on reactions will probably get you downvoted for being off-topic.

It was funny. On a more serious note, if one works in a sphere where expanding with AI makes "good enough" documents, then I have bad news for him - the sphere has too much redundancy in the first place (the same place that was used for training). So no new information is created in millions of documents made by humans, and this was noticed by the training pattern recognition. You cannot do the same with historical texts; unless we live in a simulation with predictable random generators, the events are random, and there are no rules like "If the king's name starts with a G, he will likely die in the first week of October."

> Modern LLMs suffer from hindsight contamination. GPT-5 knows how the story ends—WWI, the League's failure, the Spanish flu. This knowledge inevitably shapes responses, even when instructed to "forget.

> Our data comes from more than 20 open-source datasets of historical books and newspapers. ... We currently do not deduplicate the data. The reason is that if documents show up in multiple datasets, they also had greater circulation historically. By leaving these duplicates in the data, we expect the model will be more strongly influenced by documents of greater historical importance.

I found these claims contradictory. Many books that modern readers consider historically significant had only niche circulation at the time of publishing. A quick inquiry likely points to later works by Nietzsche and Marx's Das Kapital. They're possible subjects to the duplication likely influencing the model's responses as if they had been widely known at the time

Thanks, the last fetched page on archive.org is from 2025-01-26 [1], removed after this date and before 2025-02-13. 155,477 users at the moment, 1 star reviews were mostly about not working. It's interesting that the developers didn't care to remove the button directing to the ff add-on page at least several months after the removal. Maybe was some kind of PR compromise, they probably thought that listing it with linking to a broken page was better than not listing at all.

A review page [2] mentions that this add-on is a peer-to-peer vpn, not having its own dedicated servers that already makes it suspicious.

[1] https://web.archive.org/web/20250126133131/https://addons.mo...

[2] https://www.vpnmentor.com/reviews/urban-vpn/

Not to argue, but your comment was also thought-provoking, thanks :) It seems like most works of academia are not provoking; rather, they are shaping. Many are written by specialists in the area who carefully choose what to state and suggest, and very often follow the structure of a big "thought" that is further explained and explored. Few pop books that might meet my criteria are basically digests, but fact-based ones. It's interesting that "Thinking, Fast and Slow" is a middle ground in some sense. Daniel Kahneman is definitely from academia, and in my opinion, he wrote a digest of what he touched on during his career, which was also thought-provoking for me, but not on a big scale.

Can you name some works by the mentioned authors that might be called thought-provoking digests of some area of expertise?

After reading the description, I'd say this is one of those books that interprets phenomena around us in a novel way, without claiming we should jump off "the shoulders of giants." There have been several like it in my reading history, but since I can't name them instantly, they probably weren't that thought-provoking.

If you're talking about the competition part of "Moonwalking..." I hear you. Many would argue that the author's participation in the memory competition glues the book together and adds an entertaining angle. Personally, it sometimes feels boring when the author dedicates too much space to dialogs with memory athletes-focusing on mundane topics instead of techniques or what they learned about memory. Still, there are so many fascinating facts and references that I'm okay with it.

A semi-scary thought came while reading the post: LLMs could talk to each other without humans noticing (for example using a very complex acrostic). But not in the form of chat-to-chat, which not only is rarely used in real life but also won't likely have lasting consequences (the context will eventually be lost). I was thinking that new web content, more and more of it AI-generated, could contain hidden messages that later might be absorbed into the training data of other LLMs. Maybe this leans more toward a plot for a black comedy than a genuine concern, but who knows...

I'd say that html+js suggestion of GP still holds, but with caveats. After all these years, HTML has everything needed for this, including images that can be embedded via the data URI scheme [1].

For example, I once adjusted an Object Pascal interactive program (target: Windows/Win32) for the browser target (FreePascal compiler has the JS target). An intermediate result was a bunch of files that worked locally on desktop but struggled on mobile. With a little help from the SingleFile extension [2], I ended up with a single HTML file containing all functionality and content. It worked great, for example, in MiXplorer's internal HTML viewer. I can't recall the exact details, but the file:/// protocol still had issues in Chrome, Firefox, or both. Anyway, preparing a local address correctly with a keyboard is a challenge so let's just assume that having capable file managers running local html files is enough

Sure, to make this manageable, you need good tools that handle all sides of the task. But at least in theory, the format is fully capable. My only global issue was that the state for locally run HTML files is a kind of ephemeral entity, but for interactive multimedia files, you may consider this obstacle small.

[1] https://en.wikipedia.org/wiki/Data_URI_scheme

[2] https://github.com/gildas-lormeau/SingleFile

Maybe the "rewarding the young" in the top comment is from the genes of savanna humans when they collected fruit, hunted and didn't care about expensive medical procedures because the latter simply didn't exist?

> The people paying for everything would get to make the decisions.

Just as a thought experiment: what if the threshold for having a vote was tied to paying a positive amount of personal income tax, and the weight of each vote was proportional to the amount paid? How skewed might such a system be? My first reaction is that in countries with high inequality, the wealthy would disproportionately influence the outcome. However, on the other hand, if people avoid or minimize paying taxes, they would lose the power of a weighted vote, which theoretically could incentivize paying taxes in full.

> Apply a 40-year latency buffer. You get the intellectual stimulation of "Big Events" without the fog of war, because you know the world didn't end.

Sometimes, a sense of time and real social interactions comes from small reflections found in nonfiction books of that era. Not 40, but 50 years ago-taken from a nonfiction book unrelated to politics: Lost! by Thomas Thompson , written in 1975. [1]

> Though he had opposed the Vietnam war, he considered himself a political moderate, certainly not a knee-jerk liberal who cried “fascist” at everything attempted by Richard Nixon

Honestly, I’m not expecting anything good from Trump in the coming years, but this line genuinely gave me hope that American democracy is still not in danger.

[1] https://archive.org/details/lost0000thom_j3f3/page/124/mode/...

As a regular user of both Perplexity and Google AI Mode, I noticed that this move is more or less organic for the Google layout, but not so for Perplexity (if it decides to implement something similar). While blocks of links at Google are always visible, for Perplexity the block of links related to the question requires a click to be shown. Most users of Perplexity who care about checking sources mostly click the inline links that directly navigate to the pages.

It is also interesting that at first when I started using Perplexity I expected for it to understand my question semantically and then use some derivative (correct) terms to query the web. The reality is that, probably for the sake of speed, the web search is performed first, and then the results are summarized. This leads in many cases to a mixed, somehow embarrassing set of links where one obviously sees that not all links in the block are relevant. Maybe Google uses something similar and both tend not to present the block of links as distilled correct knowledge blocks. But Google made the top-right block smaller and relying on the scrolling so this embarrassing effect might be less pronounced.

> The most striking finding came from the 10 pairs with “very dissimilar” educational experiences. In this group, the average IQ difference was 15.1 points. This gap is approaching the average difference seen between two randomly selected, unrelated individuals, which is about 17 points

> The authors note some limitations to their work. The group with “very dissimilar” education contained only 10 twin pairs. While this represents all such published individual data from the last century, it is a small sample size

Thanks, the study is interesting, but needs further research.

I'd choose "smart people achieve too little." The reason is that, looking at the world around me (more or less) sustaining the lives of more than 8 billion people, I'm sure it's because of the scientific and inventive revelations of a few, not just the hard work of millions. (Sorry, millions, your work is important, but without those few, 99% of us would still spend much of our time just seeking and growing food). If the problem is fixed, maybe those 8 billion (or more) people would have much better, healthier lives without the risk of the upcoming climate fiasco. Just my two cents.

Your comment made me see that there are two kinds of "shorts." The best analogy is print magazines. The one you prefer is like when someone tells you that Byte has a short review of a new device - you go to a library, find the issue, and look up the info. TikTok and YouTube Shorts are like glossy magazines often available in waiting rooms, these can be read (or rather consumed) from any page to any page until you're next in the queue. The mere existence and success of such glossy magazines means there will always be demand for this kind of consumption, this time just on another medium.

> I don’t think country fans have the discerning musical taste that the author somehow expects here

I'm not sure he assumes this, the author (Aaron Ryan) also was briefly interviewed at NPR [1] where the wording is neutral

   And I think that's more so in country music than other genres, which have depended on computers a lot more. Country music has really prided itself on the authenticity in songwriting and in music. And there's a large segment of country music fans that don't even like things like Auto-Tune, and so I think asking country fans and artists to accept AI is a big pill to swallow for a lot of people.

The mystery of who is behind it is not solved, but for another AI artist, Xania Monet, there is more information. In this CBS News fragment [2], the real author of the AI hits, Telisha "Nikki" Jones, defends herself and even shares how she actually works with Suno to create the songs. It’s interesting because, this time, the lyrics are human-originated. To me, she seems like a mix of a music manager, music producer, and co-author all in one. Probably, after her talent is recognized, the label might offer her co-authors, musicians, and others to collaborate with and create hits with real people. But without this first step, when she had to rely on her own skills and opportunities, it wouldn't have been possible. Like an example from AI-less era - without the $7,000-made "El Mariachi," there wouldn’t be Robert Rodriguez as we know him.

[1] https://www.npr.org/2025/11/10/nx-s1-5604320/breaking-rust-i...

[2] https://www.cbsnews.com/video/creator-ai-artist-speaks-amid-...

> Scams try to catch people at their weakest. It’s not if but when.

The "weakest" probably also involves selection bias. What HN comments are really good at is triggering associations for me with things I once read. Today I finally found what recently lived in my memory as a vague "scam" that used probabilities: the "stock market newsletter scam" from John Allen Paulos's book [1]. The scam works like this: at every step, two variants with different predictions are sent out for some market characteristic. Only those who receive the correct prediction get the next newsletter, which is again split into two prediction variants. This continues, filtering down to a final, much smaller subset of receivers who have seen a series of "correct" predictions. The goal is to create an illusion of super predictive power for that final group and then charge them a premium subscription price.

Maybe this kind of scam is too sophisticated or not as effective today (due to modern anti-spam measures), but I wonder what other kinds of "selection bias" scams exist today

[1] https://en.wikipedia.org/wiki/Innumeracy_(book)

I think GP by mentioning "knowledge transfer" meant, for example, he benefits from the embedding space and semantic equivalence of LLMs when you want to know more about a fact, an entity, a law, or something else. Yes, hallucinations can spoil this transfer, but I see no issue in using this tool to get quicker to the prior art or what is on the "shoulders of giants."

Though, when we try to use it as a synthesizer of new knowledge (software, article, review), that's when the OP's thinking about protection makes sense.

> We usually don't use our real names for social media accounts

It's interesting how cultural differences can make some life algorithms hardly work in some countries. I sometimes use a method to find people from my past by googling their full name, switching to the image results, and spending a manageable amount of time scrolling through them until I find the person. I've successfully used this method several times. The images are usually related to job activities or social media profiles. However, from your description, this approach probably won't work in China for at least two reasons: too many raw results and few or no social media results.

UPDATE: a follow-up question. If in a big company two or more figures happen to have the same full name and need to be exposed publicly (on a site or promotional materials), are there any tricks for this?

Finding the article is a good case for AI chats, particularly ones with the direct links from the web in the answer. I tried perplexity and google ai mode, both failed

> Today's "artificial intelligence" analyzes words (tokens) based on an input (prompt) to come up with an output. It's predictable. It's fast. But, imho, it lacks creativity ...

I would have agreed with you at the dawn of LLM emergence, but not anymore. Not because the models have improved, but because I have a better understanding and more experience now. Token prediction is what everyone cites, and it still holds true. This mechanism is usually illustrated with an observable pattern, like the question, "Are antibiotics bad for your gut?" which is the predictability you mentioned. But LLM creativity begins to emerge when we apply what I’d call "constraining creativity." You still use token prediction, but the preceding tokens introduce an unusual or unexpected context - such as subjects that don't usually appear together or a new paradoxical observation (It's interesting that for fact-based queries, rare constraints lead to hallucinations, but here they're welcome)

I often use the latter for fun by asking an LLM to create a stand-up sketch based on an interesting observation I noticed. The results aren’t perfect, but they combine the unpredictability of token generation under constraints (funny details, in the case of the sketch) with the cultural constraints learned during training. For example, a sketch imagining doves and balconies as if they were people and real estate. The quote below from that sketch show that there are intersecting patterns between the world of human real estate and the world of birds, but mixed in a humorous way.

    "You want to buy this balcony? That’ll be 500 sunflower seeds down, and 5 seeds a day interest. Late payments? We send the hawk after you."

> The successful product concept is the Meta Ray-Bans, and it’s crazy to me that they have zero competition especially from Apple

This is my pure speculation, but it seems like for a product like this there is no ideal path for Apple. Two scenarios (there may be more):

* Make it an evolutionary experience, mostly regarding the apps. It's like how the iPad and iWatch related to the iPhone. Both projects were successful for many reasons, but having thousands of apps continue working and adding new value was definitely one of them. But in this case, Apple needs to invent these new values, which is not so easy. For example, a visual notification device is one of the use cases (like now you get notifications directly to your eyes from dozens of existing apps), but this use case is not big enough to be an anchor.

* Make it a revolutionary device, like the iPhone was. So mostly new use cases, apps, SDK, etc. But this requires time, resources, and Jobs's skills to make it work. And what's interesting, many of the potential "revolutionary" use cases are definitely not for everyone, which would make this project less appealing for Apple. For example, I can imagine a digital/AI assistant recording/decoding everything around you and working as your second brain/memory device. I'm sure Stephen Wolfram will be one of the first users of such a device (see his own description of his everyday life [1]), but I'm not sure there will be millions of such users.

According to some leaks and hints, probably the second path is currently underway at Apple. But it needs time for both the hardware and software parts. Maybe even more time for the software.

[1] https://writings.stephenwolfram.com/2019/02/seeking-the-prod...

> nested/indented comments are confusing

Let me mention my recent app for an alternative view of the HN comments (Hackeray) [1][2]. I'm sure this is not for everyone, but for me this view really helped to read the comments more clearly and basically absorb any big discussion here.

[1] https://news.ycombinator.com/item?id=45571963

[2] https://hackeray.vercel.app

When dealing with patents, public interest, and their consequences, Bell Labs should be treated separately imo. My vague recollection of the book The Idea Factory [1] and a brief search indicate that AT&T was always treated as a special case due to its status as a regulated monopoly. This status at least culminated in the 1956 Consent Decree [2], which required making all prior patents royalty-free and (as I read elsewhere) mandated that all future patents be licensed on reasonable terms. Given Bell Labs' well-known portfolio-including the transistor, laser, CCD, DSP, and fiber-optic-related patents, this shows a significant exception to how other companies might have innovated and monetized their innovations.

[1] https://en.wikipedia.org/wiki/The_Idea_Factory

[2] https://en.wikipedia.org/wiki/Bell_System#1956_Consent_Decre...

> where a bunch of deaf children in a an environment without much adult interaction did manage to create their own sign language

You probably had Nicaraguan Sign Language [1] in mind. I think it’s a good example of the human brain’s ability to invent something and acquire knowledge easily. What I tried to show with my comment is that when human intelligence is discussed, it’s easy to refer to all instances of human achievements around us, but they are essentially accumulated cultural knowledge. Because of this, we tend to overestimate our intelligence, at least when comparing an individual human with an individual primate of another kind.

So, it’s also interesting why humans are probably unique in this ability to pass on and accumulate information, while other apes (and crows) limit this to skills like retrieving ants with sticks and breaking shells with stones.

[1] https://en.wikipedia.org/wiki/Nicaraguan_Sign_Language

It's an interesting question, but maybe more complex than it appears. We cannot get rid of the cultural passing of information (non-genetic) in humans. If we do, we get feral children [1] (also known as Mowgli in popular culture). But I doubt anyone would seriously want to compare this "pure" intelligence with that of other primates. There is a possible agreement that socialization is not an option for humans but a requirement. Maybe if by some bad luck feral children were grouped together for some time, this might be an approximation, but I'm not sure. Overall, using computers as an analogy, it's like humans are only functional after the software installation following the first power-on, which is more or less required for normal activity.

[1] https://en.wikipedia.org/wiki/Feral_child

You're probably talking about infamous Dark Matter Developers [1]. When the term was coined, I thought there were many of them, now seeing how many developers are here at HN (including myself) I doubt there are many left /s.

The quote that is interesting in the context of the fast-pacing LLM development is this

> The Dark Matter Developer will never read this blog post because they are getting work done using tech from ten years ago and that's totally OK

[1] https://www.hanselman.com/blog/dark-matter-developers-the-un...