>The author put zero rational thought into this. If Anthropic is legally required to release the weights of any publicly accessible model, their logical response will just be to stop offering public models.
I mean... that's the entire point of their argument, yes.
>Anthropic gets most of its revenue from enterprise agreements anyway. Providing public models or providing public access to their models is their secondary business.
Based on her comments in the paper it sounds like she was aware that an AI result was coming and rushed to release her work beforehand. 240 was not a tight bound from her methods.
>QQ for all the pro players out there, for 6 or 9-ring poker, can solvers really solve every scenario - or there is simply too many permutations.
Solvers can only solve using restricted game trees. Generally only heads-up pots, for one--multiway solvers exist but they are cumbersome and there's no guarantee that their algorithms converge to any kind of optimal play. And within the heads-up game tree, you still need to restrict the size of the game tree with a limited number of bet sizes. Not really a big deal, but in theory a perfect poker player would have a continuous range of bet sizes available for them to use (eg 1bb, 1.5bb, 2bb, 2.5bb, etc), not just a few discrete ones.
>The other question I have is how do human players adapt to GTO play now? Like in other games, humans have adapted to novel strategies - I'm curious for poker whether there's ways to exploit GTO solvers (ie, if you put your opponent bot on the GTO range and chase the fat tails).
You can't really exploit GTO solvers. But a human player can't play a GTO strategy, it's simply impossible. People can get reasonably good at playing a few of the most common nodes in a low-exploitability way, but beyond that, nobody is placing a perfectly balanced strategy. It's just impossible. The strategies are way too complicated.
>And is the poker community's fair play detection algorithm good enough to catch players who use solvers like chess community - or it's easy to evade by making a few deviations.
The strongest fair play detector right now, at least at high stakes, is the players themselves. Precisely because playing a GTO strategy is impossible, if someone starts playing too perfectly, other regs will notice. And the high stakes community is pretty small too so everyone knows everyone else, or at least knows who knows them or whatever. Those two things combined mean that there's a lot of self-policing.
Also, playing a GTO strategy isn't necessarily always desirable. Equilibrium strategies are fundamentally defensive. A solver will defend at the perfect frequency against a river bet: folding just enough to make you indifferent when bluffing, so that "always bluffing" and "never bluffing" have the same EV (this is technically oversimplified but whatever). But this approach doesn't make the maximum amount of money. To win the maximum, you want to just always fold if your opponent isn't bluffing enough and always call if they're bluffing too much. Of course this opens the door for you to be counter-exploited (if you always fold, they can start bluffing all the time, and vice versa) which is why the solver lands on the goldilocks equilibrium.
In other words, a GTO strategy does not punish frequency mistakes, and Hold 'em is a game of frequencies. Of course a GTO strategy will still win money, because people don't just make frequency mistakes, they also make just straight EV mistakes. So there is certainly some risk of being cheated, sure.
Another thing that disincentivizes cheating a bit is that you have to be extremely consistent with it. In chess, using an engine in just one or two critical spots could be enough to swing a game completely. In poker, because of the imperfect information & randomness, the wins and losses are distributed far more evenly across time. It's not enough to just check a solver a couple times a session if you're bleeding money in the other hands.
Anyway, all of these are real issues, but the bigger issues facing online poker are all just related to game availability and quality: poker sites charging way too much rake, fewer recreational players willing to lose a lot of money online over time, a gazillion legal/regulatory issues that make it hard for people (both professionals and recreational players) to even play online in many places in the world.
>I spend quite a lot of time reading and writing on Reddit, and while I encounter plenty of low-quality posts and comments, I'd say that essentially none of them are written by bots.
Err... I spend quite a lot of time on Reddit as well (including moderating a medium size subreddit) and I can assure you that there are tons of bot posts and comments all over the site right now. If you think this, you're probably not very good at identifying this kind of content.
I certainly do not think that the average HN reader "could probably understand this post in a matter of days or weeks". I think the level of background information you need is something like an undergraduate degree in mathematics and at many universities that's probably not enough either.
There's a difference between understanding something and being able to do something. For example you can get a broad understanding of what calculus does in a matter of minutes: a derivative is just the rate of change of a curve, and an integral is the total area under a curve - you're practically there! Of course it'd take months of work to understand how to apply it yourself, let alone all the tricks involved, but a basic understanding of what stuff is, is very easy to obtain.
And I think that generalizes to math. Its normal presentation is completely impenetrable due to a large use of symbols and terms with no outside meaning (or, even worse, a meaning that contradicts the colloquial usage). But I think if we somehow resolved that issue, the average person would be fully capable of following along even if they're going to have to take everything at face value as opposed to being able to meaningfully understand the exact methods and tricks being used to get from A to B.
I basically completely disagree with your analysis. I am already taking into account the "understanding vs doing" distinction in my assessment. I think it takes ~an undergraduate math education to have the background to work through this post, but there's no way that most people with that background would be able to produce it.
If, like the vast majority of people, you have never studied math past a college calculus sequence and perhaps a (practically-oriented) linear algebra class, then you are missing some pretty fundamental ways of viewing and thinking about math. The basic mechanisms of algebra, for one: spaces and operations, morphisms, products and quotients... You don't know what a group is, never mind the idea of a group action. You don't have any familiarity with some basics of geometry: the Riemann sphere, Mobius transformations... And you certainly don't know anything about algebraic geometry: what an affine variety is, what birational equivalence is, what a fiber is.
All of these are concepts required to understand this blog post. And it's not just a matter of understanding the definitions of the terms, but having at least some intuition of what they really mean and how they work. Most people cannot go from zero to understanding all of this in a few days or a couple weeks, at least not for any reasonable definition of "understanding". There's too much basic mathematical background that's missing.
To be clear, this is not the kind of thing where a Lean formalization provides any value at all. It's like formalizing the answer to a high school algebra problem. The counterexample is obviously correct.
Indeed. I was mainly responding to the comment about waiting for "independent seasoned mathematicians to verify", whereas in this case it is easy enough to convince oneself of the counterexample's correctness.
They are certainly much less accessible than they used to be. Not only has torrenting lost a lot of mindshare and general cultural awareness from 15-20 years ago, but the sites are generally harder to get into. If you're starting from scratch, it takes several years of waiting and uploading hundreds of torrents in order to get access to the best TV and film trackers. People are stingier than ever with inviting others too.
>It's important to remember that to this day, streaming sites do not have a full archive of the music out there. There is still a need for music piracy
Ehhh..... I'd wager that pretty much anything that most people want to listen to is on music streaming sites. Streaming is how everyone consumes music these days, so everything new gets released there, and by this point the catalog from the CD era is extensive. Music streaming has more music than What or Oink ever did. Streaming also has huge value add over piracy: it's really easy and convenient, it's better socially (shared playlists), and recommendations/discovery are waaaay better.
The vast majority of people do not "need" music piracy any more. If you want ten different versions of every REM album with slightly different mastering then sure, join RED. But it's a niche interest these days.
It's a huge contrast to movie piracy, which is thriving and which provides enormous advantages over any other way of watching movies at home, not just in cost and convenience but also in access and in quality.
> I'd wager that pretty much anything that most people want to listen to is on music streaming sites
If you have simple tastes, easily accept holes in their catalog and don't care about being served butchered "remasters". People who actually care don't use Netflix/Spotify.
>If you have simple tastes, easily accept holes in their catalog and don't care about being served butchered "remasters". People who actually care don't use Netflix/Spotify.
Oh please, spare me the condescending bullshit.
Sure, there exists music that is on RED that is not on Spotify. There also exists music that is on Spotify that is not on RED (some of which I even listen to!).
I said "pretty much anything" and "most people". I stand by this. Most people do not experience major holes in the Spotify catalog and are perfectly well satisfied by the breadth of the catalog. If you aren't, that's cool, but you're in a minority.
If this weren't the case, music piracy would be more popular. It's not. RED has more music now than What.CD did, but the community is smaller. It's telling that it doesn't even get a mention in the OP. A lot of people who join aren't even particularly interested in music piracy but just want to use it as a stepping stone to other communities.
I'm not saying that music piracy sucks or whatever. I'm just saying that most people don't feel much need for it and are well-served by Spotify--which, again, has some huge advantages over piracy that I gave previously. I think it is useful to be realistic about this because it's easy reading an article or thread like this to feel a kind of FOMO and I think it's valuable to push back against that.
I think you and OP are arguing different things. Netflix/Spotify are the McDonalds/Olive Garden of media. They are basic, uncomplicated, and flawed, yet they serve most people's tastes. People whose needs they don't serve well go somewhere else.
I'm not sure why people keep bringing up Netflix. Last I checked, they weren't a music streaming service.
There's a huge difference in Spotify for music vs Netflix for TV and movies. Netflix's catalog is absolutely tiny. Spotify's is not.
I also continue to disagree with the idea that Spotify's catalog is only for people with basic or mainstream tastes. This is far too reductive. Like I said above, the vast majority of newly released music is released onto Spotify. For many new or small artists, streaming is the only way their music is released. There is no "somewhere else" to go.
Personally, I listen to a lot of contemporary French indie pop, including a lot of pretty obscure stuff. This stuff is most available through streaming services. To the extent it is available via pirate channels, it's because someone bothered to take the time to rip it from a streaming service. There is nothing which music piracy offers in this domain, and there is a lot that it lacks (both in availability and in all the service factors that make streaming appealing, as I've said--convenience, sharing, discoverability). A further factor I haven't brought up yet is that it just feels shitty and pointless to pirate this kind of stuff when listening to it on Spotify at least provides some benefits to the artists themselves: royalties (tiny as they may be), increased play counts, exposure via recommendation algorithms.
A large fraction of what is uploaded to music piracy sites currently is sourced from music streaming platforms. Not all of it, certainly. A large fraction of the CD- and vinyl-sourced uploads are also available on music streaming platforms too though. As I've mentioned a few times previously, Spotify has a bigger catalog than RED does (even if you ignore all the 0 play AI slop spam that's been flooding Spotify lately).
I would say that the median user of private music torrent sites now is one of two things: either they are there because they want to join other communities for things beside music (TV, movies, video games, anime, porn, whatever), or they are someone who has been pirating music for a long time, probably Gen X, listens to mainly music from 30+ years ago, and has made music piracy a part of their identity and self-perception on some level (you can see a certain sense of smug superiority about it even in this HN thread).
All of this is to say that I think that the comparison to McDonalds or Olive Garden is just way, way off base. Spotify and Netflix do not belong in the same conversation either. Is there some stuff that's not available on music streaming platforms? Sure, it exists. If people find that streaming is not meeting their needs, of course they should look elsewhere. But the picture that is being painted in this thread is simply far off from reality.
While I must sadly agree with your usage of "most people", I still don't understand why you thought it changed anything to the original claim that there is still a need for music piracy (for the aforementioned people who care). And note that I didn't pick particularly obscure artists, we're talking 10~100k ratings/followers on RYM.
If you take issue with the claim that everyone streams music these days than the only way I can understand your comment is by assuming we live in vastly different cultures.
Do you think the difference between film piracy and music piracy is inherent, due to the differences between film and music; or is there some alternative reality where we ended up with a one-stop shop for films, as well?
For the history of music piracy, I found" How Music Got Free: The End of an Industry, the Turn of the Century, and the Patient Zero of Piracy" was a good book to read.
I would wager the effective piracy rate of stuff that on prime and Netflix a few years back was close to the effective music piracy rate. IMO the difference is that with Spotify, tidal, Apple, YouTube or Qobuz - you mostly get access to everything. With film, you can pay for Netflix, Disney, Hulu, peacock, HBO, and _still_ not be able to get access to major releases without paying more on top of the subs.
That's an interesting question. I'm not sure. We sort of had that one-stop shop experience with Netflix's DVD service, where you would pay a subscription fee and in exchange you would get to watch movies from a huge catalog. But this didn't translate to the streaming era.
P2P film piracy, at least for the quality-minded, has a few strong competitive advantages over film streaming. It doesn't have to deal with rights issues, for one, which can present huge roadblocks to film distribution. Films are also huge files and the interests of a streaming platform (low bitrate) are in tension with interests of quality. Even in comparison to physical media--the highest quality release of a film might be from a different market than yours, or there might be many competing releases over time. There might be different factors that are better in one release and other factors better in another release, where the pirated copy can combine all the best parts. It's actually somewhat remarkable how good film piracy has gotten these days for those who care.
I don't think quality is really much of a concern for the majority of people, only enthusiasts. I suspect, analagous to 128 kbps opus (on youtube music), most people can't really tell the difference between a 1080p bitstarved stream and a 4k bluray rip.
The library for the music streaming platforms is much bigger than for films, of course (about 250 million for Spotify), but there's also a much lower barrier of entry. So perhaps the higher work needed to produce a film necessitates more profit, to a degree that only the fragmented streaming platforms can fulfill. After all, netflix started making originals to counter studios launching their own streaming platforms to raise profit margins, and pulling their content off netflix.
> I don't think quality is really much of a concern for the majority of people, only enthusiasts. I suspect, analagous to 128 kbps opus (on youtube music), most people can't really tell the difference between a 1080p bitstarved stream and a 4k bluray rip.
They might not care but anyone who literally can't tell the difference of a good 4K source and a 1080p source on an appropriate display needs to go see an eye doctor. But that most people don't care about quality also isn't particularly shocking.
And as gp indicated, quality isn't just about the encode resolution and bitrate but also about the master itself. Unfortunately not all directors and companies behind great movies have the resepect for their creations that it deserves and the current release which might be the only release on streaming platforms might have significant flaws such as unwanted cuts/restorations, missing audio tracks, replaced sound effects, inaudible mixes, missing subtitles, bad upscales, excessive denoising, reframing from the original aspect ratio that cuts of content and/or shows parts of the originally captured film frames that were never meant to be seen, or various other "enhancements".
But physical DVD rentals is even more restrictive than a streaming library. You'd literally hit a physical limit on the number of patrons at any one time.
Music being generally 3-10 minutes long while film is 1h30-3h makes a big difference here. A film is a bit more of a commitment than a playlist entry; you can just put music on the virtual sushi belt and grab what comes past, while sitting down for a film is more of a time commitment.
There are fewer music rights holders, so it is easier to get them together in a room and agree to, for example, a piece of Spotify in exchange for licensing the music. Thus, Spotify becomes like a defacto standard with reasonably all popular music. Just one subscription for what the average listener will want.
Right now, there are too many film distributors and services, let alone TV, plus a lot of exclusives that people want to watch. These video streaming services seem to be trending towards consolidation, but I think film distributors remain diverse.
This is very far from truth unless you only listen freshs pop music. And even then it is easy to click a song only to receive "not available in your region".
Yes, same can be said for Vkontakte, for example (thanks to its piracu police and user uploads)
But this is kind of a different story.
Likewise at somepoit Russia passed a law that made all soviet movies and cartoons public and obviously studios started using YT as their hostding platform.
But this is not a default case. Just a lucky coincidence.
TPB were actively advocating and facilitating the upload of owned content. YT reacts to takedown requests and even re-directs advertising revenue to the rights holders.
You know TPB is a company right? They even run a paid VPN service. And half of their business was "bulletproof hosting".
Plenty of pirate material on yt, and I couldn't care less about what organizational format tpb have. I'll support them against Google every day of the week, since they are not hellbent on creating a socialist surveillance society.
First of all, vinyl is still relatively niche in absolute terms. Second of all, the popularity of vinyl, such as it is, has absolutely nothing to do with availability. It's largely driven by a kind of retro nostalgia (as the technology itself is, of course, inferior from a technical sense of faithful reproduction) plus a desire for personal physical ownership of something.
I was a zealous collector of records when we made the switch to digital audio. And let me tell you, there was a significant artistry to the thing: a vinyl record was a standardized work of art. Every album cover was the result of careful design, photography, layout. The inner sleeve oftentimes contained more art, or if we were lucky, all the lyrics we wanted to sing along with. The record label itself, a masterpiece of design. All the smells and all the feels of merely collecting vinyl--even without playing it--are indescribable today.
When CDs came into the market, they were horribly clacky and just clad in layers of tacky plastic. The album art was shrunken, misshapen... and the objects themselves stank of polycarbonate, rather than delicious vinyl. Sure, they sounded great and they lasted a long time, and maintenance was dead simple. But so much artistry was lost. I was still collecting lots of CDs when purely digital distribution hit us, but by then, the smells and feels and experience of collecting vinyl were distant memories.
And that entire experience may be why people argue for the technical superiority of vinyl recordings, and analog tube amplifiers. Because it was all self-reinforcing, and it all fell apart once the clacky, tacky, plasticky CDs took over.
Yeah, that's all part of what I meant by "a kind of retro nostalgia". People enjoy the experience of buying vinyl and of putting on vinyl records. I don't think there's anything wrong with this, for what it's worth. I'm just claiming that this is what's driving vinyl's popularity right now, rather than because people are turning to it due to music availability issues (which is much more of a factor for DVDs/Blu-rays right now).
There's a shitload of valuable vinyl records that don't have any art at all or even any cover. Paper and cardboard doesn't last as long as the record itself.
Look, I am telling you about my own lived experience with collecting vinyl. You can speak for yourself, but I carefully stored all my items in archival sleeves, and the jacket, art, and inner sleeve were often just as important as the disc and the music encoded on it.
There was a real thrill and reward that came from collecting LP albums in particular, and that meant 12" discs, and I also had a particular specialty in finding 12" remixes and DJ versions of singles.
Yes, there were shaped discs, and colored vinyl, and white-labels and acetates that came with no art or plain sleeves, and I collected those with just as much alacrity, but it really was a pleasure to flip through my collection, or someone else's, and drink in that large-format album art.
Interestingly the higher-end physical movie releases have moved to metal cases and/or paper sleeves to hide the plastic. Perhaps that's also what digital music would have moved to if the physical market didn't get eaten by streaming. The artwork would still be smaller than vinyl but if you have broad tastes and limited space that may actually be a feature.
You are correct that you can't lossily compress a vinyl, but you absolutely can master a vinyl for maximum loudness. It's harder to do, and some of the techniques are different, but it can be done, people did it, and this was actually where the loudness war started. And notably, unlike digital, vinyl does not impose a maximum loudness wall. If you find a way to make the groove wobble more than before, the player will absolutely produce a louder signal than before, whereas with digital you have a strict limit to your levels at +/-32768.
To make matters worse, people aren't doing separate masters for audiophile formats anymore, so vinyl is getting the ultra-compressed, low-dynamic-range master anyway. That is because the vast majority of people buying vinyl were doing so as merch, not so much as a way to buy better-mastered albums.
Give me a break. CDs aren't compressed and sound flawless if the mixing is good. it also isn't dependent on things like the quality or age of the disc. And sure, a bad player might sound a bit worse than a high quality one, but that mostly comes down to the DAC in it, whereas with vinyl depends on the player a lot more by virtue of being a mechanical format
This is complete bullshit, sorry. Digital is more capable of faithfully reproducing sound than vinyl is. Vinyl does not have technical advantages when it comes to faithful sound reproduction. I mean, it literally degrades over time, for chrissake!
I don't know what kind of "compression issues" you're talking about but I strongly suspect you'd be well served by learning about the Nyquist–Shannon sampling theorem.
> Digital is more capable of faithfully reproducing sound than vinyl is.
That's exactly the problem that makes digital unsuitable.
Theoretically digital can reproduce sound faithfully, but if the medium allows sound engineers to compress the hell out of music, then they will abuse the opportunity.
Vinyl is a very limited format and you can't really do any sort of "creative" audio optimization bullshit with it.
Compression in the audio waveform sense has nothing at all to do with the medium. You can absolutely press the same compressed masters on vinyl or release the uncompressed masters digitally. It's a matter of what the customer base is looking for. Digital music is made for the mass market where "louder" means standing out and thus more success while contemporary vinyl is made for the enthusiast who can afford high-end equipment and clean listening environments.
here's a counter example: the opening to mirror's edge is not hosted on spotify (last time I checked, in my region at least), and it's an old favourite of mine (and I'm sure many people who played the game too)
It still makes absolutely no sense. First of all, this is not currently a bomb threat up until someone actually makes a threat. Second of all, in the event that somebody does make a threat, the existence of a Bluetooth device named "Bomb" doesn't make the threat any more credible or serious.
It doesn’t have to be an intentional threat to be worth responding to. One might reasonably think they’d stumbled on an (admittedly poorly executed) attack.
I mean... that's the entire point of their argument, yes.
>Anthropic gets most of its revenue from enterprise agreements anyway. Providing public models or providing public access to their models is their secondary business.
Enterprise agreements are public access.
reply