While the article is mostly about cryptocurrency, my response is in re:
> But what do you do if you already live under tyranny? The rule of law is a great defense, but cryptography alone can't bring about the rule of law. What is the role of technology in this foundational struggle?
My colleague Moti Yung has studied this and there are some surprising results
> Anamorphic Encryption: Private Communication against a Dictator
...
> In this work, as a technical demonstration of the futility of the dictator’s demands, we invent the notion of “Anamorphic Encryption” which shows that even if the dictator gets the keys and the messages used in the system (before anything is sent) and no other system is allowed, there is a covert way within the context of well established public-key cryptosystems for an entity to immediately (with no latency) send piggybacked secure messages which are, in spite of the stringent dictator conditions, hidden from the dictator itself! We feel that this may be an important direct technical argument against the nature of governments’ attempts to police the use of strong cryptographic systems, and we hope to stimulate further works in this direction.
The hypothesis that companies today need to really focus so they can make money ignores the fact that, in fact, most or the major AI companies are not making money relative to their costs.
When was the last time a public pressure campaign caused a company to give up making money? And Google in particular? The company that bans entire GCP accounts on a roll of the dice?
Building a tool that tries (and probably fails) to remove the watermark (due to the arms race that large corporate machines will win) is tacitly accepting the barcode. The hacker ethos should be, first and foremost, to run open source models locally without relying on a corporation.
>due to the arms race that large corporate machines will win
Much like how the entirety of Hollywood, book publishers, academic publishers, and game developers have won against piracy despite being some of the largest corps on earth and dedicating untold billions to the issue over the past 30 years?
They didn’t win because of DRM. They won because of the regulations that grant a monopoly for a specific term in the form of a copyright. Society has recognized that incentivizing creative acts requires a temporary grant of monopoly to ensure the necessary scarcity to make money and recover the costs of creation. The real problem is Disney keeps expanding that time period so things never enter the public domain
This is again conflating at least two things and this is so prevalent in this context. Let us not conflate how annoying DRM:s are to us users that buy the things, with pirates thinking they somehow have a right to use any software without paying fairly for it. I would even go as far as to say that you pirates are the reason I have to have a DRM in the shit I bought and paid for.
> I would even go as far as to say that you pirates are the reason I have to have a DRM
I think this is largely an incorrect take. DRM is anti consumer, not anti piracy. In fact, it has done very little to deter actual piracy (and remember it only takes ONE person to break the DRM), while affecting some casual pirates and all legitimate users. In the process, they got rid of reselling stuff you own.
It's anticonsumer, not antipiracy, never forget that. It means something like this would have happened regardless of pirates.
They succesfully did away with 2nd hand markets and the concept of "owning" anything. So yes, I would imagine DRM would continue to exist without piracy.
I think so, because their main goal is to prevent unwanted use of the digital product -- to the detriment of end users -- in more ways than just piracy. In fact, they don't solve the piracy issue.
I am not sure how I am conflating two things, it would be helpful if you could expand or connect to my argument. Perhaps I am misunderstanding.
My argument is that the grant of monopoly is a regulatory decision and the real cause of "winning". No amount of DRM would confer the same benefit because the ability to bypass it through piracy would be totally legal with no economic or other consequences and so a robust cracking and distribution ecosystem would emerge. Thats a drastically different story than when napster gets shut down, and limewire gets shut down and pirate bay gets shut down every time it relaunches. Imagine a world where there is are 1000 pirate bays
Piracy is as easy now as it was pre-DRM. DRM is the digital equivalent to security screws on electronics, in that they’re a mechanism for lawyers to argue their client made an attempt despite being easily bypassed with a trivial amount of effort.
Exactly this. The real power is in the regulatory grant of a monopoly that comes with rights like the ability to sue for damages, issue take down notices etc. the DRM does allow restrictions on distribution because many people can’t be bothered to remove them, but more importantly the act of removing them is evidence of the intent to knowingly violate the copyright which might be harder to prove otherwise
And now that they're trying to push up the margins and the streaming ecosystem is fragmenting making everything into a series of bundles again, piracy is on the rise again.
They did win for a while because they stamped out 99% of piracy. In the early days of streaming it was legitimately difficult to argue for piracy. Streaming was just too convenient and too cheap.
But, they are greedy above all else. And so, we are once again seeing a resurgence of piracy. Large corporations seem to snatch defeat from the jaws of victory.
By the time you're building (or buying) the necessary highly esoteric and expensive ultracentrifuge setup I think you would be well outside the realm of "hobbyist" unless someone insists on the most unreasonably pedantic definition for the term.
Unless we're only considering final assembly. Just gotta get that weapons grade fissile material supplier lined up. That might or might not qualify as rich hobbyist territory depending on how high a price tag is permissible.
In theory, there's also direct laser-based isotope separation. It's a technology that is being actively suppressed, and that's one case where I very much in favor of that.
This subthread starts off with the argument that the big corps will never beat the little determined hackers, one of the founding myths of the early internet. And then every now and then a strong little branch of the argument runs up against an example and it becomes well sure, the little hobbyist hackers don't have anything there but that is because the big corps/gov/billionaires/whatever put so much into beating them.
I mean reading it all certainly sounds like the people on the little guy's side are overestimating the value of pluck, an observation Hollywood generally makes just before the heroes with pluck win for ever!
> And then every now and then a strong little branch of the argument runs up against an example and it becomes well sure, the little hobbyist hackers don't have anything there but that is because the big corps/gov/billionaires/whatever put so much into beating them.
It's almost never about the level of resources the organization puts in. The usual reason is that there isn't enough incentive to do it. What is a hobbyist going to do with a nuclear weapon? Why spend your time creating one if you, like the overwhelming majority of people, have no desire to blow up a city?
Preventing something that hardly anybody would be trying to do even if it wasn't being suppressed is a lot more practical than preventing something millions of people would do given the chance.
Yes. Winning against piracy doesn't mean you completely eliminate piracy. It means you scare enough people into not doing it and make it a bit harder to do for others.
Losing to piracy would see companies like Netflix and Spotify not thriving.
> It means you scare enough people into not doing it and make it a bit harder to do for others.
By which definition they utterly failed.
> Losing to piracy would see companies like Netflix and Spotify not thriving.
Not at all. Netflix and Spotify do well because they are a good value proposition for the average customer. Piracy is free at point of "purchase" but is (and always has been) expensive in terms of various sorts of overhead.
As long as enough people keep the pirate bays open, it will be there as an alternative when the services start their inevitable enshittification.
I for one do not enjoy the “Which service has the classic film I wanted to watch this week?” Nor having to switch services every time I want to see a new TV series.
We need (and have!) similar “free” alternatives to the watermarked generative services. Just like I hate the yellow dots on my printed images, I am not happy to have my creative assets (I do nothing nefarious) stained with SynthID.
> Winning against piracy doesn't mean you completely eliminate piracy.
But this is moving the goalposts. You can win against piracy either by making piracy less attractive or by making the paid offering more attractive. The first has utterly failed, piracy remains easy as a rule, and to the extent that they've succeeded it's not only disproportionately by doing the second thing, the DRM itself is a net negative because it has such a small effect on the ease of piracy while making the paid offering worse.
It doesn't make any sense at all. That's like saying browsing the internet with an ad blocker and other privacy tools is a tacit acceptance of tracking and ads, and that you should only visit websites that doesn't track or have ads.
> Use cases where the threat model fits: You are preserving art or historical record against false-positive "AI-generated" labels.
Sorry, how does using AI to generate images have anything to do with this? Image generators cannot insert watermarks into things they did not generate, and it seems highly unlikely that you will get a false-positive watermark on human-generated art, especially if, as the readme says, these watermarks have high enough fidelity to trace to a specific session id. Plus the modifications to the image needed to erase watermarks would necessarily change the thing being "preserved."
[edit]: the more I read the more I'm convinced, the claimed use cases in the README are bullshit and the real reason is to provide a tool that helps people bypass "AI-generated" labels on social media for AI slop.
I mostly agree about the justification in the repo being wrong, but wanted to engage about this point:
> Image generators cannot insert watermarks into things they did not generate
It's actually very easy to take a real image, ask Gemini/ChatGPT to modify some tiny part of it (could be something as silly as lighting/shadow/etc), and often the resulting image will be detected by their watermarking tools. This way you can easily present any real image as AI-generated.
Ignoring that a watermark removal tool does not help with this threat model, the claim is still true: the original image can not be changed, and instead a copy is created.
So what? I can also open an image in Photoshop and make sure it saves out some Photoshop specific EXIF data and try to claim the image was doctored. What I can't do is go and put my deceptive altered file up in place of the original in all the places on the Internet it exists.
I had to think about it, how about if the claim were:
If you take a photograph that is misidentified as AI generated, you can “preserve the historical record“ by using this tool before publishing the image.
(Anyone know the false positive rate with watermark IDs, would’ve hoped it’s like zero)
This article has no references to back up its claims, some of which seem like a stretch without further evidence (e.g., "The Cheshire Cat is a property without a carrier" being a critique of group theory). Are there references that back these claims up?
I believe it's because this is random AI garbage. I swear ChatGPT can be used to justify anything. The "jokes" are nonsensical and unfunny, even for early-19th-century Oxford professor inside jokes...
I'm surprised to see, on HN of all places, so many people taking the AI bait so often. Seems like more than half of blog posts and articles posted here are AI generated, but people skip reading the content and just go straight to the comments to discuss the title out of context.
> "Grin like a Cheshire cat" was a common phrase in Carroll's day. Its origin is not known. The two leading theories are: (1) A sign painter in Cheshire (the county, by the way, where Carroll was born) painted grinning lions on the signboards of inns in the area (see Notes and Queries, No. 130, April 24, 1852, page 402); (2) Cheshire cheeses were at one time molded in the shape of a grinning cat (see Notes and Queries, No. 55, Nov. 16, 1850, page 412). "This has a peculiar Carrollian appeal," writes Dr. Phyllis Greenaere in her psychoanalytic study of Carroll, "as it provokes the fantasy that the cheesy cat may eat the rat that would eat the cheese." The Cheshire Cat is not in the original manuscript, Alice's Adventures Under Ground.
It continues with a full page on the topic, none of which are anything to do with math jokes.
He did actually work with permutations and cycles in voting ( https://en.wikipedia.org/wiki/Dodgson's_method ) but it's combinatorial and not very group'y. Agree with comments that this is AI generated and that the highlighted stuff isn't necessarily interesting, deep, or correct.
As for the cat's smile, analytic philosophy substance/property stuff goes back to Leibniz if not Aristotle. Dodgeson basically predates much of Russel's career, but he could have been an influence on his idea of "bundles" and he definitely influenced Quine. He wrote a few textbooks if you want to dig into his research interests but IIRC it's more along the lines of Boole and DeMorgan, even if fictional fun is arguably anticipating the next wave. I linked the haddock's eyes elsewhere in thread.. good fun but also some rich implications. Since he's preoccupied with self-reference you could argue it anticipates Godel. https://en.wikipedia.org/wiki/Use%E2%80%93mention_distinctio...
A lot of these are very reminiscent of asking ai to find references to a subject. You end up with very tenuous links that take multiple steps to get there and i’m never convinced that the author really did intend that meaning.
Of course this could also be traditional literary analysis. It’s hard to tell.
Exactly, even having a father in the 1800s reading this to their kid and instantly getting that it's a base 4 joke is such a stretch that the article lost me right there. Even after knowing what it is, it's pretty obscure to spot.
> MIT recently became one of the first universities to get a grad student union
University of Illinois at Chicago (my alma mater) had a graduate student union in 2011, and I don't think a grad student union was so uncommon at the time...
The card check election was in 2004. I was part of the drive at the time (although I graduated before the union began—that was one of the challenges in the drive: a lot of us were essentially working for the benefit of those who came after us and wouldn’t be able to enjoy the fruits of our labors). It wasn’t the first, but it was still a relatively new thing at the time.
There's a reason we're not reading monospaced here, and a reason we do read monospaced code.
But the beauty of this moment is that if you want a really good SwiftUI monospaced Markdown reader, you can have it before dinner. This is exactly what I'm talking about. You have an idiosyncratic personal preference, and it's now reasonable to expect software to shrink-wrap around that preference.
Generally I just don't appreciate when someone jumps from "I care about this" to "everyone cares about this for obvious reasons." Focus on what something means to you, and being sincere about it. But that is just my advice for writing, take it or leave it.
Also, are browser text area inputs monospaced by default for everyone? Or did I configure that for myself long ago and forget? If it's not just me, maybe the "reasons" you're alluding to are not so obvious. Anyway, I have no trouble at all reading the long comments I type into text areas.
Legacy decisions as a remnant from a time when taking more space on paper cost pages and therefore resources, remaining as a default from centuries of inertia in how text is printed?
No, prose set in monospace is harder to read. The "legacy" is monospace! We went way out of our way to to get proportional typesetting working.
But seriously: you do you. There are people who code in proportional typefaces and they're as baffling to me as you are right now. Let a thousand Markdown viewers bloom.
> The "legacy" is monospace! We went way out of our way to to get proportional typesetting working.
The legacy is proportional, at least in Latin script and its ancestors. Handwriting was proportional, of course, and so was Gutenberg's printing press. Books and newspapers have virtually always been printed in proportional type.
In Chinese and Japanese, monospace is legacy in both handwriting and print... and also still universally used today. All Chinese and Japanese text is monospaced by default. Billions of people are getting by just fine reading monospaced prose.
I don't really know where this conception that monospaced is somehow objectively harder to read is coming from. Actually, this is the first I've ever heard of the complaint. I can't help but wonder if you've been subjected to some very bad monospaced fonts in prose or something.
Monospace text is objectively less dense, which means you have to move your eyes more. Every eye movement is an opportunity for error. Monospace text only makes sense when seeing exact character counts matters (which it often does in computer code).
One could argue that less density, as well as standardised widths, significantly reduces opportunity for error compared to cluttered text that is constantly varying how it is displayed. Perhaps moving your eyes more increases opportunity for error by 10% but easier-to-parse characters decreases the opportunity for error by 20%?
Turn off syntax highlighting for your code, translate it to COBOL, and pass it through a formatter that converts it to continuous word-wrapped text. Then we’ll talk again.
There is a real use case for a viewer if you have a lot of formulas. Yes you can read the raw latex but you go cross-eyed after a while. Maybe I am a softie though.
I agree, but I don't think the author of this blog post is coming from that perspective, and markdown renderers of the sort described in the post tend to do pretty poorly with math typesetting.
Nothing personal but I hate this take with a passion, and I literally think it's representative of the worst attitude in computing because it's the literal opposite of software SHOULD BE.
The whole entire point of computers in their best light is changeable software, the whole point should be "let people read how they want to."
reply