Hacker News new | comments | show | ask | jobs | submit login

This gets very meta very quickly.

If linking to copyrighted data 'should' be illegal (SOPA), then what about descriptions of that data that are sufficient to identify the original, but not reconstruct it (magnet links)?

And if those were made illegal, then what about descriptions of those descriptions? You can recurse infinitely on this.

Beyond mere amusement, after just one or two recursions, you get to the point where it would be difficult to write a law that would criminalize magnet links without also criminalizing people who link to a Sparknotes-like summary or commentary for a piece of media.

You're thinking about it the wrong way. From a programmer's POV, the link to the torrent can be abstracted endlessly into new and distinct forms, each of which you believe needs to be legislated away in turn. From a lawyer's point of view, the specifics are really not important, but rather the end result: is the user illegally procuring copyrighted material, or is the distributor providing them with a readily accessible means of doing so?

Law has certain resemblances to regular code, but folks here seem to think that if something isn't properly specified that the law will break in the same way that a program will fail to compile or run properly. But that's not how it works. Poorly drafted laws can fail, certainly, but it's not that hard to draft something that focuses on the end result.

Consider ordinary offences, such as robbery. You wouldn't get anywhere by arguing that you're alleged to have put your right hand in your pocket and pulled out a small hatchet, and that since there's no law specifically forbidding right-hand wielding of hatches, you should go free. The technicalities of how you committed the robbery are irrelevant as long as it can be established that you took someone's property in a violent fashion. I'm a little perplexed as to why folks think torrenting/piracy/filesharing etc. is so different that it can't be addressed legally. Sure, the law needs to be clear and logical, but only up to a point. It doesn't need to be absolutely exhaustive, and 'beyond a reasonable doubt' has never meant 'beyond any imaginable possibility'. People do make arguments like that in criminal defense cases from time to time, but they typically fail because the doubts they attempt to raise are absurdly far-fetched.

Totally true. As someone commented on my own story on this, you can kill someone in war and be a hero, or kill someone at home and get life in prison. Context and the definition provided by the law are very important. And as you say, the law doesn't have to be needlessly ignorant of the reality. It can easily encompass intent, result, context, etc.

But! There are still interesting problems to be acknowledged as far as how you define the transgression. A torrent like this from one perspective is worth billions, from another is worth very little. And practically speaking, if someone were to really use it the way you might expect a pirate to use it, it is practically worth only a tiny fraction of its potential worth. These are things that are difficult to define or restrict, but are immediately obvious to anyone who would use it. That's why it's interesting to me.

Sounds like we're on the same page. Legislative drafting is often overlooked in these things, I agree. A legal friend in the UK friend told me that way back when, the very best lawyers would get recruited to work at parliament writing and 'debugging' legislation, in order to maximize its legal reliability. Around the 1980s this came to be regarded as a waste of money and the budgets were slashed, and intead of the best lawyers legislative drafting fell to those who were unable to get decent jobs in the private sector...and the quality of legislation fell accordingly, while the cost of litigation soared. We're seeing a similar problem in American lawmaking nowadays, as best I can tell.

That's what this is already, eh?

A description of copyrighted data would be the torrent file, which content producers would probably like to argue are infringing.

A magnet link is a hash of the torrent file, so it's already two steps removed.

Of course, the Pirate Bay magnet dump is itself a torrent, so it's a hash of a hash of a hash of copyrighted data.

And that torrent itself has a magnet link: 938802790a385c49307f34cca4c30f80b03df59c is a hash of a hash of a hash of a hash of copyrighted data. (In the MP/RIAA's ideal world, I've just committed criminal copyright infringement with damages reaching into the $billions.)

Theoretically, the Pirate Bay dump could include the torrent for the Pirate Bay dump, and be an infinitely recursive description of itself... but that's probably an intractable cryptographic process.

And, continuing back from hexadecimal, we have the splendid:

in base 10 for the magnet link of the torrent containing the list of magnet links for the pirate bay's torrents. Shall we add to the list of illegal numbers?

Perhaps its prime factorization also belongs on that list.

Best to include the Roman numeral version of the number, as it's equally illegal.


It would also be wise to outlaw the URL I've just linked, and perhaps also the combination of letters "JROu8" as they also contain the information in question given proper context.


So, going back to hex and turning it into a flag just like the Free Speech flag in the wiki link, would that make this image:


the most copyright-infringing flag of all time?

What will they do if we find a crazy enough way to make, say, the number 5 infringing? And who owns 5, anyhow?

Even worse, they will make pointers illegal.

I can beat that. With a bit of indirection, the URL http://thepiratebay.se references all of the above.

> ... but that's probably an intractable cryptographic process.

Not always.

Algorithms that append a hash to a file, preserving the same hash, i.e.

  hash(s) == hash(s ++ hash(s))

A program that prints (though does not contain) a hash of itself:


(edited for formatting)

Ha. I bet one of those hashes is a hex representation of a copyrighted song lyrics

The more one studies computer science, the less one believes that information can be owned in any meaningful way. I remember reading a paper on a "lightnet" that basically XORed arbitrary blocks of data together, then produced a recipe on how to recover some original data by XORing the appropriate blocks. With just this recipe you could ask other nodes for the random blocks, then recover the original data locally. The paper made a good argument that however you try to define ownership of a block, it will lead to some contradiction. The system was implemented but I can't recall its name. I think it was hosted on SourceForge.

I'm sure most everyone on HN has read it already, but "What Colour are your bits?" http://ansuz.sooke.bc.ca/entry/23 is always a good reference in this sort of discussion. The system you mention sounds like what the Colour essay is railing against; trying to create a mathematical solution to a problem which is rooted in the law, and cannot be resolved by mathematical manipulations, but only by legal manipulations.

Do you recall the title or author of that paper? I'd be interested in reading about the contradiction argument you mention.

I think parent is talking about Monolith[1], which in fact that article specifically references:

    I think Colour is what the designers of Monolith are trying to challenge, 
    although I'm afraid I think their understanding of the issues is superficial
    on both the legal and computer-science sides. (...)
[1]: http://monolith.sourceforge.net/

I like that article as well, insofar as it helps both sides understand each other better, rather than continuously talking past each other. In order to do so effectively, that article pointedly avoids taking a position on the issue. However, the article also doesn't suggest that people shouldn't hold a position on that issue.

I personally feel comfortable taking the position that anyone claiming that bits have color has an objectively wrong view of reality. I don't think that position needs much advocacy, though; reality always tends to win in the long term. It could certainly use a little help sometimes, though.

The followup "Colour, social beings, and undecidability" http://ansuz.sooke.bc.ca/entry/24 is also pretty good.

I feel this is dealt with by the fact that law has a concept intention. Handing out the recipe for reconstructing the pirated material counts as aiding piracy, I would say.

There's definitely the undecidable concept of intention, but there's also an undecidable concept of identity: for example, when is an MP3 file "the same" as a copyrighted song? You can encode a copyrighted song into different bitrates, swap left/right channels, slightly pitch shift, apply equalization, etc. Most likely, given some algorithm that tests audio identity, you could always come up with a way to create a copy that sounds the same to a human but is missed by the computer.

Granted, the fingerprinting algorithm used by YouTube is pretty good: http://www.csh.rit.edu/~parallax/

>when is an MP3 file "the same" as a copyrighted song?


I have to say, I find the concept of intent a little difficult to grasp when it comes to this stuff.

As an, albeit somewhat contrived, example:

Let's say I have a music player on my computer that, when fed the works of Shakespeare, it plays the Gaga's latest hit and when fed with the works of Sir Arthur Conan Doyle, it plays Jingle Bells.

My actual intent is to read Shakespeare and to listen to Jingle Bells, but that is going to be a pretty difficult case to prove in a real court of law. The assumption will be made that I had the player and copy of Shakespeare so that I could listen to Gaga.

Theoretically, unless Shakespeare has been interpreted by said program, I have not committed any crime. But precedence disagrees. Just having the right sequence of bits on your computer is enough to prove intent, whether your actual intention was to use it to violate copyright or not. That is where I get lost.

Yes, in much the same way that any given number could be random, pseudorandom or nonrandom given its context, it can also be in violation or accordance with copyright law depending on its legal context.

Don't confuse a random number with the value returned by a function that returns nondeterministic values with a somewhat predictable pattern.

In case anyone is interested, the program to which I referred is "OFF System". Its home page is supposed to be http://offsystem.sourceforge.net/, but unfortunately it's down now and the project seems dead. The paper was on the site, but I can't find any link to it now.

EDIT: I think I found it at http://www.findthatfile.com/search-31270472-hPDF/download-do... . The name of the file is "CopyNumbCJ.pdf".

> This gets very meta very quickly.

You're thinking legality, but I'm thinking efficiency. So we can now distribute 1.5 million torrents (of a total of several thousand TBs, no doubt) in a file 90MB big, in a torrent which itself has a magnet address that takes up...20 bytes?

The size savings as you go up the tree are incredible. I see no reason why you couldn't create an almost-entirely distributed torrent site in this way.

Think about: Torrent discovery could be done by regular distribution of index torrents, and the clients use that to find out what can be downloaded and where.

In fact, in the world of magnet addresses, "uploading" a torrent would be as simple as requesting that its URI be put in the day's index. So running a torrent site would be as simple as curating a list of magnet URIs each day into an index, then publishing that torrent's URI somewhere. Like Twitter. You could run a torrent site entirely from Twitter.

edonkey has had distributed search for a long time. It's possible to maintain an keyword index of magnet links in a dht, and then you remove the need for the torrent site completely.

I think that requiring torrent files and trackers was a policy decision to deflect liability away from the client implementer to multiple third parties. That's why bit torrent is still around and Grokster isn't. There's no technical need for them.

BitTorrent has torrent files and trackers because it was designed for non-infringing use-cases (e.g. distributing Knoppix images). The centralized elements substantially improve reliability (and, at the time, performance) in those use-cases.

Java should be illegal. You can write code that downloads torrents with it.

Your comment should be illegal; it gives advice on how to go about this process.

Your comment should be illegal; it clarifies that the previous comment was advice and therefore assists in illegal action.

Someone could suggest a system, whereby spelling mistakes are used to encode partial information about a magnet link.

One switched pair could give you the position in the magnet-link key, the other switched pair could give you the value. That way, you could never pin down exactly who gave you what information.

Or maybe I shouldn't suggest it?

That could have interesting unintended consequences - ISPs being pressured to ban users who post blog comments with spelling errors.

I approve of this

Lawmakers can easily avoid this meta situation by writing simpler more encompassing laws.So rather than being specific about how the pirated material is accessed they can write a more open law along the lines of "a site that's main use is assisting the distribution of pirated material".

SOPA/PIPA tried to do that.

> then what about descriptions of that data that are sufficient to identify the original, but not reconstruct it

Such as the name of the copyrighted work?

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact