Slashdot Mirror
Reputation System Fights P2P Junk
yeejiun writes "Many of the files that are shared on p2p networks tend to be junk. Organizations such as the RIAA and music labels regularly pollute these networks with nonsense files masquerading as real music/video files. These junk files make it difficult for users to find what they want on such p2p networks. Some researchers at Cornell University have developed a reputation system called Credence, that works on the Gnutella network, allowing users to tell the good files from the bad ones."
quit downloading crap off of kazaa/grokster/morpheous/etc. dont trust brittneyspearsporno.avi.mpeg.exe
lameness filter thwarted.
Gotta love the torrents!
argumentum ad fallacium: Fallacy of defining a fallacy which allows one to dismiss the argument in question.
I thought the primary purpose of P2P filesharing was to share legally swappable media files as well as other files like documents and useful freeware applications. Is there some nefarious entity flooding the P2P networks with garbage disguised as those files above? Why would you need to know the quality of the file's reputation?
Jesus saved me from my past. He can save you as well.
if the RIAA is willing to create junk files, you really don't think they are going to create fake accounts to rate their junk files as "good"? ANY system you put in place that gathers "votes" from users can be manipulated.
Doesn't the eDonkey2000 network already have a system like this? Users identify fakes and report them, then the phony file information propagates throughout the network and the fake file dies.
Don't you mean the real illegal files from the fake illegal files? Seriously, it is no surprise to me why P2P has gotten a bad rap. Many of the users simply use P2P apps to commit piracy.
Yes, there are legit uses as well. But honestly, if you are looking for free music from a band that has released it as such, you can usually find it. It's the copyrighted commercial music and video that have tons of fake files, porn movies, etc...Not Jim Blow Sings the Blues, Live from Natrona, PA!
How is this any better than Bitzi and its Bitprints, which are already built into popular Gnutella servents like BearShare?
"Our client provides a peer-based judgement that a given object will possess the properties with which it is labeled and enables users to evaluate search results for authenticity before downloading."
Sounds exactly like Bitzi to me...
"Many peer-to-peer reputation schemes have been proposed in academia. Credence is the first practical implementation of a peer-to-peer reputation scheme."
I don't think so.
"BSD: Free as in speech. Linux: Free as in beer. Windows 10: Free as in herpes." --Man On Pink Corner in #52607549.
...which only verifies file integrity. It doesn't check if the file is what its filename says it is. It only ensures correct data transfers, not correct data.
If a file appears to by RIAA-affiliated music, treat it as a junk file.
Why bother with music the artist doesn't want you to have? Just forget about it altogether and discover new music, even new types of music that you'd never realize existed, much less that you could enjoy.
OVERVIEW
.
Credence is a robust and decentralized system for evaluating the reputation of files in a peer-to-peer filesharing system. Our goal is to enable peers to confidently gauge file authenticity, the degree to which a file's contents matches its advertised description.
At the most basic level, Credence employs a simple, network-wide voting scheme where users can contribute positive and negative evaluations of files. On top of this, a client uses statistical tests to weight the importance of votes from their peers. And finally, Credence allows clients to extend the horizon of information by selectively sharing information with their peers.
Authenticity and Pollution
We define pollution broadly as any file with content that does not match its description. An authentic file, by contrast, has content that is accurately described by its metadata. We find in practice that pollution in current networks can be easily identified by users without any special knowledge or expertise. As pollution becomes more sophisticated, more advanced detection techniques will need to be developed to help users safely identify malicious content.
Voting
The Credence system relies on individual users as the first line of defense against pollution. After a user downloads and uses a file, she is given a chance to submit a single vote to the Credence system: a positive (thumbs-up) vote for authentic files, and a negative (thumbs-down) vote for a polluted file. Each vote is cryptographically signed and entered into the system.
Vote Gathering
Credence uses these votes collected in the network to determine the authenticity of content. Credence displays a rating for each file that appears in response to a user query.
First, the client software executes a search for votes, and downloads a number of votes randomly selected from the network. These votes are then aggregated into a single estimate of the authenticity of the file in question.
Each vote collected from the network is not used directly, however, since some peers in the network may accidentally vote incorrectly, or even lie intentionally about the file's authenticity. Therefore we assign to each peer a correlation coefficient, or weight, reflecting the historical usefulness of the peer's votes. In effect, this helps remove the incentive for an attacker to lie about the authenticity of files. A consistent liar is, after all, just as useful as an honest peer when it comes to distinguishing authentic files and pollutions. And an inconsistent voter will come to be be ignored by others in the network.
Information Sharing and Transitive Correlation
Peer-to-peer networks can grow quite large, and many clients might participate rarely, sharing and voting on only a few files. This means that alone, a client may have trouble quickly discovering peer correlations and other historical data. To alleviate this problem, Credence uses a technique called transitive correlation to quickly spread information among small groups of peers and help clients expand their horizon
In Credence, a client periodically requests historical data from selected peers in the network. This data contains information on how the peer voted in the past (cryptographically signed, as before), and information about how the peer is related to other peers in the network. The client can then validate this information for authenticity, then integrate it into its local databases. In this way, not only does the client take advantage of the work other peers do in evaluating files for authenticity, but also gains insight into the behavior of peers in the network. All this is done without need for user interaction, or any peer trust values, which can be difficult for a user to accurately determine.
Changes to the LimeWire Client and Gnutella Network
Credence is integrated into the LimeWire client, and works on top of the Gnutella network. The implementation is built entirely on top of existing primitives in the Gnutella protocol. It opens up no additional ports
Shocking.
I don't know that their tactics are effective - after all, networks like eDonkey|eMule seem to be pretty good at self-policing. But it's amusing to see the undercurrent of outrage in these 'stories'.
We all know damn well why the *AA folks do what they do.
Web2.0: I love when people Flickr my cuil and digg my boingboing until my google is reddit and I start to yahoo
For those of you that can't be bothered to RTFA, this system takes a profile of how you vote on files and matches you with other people who voted similarly. Thus, the spammers would see different ratings than 'normal users.'
Illegal? Samir, This is America.
It is already very east to tell the junk files from the good ones. The junk ones will come from a very limited IP range. What usually happens is that the *AAs, and the companies they hire to pollute the networks will use the entire IP range they own to do that, but that usually still only amounts to a few class Bs. The good files on the other hand will come from all different class As.
The fact that I didnt get to play HL2 was compensated by the 2 hours of dwarf porn.
-knowles
I like this idea. Media hordes, read as RIAA and MPAA, will constantly try to find technical ways to put the P2P genie back in the bottle.
/. mobs will just mock them.
For every Napster (Kazaa, etc.) they close, another will be spawned. For every fake or intrusive system they create to battle downloaders, another downloading method will be innovated. For every commercial they feature a celebrity crying copyright heresy,
It's no shattering concept there'll never be a checkmate for either side.
Some aim to please, I aim to tease.
Like what Slash does.
I think the main insight and contribution of the system is that the reputation of a peer according to you is determined by whether he/she votes in a similar manner as you.
So if the RIAA starts spamming Gnutella with lots of junk stuff, you will never vote in the same way as the RIAA dummy accounts, and you don't take their votes into account.
In fact, it seems the system is even smarter than that - it can take votes from people that are strongly uncorrelated with you and use that as negative information. So anything these people vote as valid files, you can treat as garbage as their definition of good/bad files is completely opposite to yours. And assuming you trust your own judgement, that means those files must be bogus.
Reminds me a lot of the google pagerank system, but with explicit learning/training instead of using back-links for determining correlation.
Seems the trust system is prone to spamming itself. If the RIAA (or anyone for that matter) flood the system with bogus votes, then the "honest" votes will get ruled out.
I haven't read the description closely, but it's hard to see why flooding the system would matter- it isn't majority rule, it's who do you trust and who do they trust. If the RIAA has ten million bogus users, I and a few hundred other people vote thumbs down on them but thumbs up for each other, then we have our little corner where a set of honest opinions exist (although it may take a while to initially connect to that group).
The way you would have to spam the system would be to vote honestly for a time and then switch abruptly, but even then the damage would be quickly mitigated as your credibility disappears.
You misunderstand what you quoted... if they flood the system with votes, it matches them with the type of vote they make, when you use it you are matched with the type of vote you make... Thus, if you mod real files up then another user who mods real files up will trust your mods more than somebody else. If the evilpeople mod real files down and bad files up, then they will trust the mods of other evil people, but they won't trust your ratings, and you won't trust them.
Thus if you wanted to have a really easy way to find a list of crap files, you just have to mod down every real file you have, and mod up every piece of crap you have, then do a search. Your results will be clustered by the trust that the file you are getting is "like yours" or in that case, a fake.
Gravity Sucks
True... But a bogus torrent usually doesn't survive too long and certainly doesn't see too many seeders. If it's been up for a day or two you can be reasonably sure it's valid.
Also, even the "pirate" torrent sites are centralized and often even have administrators, sometimes even comment boards. If a torrent is bogus, someone will take it down. (Not that I've been to those sites, of course...)
Of course this could all be manipulated, but AFAIK it hasn't been yet by the powers-that-be... And I don't see why they'd bother, when a threatening letter is all it usually takes to take a torrent site down, and it would take considerably more effort than turning a bunch of scratchy mp3's loose on kazaa.
So a 'good' rank is dependent on the whims of people who usually vote the same way that you do. So spammers will see high rated spam and non-spammer will see high rated non-spam. Simple.
The research and motivation for this is important. If peer to peer networks can be subverted, then they have lost their usefulness. IMO, the sharing of copyrighted data is unavoidable, and sacrificing the freedom of a protocol in an attempt to prevent it is shortsighted.
It probably would have been better for Cornell if it had been left as a paper, rather than implementing it.
"A language that doesn't affect the way you think about programming, is not worth knowing" - Alan Perlis
I'm thinking this is a troll but what they hey, I'll bite. The problem with the P2P flooding is not that people don't know how to use it. I've used P2P for quite awhile now and would consider myself somewhat "skilled" at searching for things I need. The junk files are the exact same size as a normal file would be. A song for example, will show that it's a 192KB song, lasts 4:30, and is shared by 40 people. So you download it and start to listen. The first 30 seconds is perfect audio, then it's followed by 4 minutes of either dead silence or some sort of annoying tone. I can almost guarantee that there's more junk out there than real files. And as a side note, you may want to know what you're talking about before you start calling people dumb.
My sig of choice is Marlboro
Many hardcore file shares and hosters, dare I say most that would call themselves hardcore, are not in it for getting free content on demand when they want it. They are into collecting absolutely anything and everything they can get their hands on. In some collections, people wouldn't possibly, in their lifetimes,be able to listen to all the music or watch all those movies. But just the thought of having it makes many hoarders happy. And it's not even necessarily reputation amongst others. It could be in many cases, but not always. They just have to have it.
What's my point? Well, this is the greatest strength and weakness of peer to peer. Hoarders ensure a healthy flow of files, but they rarely actually check what they have. They don't check to see the software works, or if the music is a complete copy, or that the movie was cut down to a quarter of the original screen size.
This is what companies take advantage of, both those who want to hurt swapping, and those who just want to seed files for the purpose of installing some evil spyware. It's nice to have a bunch of people trying to seed the masses but cmon the point of file sharing is to pool our independent resources. For someone who doesn't have all day to search for files and test quality and whatnot, it is sometimes less painful to just go buy the CD than it is to actually try to download it amongst the mess of files that are out there.
"All great wisdom is contained in .signature files"
Many many companies (and individual artists) have faced SERIOUS economic damage by attempts to thrawt P2P from being absolutely ubiquitous and maximally effective. Estimates are in the BILLIONS of dollars (US only) of lost sales in broadband connections, blank media disks, large hard disk drives, software support, consulting fees, home audio/video equiptment, and the like. And Western countries are fast falling behind as the majority of educated citizens from developing nations take advantage of the black market for these goods and services while Western citizens are blocked in droves by propaganda, political corruption, inferior substitutes, and FUD from fully participating in the open exchange of science, the arts, poltical discorse, and culture in general.
Credence will hopefully bring us a bit closer to reaching our current potential.
Use a P2P program that actually includes some 'anti-junk" features. I typically use Shareaza (probably not the best, and I'm sure someone will state a better P2P but the points still remains, Shareaza does offer some features these clients do not -- including a rating/comment system that goes with the file whenever anyone finds a search result for it). Usually I know if the file is a fake before I download because I use some obvious signs:
I prefer the client program including these features, especially when it's available to connect to several networks at the same time. Nothing worse then getting a 100MB+ file and realizing you wasted the bandwidth for not, or the program you downloaded wasn't the same as the file name (more legit, but not what you were looking for).
Do be careful because some files that are really a virus can be detected by AV as 'ok'. Thankfully I found the virus before it did much damage and by reading the Symantec AV report I was able to make sure I removed it completely. Just because one 'setup.exe' claims to be a setup program don't trust it unless you trust the name of the setup program -- "Program Setup Wizard" does not cut it!
Since Shareaza also supports torrents I usually go through torrent sites and have rarely had any 'junk' files from the torrents. The more junk the RIAA (and other companies!) try to spread the better we get at ignoring and working around it!
Actually, while I doubt the OP intended it, he has a good point.
See, let's be honest about this. While there will *always* be jackasses out there who spam networks just because they can, and a few more people trying to shove spyware down people's throats, a pretty big chunk of the folks producing spam are those trying to prevent their copyrights (however overly-long-lived they may be) from being infringed upon.
So, the point is, that it's a good bet that a sizeable chunk of the files being shared aren't exactly legal.
Which means that you don't really want to make it especially obvious that you're sharing said file.
What this system does is provides a cryptographic signature on a small, publically downloadable piece of data that establishes that you have downloaded and *consciously examined* the file.
Frankly, this is pretty good evidence for someone trying to push an infringement lawsuit that you have infringed upon their copyright (yes, our work has MD5sum "foo" the same as the thing this guy is rating.
That being said, I do think that a more sophisticated method to this approach will win.
The largest problem on the Internet has always been rating and attributing data -- Google takes a pretty decent stab at some of the problem, and look how essential they've become. This just does a much better job.
Any program relying on (nontrivial) preemptive multithreading will be buggy.
Organizations such as the RIAA and music labels regularly pollute these networks with nonsense files masquerading as real music/video files. ...as do the "renamers". I wonder if anyone has studied why such people rename files in this way?
"You can justify anything by putting it in quotes, adding a famous name and making it a sig" - Albert Einstein
So, when's Slashdot going to impliment this "golden" system?
From the FAQ:
Join moola.com, play games to earn money.
Whoops, posted too soon. The second potential problem you describe is more in line with how Credence is described to work, but I think it's unlikely to be a very big problem. Yes, the system will probably allow for "mistakes," but it will cull those mistakes out. So if the spammer rates most good files good and bad files bad, but rates their one spam file also good, then it is possible your client will report that spam file as having a high credibility. But, once you (or anyone else) download and find that it is not a good file, you will rate it bad, and as more people rate it bad, its credibility will go down. It's a case of diminishing returns for the spammer.
Join moola.com, play games to earn money.
Ummm, yes there is. For instance, VLC media player will play partly downloaded videos.
http://www.sydney-webcam.com
This may automate the reviewing process
1. Mark a bunch of good files as good
2. Mark your bogus file as good
3. Spread your vote list on zombie network
4. Your votes corrolate highly with "good files", and there's no counter-votes by others (yet)
5. Trick lots of people to download it (the rating goes to shit eventually, but...)
6. New bogus file. Goto 1.
In addition, you have an issue with semi-good files. What if the encoding is flawed, should you mark it as bad or good? Either case can put you at odds with the general opinion.
Third, you have an issue with files trolling for incorrect votes. Create a "non-obviously" bogus file, which some people will mark bad, others good. You'll create a lot of conflicting votes and "noise" in the system to make attacks like above possible.
Kjella
Live today, because you never know what tomorrow brings
True, but Torrents rely on the community, while with things like kaaza, many times what you want is hosted by one guy, and it's hard to kick fakes. In torrents, fakes die very quickly, thanks to the 'OMG fake' comments on the torrent sights.
Who actually searches for files in the P2P client? Normally you visit some site where the releaser himself posted a torrent or an ed2k link and you download that.
I can't remember the last time I actually searched in eMule.
Of course it runs NetBSD. BTC: 1NT7QvbetmANwaMzhpVL6
couldnt agree more. P2P is superb stuff, and has all kinds of legit uses, but to pretend that its not 95% used to download copyrighted music and movies and thus save a few bucks is just denial.
There are far too may slashdotters who reply to any article on copyright with "get with the system dude! copyright is over!" usually they seem to be 13 year old kids who dont understand what its like to have your income and career based on developing electronic products.
Do people really think that Lord of the Rings deserved to sell just 1 copy, to the p2p hacker who ripped it?
DRM-free indie games for the PC and Mac: Positech Games
Can this also be used as a metric for the RIAA and MPAA to decide which people to take legal action against? Go for the most trusted, most highly rated individuals and take out the most influential (central? critical?) nodes. In the same way that cliques of poisoners would stand out.
Xix.
"Everything is adjustable, provided you have the right tools"
Do people really think that Lord of the Rings deserved to sell just 1 copy, to the p2p hacker who ripped it?
There was a time when home video didn't even exist, and yet movies still got made. After their initial run in theaters, movies would only be shown every now and then on tv late at night. They were basically filler programming for what would have otherwise been dead air. One might even argue that the average quality of new movies has been on a steady decline since then.
Just because I can package something for individual sale and ask people to pay for it doesn't mean it's right, copyrighted or not. Especially when the producers more than recouped the cost of production long before going to video and that copyright is going to be extended ad infinitum.
I disagree that these scientists are breaking any *legitimate* law, but if you accept as a premise that they are, then they are in fact breaking the law using taxpayer dollars.
Instead of modding that down it should be modded up so more people can discuss the ramifications.
Do we allow taxpayer dollars to be spent on civil disobedience? On that issue, I am very unsure.
--- Grow a pair, liberals... stop letting the Republicans bully you!
Despite everyone's views on the use of p2p networks, isn't it a dangerous precadent to set to allow these companies to steamroller over *anyone* who dares share copyrighted material. Is living in a DRM world where consumer rights are constantly reevaluated as to give us the least amount of enjoyment and freedom from our purchases worthwhile? It doesn't matter *who* these p2p sharers are, isn't setting the precadent of removing consumer rights by DRM (to copy, rip, burn for backup etc.) far more demeening and indefencible?
"Just because I can package something for individual sale and ask people to pay for it doesn't mean it's right, copyrighted or not." Hi, this is reality talking, we'd like to know where you could have possibly formed such a delusional idea that people don't deserve to sell the things they made. I guess I will just go over to McDonalds now and grab a big mac without paying. I don't think it's right to pay and I haven't eaten for about 2 hours, so why should I have to pay?
Too bad he wrote as AC, but he has one of the most insightful points of the entire thread, an entire aspect that is overlooked.
The basic premise of the slashdot story is how cool it is that researchers are defending the acts of people to trade in uncorrupted *illegal* file trading.
After all, it seems the most if not all corrupted files are ones that, if they weren't corrupted, would have been illegal to trade anyway.
I think the RIAA and MPAA are scum sucking pigs who need bacon carved off their arses and handed to them. Still, I also think their concern about massive illegal fileswapping is legitimate, even though the leaders of their respective industries are the ones truly responsible for raping their own artists...
The system seems like a tool to use against the RIAA/MPAA to block pollution efforts. However, then the other shoe drops, and the RIAA/MPAA has a tool to target the highest ranked nodes/cliches/people. No longer do they need to figure out how many files you have.
They just have to find one file, extrapolate your rank to the average system rank, run a few numbers (and maybe a few inflated costs in there too), and bam... for sharing Happy Birthday To You.mp3, you get slapped with a $1 million infringement case because you happen to rank as a very high legitimate link.
On the other hand, this might be benefitial to take the heat off of the majority of the file trading community that honestly is NOT costing them any money. They don't need to target the casual "weekend downloader", who's rank should be significantly lower (being a new node on the network) than some guy with 4 160GB HDD's of the latest releases to theater and DVD. Nobody feel sorry when these guys (or gals) get busted. When 14 year old choir girls get busted, there is PR hell to pay. This system allows them to do that.
Didn't RTFA, but that's my first impression. A use to boost network quality, a use to increase (not decrease) the reach of the **AA's, and a use that may help both sides.
"Every tool has at least 2 completely unassociated uses. A spoon can serve food to your mouth, or gouge the eyes out of your enemies." - Me
I8-D
I haven't tried VLC, but mplayer will usually play partially-downloaded torrents if they are mpeg files. It just skips the bits that haven't been downloaded yet.
It's less reliable with avi files: it doesn't seem to like it if the first part of the file is missing.
And I don't see why they'd bother, when a threatening letter is all it usually takes to take a torrent site down
That's not really true. Depending on where the site is hosted, legal threats could be more humerous than scarry.
Case in point.
Btw, if you've got a few minutes to kill, you should really check out some of the emails to and responses from thepiratebay.com. They are hilarious!
Will this make finding CCR on P2P harder?
"As you can see, your honor, according to a ranking system on the pirate file-sharing network, the accused had a high rank for carrying real, pirated files."
No, thank you.
vk.
" if p2p files are legit, why do you need checksums.."
To verify that the file is, in fact, legitimate. There are a number of unscrupulous folks out there that would just love to have even just a few people install their trojans. As Ronald Reagan said "Trust, but verify."
- Fake files. This is clearly a more primitive tactic and can be thwarted by clients that can be set to download the first parts of a file first.
- Incomplete files. The seeder reports having the entire file, but will never deliver certain parts of it. Thus, downloaders get stalled at 98.5%. And it's amazing how long people will wait for that last bit.
- Fake seeds. Haven't confirmed how this one works, but sometimes you'll see a torrent with an improbable number of seeders (e.g., 300 seeds and 100 leechers for a fairly new torrent). Lots of seeds attract more people.
- Timing. For example, demand for a movie will rise in the days shortly before its release. If you get your fake tracker up and running during that critical time before there's a real pirate version out, then you'll attract downloaders and waste their time. And there's a snowball effect: when people go to download from BT, all of things being equal they usually go for the tracker that has the most people on it.
Combine the tactics, and you've got a serious problem. Every user adds to the strength of the distribution network so tying up one client with a fake not only prevents that client from getting the material, it also keeps that client from helping others get it as well.If you're patient, persistent, and knowledgeable, you can avoid or minimize the impact of these spoofing tactics. But patient, persistent and knowledgeable don't really describe the average pirate (or just about anyone else, for that matter). The dedicated pirate simply won't be stopped, and the content producers know this.
Like you, I once assumed that the various forms of moderation on the torrent sites would mitigate this. But the countermeasure are slow to work, as I've seen fake torrents stay up for weeks. It's easy to post multiple new fakes. And users are incredibly clueless. I have, on several occasions, seen comment threads where several people will post "This is a fake, don't bother," but the torrent will still have thousands of people downloading and the very next comment will be something like "I've been stuck at 99% for three days, will somebody fucking seed this!!" Remember, the goal isn't to elimiate the network. The goal is to make it so untrustworthy and unreliable that it's too much trouble for Joe User and he'll go to the theater instead.