Hi there .. I'm currently sharing this file on the network, however, as I'm constantly hitting my maximum number of uploads and topping off my upstream bandwidth, I'm also copying and pasting it here. Please create a text file and share this on the network. And please pass this on to any coders you know.

Anyone coding a gnutella client for Windoze or Linux, please read this. Please pass this file on and share it.

Freeloaders. No one likes em.

First, for those of you who are no longer sharing files because you are tired of freeloaders, all I can say is that you are part of the problem. This kind of mentality will only speed up the end of this network rather than help it. Maybe you don't care, but as someone who has nearly 2,000 files shared, I think pulling your files out of your shared folders just plain sucks, no matter what the reason.

However, given the *current* interface, there's little immediate incentive to share files.

What needs to change? HOTLISTS! Well, improved hotlists.

Say Joe starts downloading a bunch of songs, or whatever, from my host. If I can right click on the download and see all of Joe's files (for instance, a shortcut to his web page list), that tells me whether or not he has files in his shared directory. Now, if take that one step further and determine whether or not Joe is allowing others to download his files, we have usefull information.

How, you may ask, is that usefull? Simple. If I find out Joe's a freeloader, I'll boot him. If I see him again and he's still freeloading, I'll block his ip. Sooner or later, freeloaders will get the message and either start sharing or get lost. Either way, the network will improve.

The only way this is gonna work long term is if we start treating this like a community instead of a common grazing field. Communities have cops. Nah, I don't want people poking around in my files to determine if everything I'm sharing is 100% legal. The kind of cops I'm talking about a modified version of the netcops in usenet. What I'm talking about is providing everyone with the type of information, in a quick and easy way, that we need to start enforcing the golden rule.

P.S.

I'll pay money (just my fair share ... I can't afford to finance development) for a client that will do this for me. Contact: lurker701@yahoo.com

Lurker,

Lets slow down and think about what you are saying here. You aren't getting your downloads as quickly as you would like and you have decided it's because of all the freeloaders. I'll assume You havent recieved the certified letter from your ISP outlining your "Terms of Service" and how it applies to the sharing of copyrighted files. Some users have very good reason for not sharing your quota of files. These "freeloaders" may still be benefitial to the network as a whole by serving as supernodes - nodes handling a very high bandwidth of network traffic, linking thousands of users and directly contributing to the success of your file searches.

You seem to think that you should have the ability to inspect the shared file library of every user that requests a download. Then you can put those worthless freeloaders in thier place, Right? OK, maybe you could stop a few uploads to users whose share library doesn't meet your criteria, but you may very well be the only person still sharing anything. Most of us know big brother (Metallica) is watching, and are not inclined to declare the entire contents of our shared file library with every download request. IMHO you would see more sharing if the security and anonymity of the network were improved, not relaxed. We don't need any more netcops, modified version or not.

As I am sure You are well aware, with every implementation of this sort You are faced with the prospect of people trying to beat the system. Well I could take a screenshot, encrypt it, multiple copy it,wrap it all up in a zip and rename it to something like "Metallicas_ Favorite_One.MP3". Great!, now I have a legal, self-made 5MB file that fits your criteria. After I've done this numerous times,using differend file sizes and names, I would seem to have a portfolio even the most judgemental would deem worthy. What's even worse is the fact that other users would download these bogus files and leave them in thier shared folder for no telling how long befor they are finally deleted, all the while sharing them with even more users. We don't want to give people any incentive to try something like this. Just look at what is happening now with the fact that some clients and users give preference to other users with more than a set number of shared files. We are getting reports of an enormous number of files avialable, but without the expected increase in MBs. People are putting irrelevent files in thier share folder just to pad the file count. These irrelevent files all have names and thus are prone to returning bugus finds to legit searches, needlessly driving up network bandwidth.

Just a thought, not necessarily meant as flamebait.

Anyone else have suggestions or opinions on dealing with freeloaders?

The Gnu Age Philosopher

Point taken. However, I don't think our goals are mutually exclusive. You want more annonynimity, and I want to be able to see whether or not people who are taking up space in my upload que are actually sharing files with other people. Fine. If we can make the file sharing process anonymous, we can certainly make the list of files shared anonymous.

My understanding is that future versions of gnutelliums will provide an increased degree of anonymity in the file transfer process. Of course, no p2p protocol can be totally anonymous, and an experienced hacker will be able to tell who is who and what is what, no matter what privacy measures are added to the protocol. Still, if they can tighten the anonymity on file transfers so that a casual user can't figure out exactly who or where you are, they should be able to do the same thing with file lists.

Quote:

You seem to think that you should have the ability to inspect the shared file library of every user that requests a download. Then you can put those worthless freeloaders in thier place, Right? OK, maybe you could stop a few uploads to users whose share library doesn't meet your criteria, but you may very well be the only person still sharing anything.

I doubt it. There are plenty of people sharing files now, and most of those who are sharing also have their lists available via a web interface. I don't see why those people would stop sharing simply because it's easier to view that list

Quote:

Most of us know big brother (Metallica) is watching, and are not inclined to declare the entire contents of our shared file library with every download request. IMHO you would see more sharing if the security and anonymity of the network were improved, not relaxed. We don't need any more netcops, modified version or not.

Heh ... let's face it, Metallica ain't that smart. Maybe MS is, but I doubt it. They're interested in keeping an eye on gnutella atm, but I doubt they see it as a serious enough threat to start going after individual users on a large enough scale to affect the network. And the truth is, if more people were sharing files, they wouldn't be able to if they tried. The biggest threat to the network is from within, not from the outside.

Quote:

As I am sure You are well aware, with every implementation of this sort You are faced with the prospect of people trying to beat the system. Well I could take a screenshot, encrypt it, multiple copy it,wrap it all up in a zip and rename it to something like "Metallicas_ Favorite_One.MP3". Great!, now I have a legal, self-made 5MB file that fits your criteria. After I've done this numerous times,using differend file sizes and names, I would seem to have a portfolio even the most judgemental would deem worthy. What's even worse is the fact that other users would download these bogus files and leave them in thier shared folder for no telling how long befor they are finally deleted, all the while sharing them with even more users. We don't want to give people any incentive to try something like this. Just look at what is happening now with the fact that some clients and users give preference to other users with more than a set number of shared files. We are getting reports of an enormous number of files avialable, but without the expected increase in MBs. People are putting irrelevent files in thier share folder just to pad the file count. These irrelevent files all have names and thus are prone to returning bugus finds to legit searches, needlessly driving up network bandwidth.

Yeah, well, there will always be those attempting to break the system. That's part of the reason gnutella exists in the first place. But let's face something else: most freeloaders ain't this smart either. Those who are already padding their file counts will continue to do so. Those who aren't padding aren't likely to start simply because of my suggestion being implimented. My "system" wouldn't provide any *added* incentive to go through all the trouble you've described.

Bogus files are something that you have to deal with if you're going to use this system. And so are freeloaders. While we can't totally eliminate freeloaders, we can discourage the activity. That's what I'm suggesting we do here.

Quote:

Just a thought, not necessarily meant as flamebait.

Normally, when I see this in a post, I automatically assume that it *is* indeed flamebait. I was pleasantly surprised.

Lurker701

I seem to have read that something like 95% of the files transfered over the network are supplied by 5% of the users. I'm assuming this is media hype because I dont think, with the present system it can be catigorically stated how much any host, even an anonymous host, uploads. I'm afraid if it was a proven fact, built into the protocol in a tamper proof way to show how much even an anonymous user uploads it may spurn the dark side to devote the resources necessary to ferret out these hosts, and instill in them the proper respect.

I completely agree with you that freeloading is a major problem and seems to be getting worse. I just dont want to inadvertantly give anybody the rope to hang us with.

Give me your honest opinion on this idea. Along with every download request the requesting client would send a 2 digit number called a GQ (generosity quotient). This GQ would be based on the ratio of uploaded MB to downloaded MB and would be a pain in the *** for the user to manipulate. Users should be given credit for some percentage of upload network bandwidth and not just uploaded files for those who have recieved the ISP's letter. A brand new user would start off with 0up and 0down balance = 50GQ, not bad, a respectable user like that would have a 50% chance of getting thier request honored on the first request. After that first download however thier GQ suddenly goes to something like 1 (1%up-99%down), they have now joined the realm of the freeloaders. These unfortunates now have
a 1% chance of having thier request honored on thier first request. The only way out of the freeloader bracket is to stay connected and contributing to the network by handling network traffic. After doing this for awhile someone downloads thier file ... Yes, out of the freeloader bracket of lets say anything under 10GQ, hopefully never to return. After just this one upload they are now able to download 8 more files (with a decreasing GQ after each file) before again going below the cutoff point of 10GQ. They have been given a basic lesson in the value of sharing. Why am I talking about a cutoff point below 10GQ when I just said rank freeloaders still have a 1% chance? Because every 100 requests the client would send a totally random number
as the GQ, this would generate enough bogus 98 and 99GQs to prevent anyone from tracing down hosts that honestly earned thier high GQs.
I too was pleasantly surprised to see that you had taken my comments in the spirit in which they were offered. My posts sometimes seem to generate flaming responses for some reason. It seems to me that some people seem to take offense when none is intended. Others seem to flame just for the sport of it. I'm glad to see Your not like that, cool ideas rarely seem to come out of flaming discussions.

Regards,

The Gnu Age Philosopher

The idea is indeed apealing ... although it would take some tweaking to make it work. Take, for instance, me, who has a little over 10 gigs consisting of roughly 2000 files shared. My GQ under your system would probably give me a fairly low score, because I have a cable modem with upstream speed capped at 128 kbps. Because of this limitation, I limit file uploads to 3 users at a time, one file per user. This way, people get reasonably decent speeds.

I consider myself a good user ... I often remove the bandwidth limiter, especially if I'm asleep or away from the computer. And I leave my servent running as much as possible. My upload ques are always filled, and I get lots of hits on my material. I also connect to betwen 7-10 hosts (more if I'm away from the computer), increasing the conectivity of the network. Still, because I upload a lot slower than I download, your system would probably end up penalizing me.

Secondly, this might discourage sharing of rarer, but cooler, files because people want to keep their upload stats inflated.

Third, this doesn't really get around your objection re: bogus files, since people could inflate their stats by encouraging uploads of crap, particularly as a newbie before they've accumulated lots of files.

Fourth, you're talking about a drastic change in the protocol in order to allow the GQ stat to be reported along with the rest of the information returned in searches, etc.

On the other hand, a similar quotent based on the total size of files shared would be easier to impliment, and fairer to me (remember, in this community, it's always about me .

Simpler because the total number of files shared and total size could be calculated based on the listing of users' directories. This information is already available for many servents via the web interface, and it would be relatively straight forward to enhance this so it worked more efficiently, without requiring servents to respond to searches differently.

Fairer because those who are sharing as much as possible, but limited by their isp in their upload speed wouldn't be hurt.

Still doesn't address to issue of bogus files, but there's no real way to address that issue effectively. I'll simply restate that your option doesn't effectively deal with the issue either.

Your idea has the benefit of being totally automated, though, whereas mine would require some human intervention. Although, i guess it's possible to have some sort of criteria in the upload window "[]Only allow uploads to servents with _______ MB shared"

Either way, we're talking about adding a new dimmension to the network, which is in effect finding ways to enforce the golden rule.

That makes you a netcop too Your badge will be delivered soon.

-Lurker

Anyone who would actually take the time and pad their download folder instead of just sharing what they have...?? Man, that seems like a lame waste of time.

The problem with a "GQ" is that the number of files shared (and thus the GQ) is determined in large part by bandwidth. Dial-up users might share 40 files (which is more than fair) and yet only get one upload per every 10 downloads. A T1 user might share 2000 files and get 10 uploads per every five downloads. The point is, if you have a slower connection, you shouldn't have to share as many files, and what's more people are less likely download any files you are sharing if someone with a broadband connection is sharing the same file. IMO, dial-up users shouldn't have to share anything to participate.
I think the best thing to do about freeloading is simply to design better clients. No client should allow a file to be downloaded to a directory that's not shared. Of course you could still move the files out of the shared directory after downloading, but I think most people wouldn't. Another thing that contributed to the high sharing levels Napster had though was its internal media player (since users would connect and share files just when they wanted to listen to their music). I personally prefer Winamp/XMMS (as do most developers I imagine) but that doesn't mean Gnutella programmers shouldn't outfit their clients with media players simply to promote greater sharing.

If you're connected to Gnutella through a proxy, I don't care what kind of hacker you are you can't figure out who the file sharer is unless you are the proxy machine (or compromise the proxy machine, which probably isn't possible for a simple Gnutella client). Not totally anonymous, but that kind of anonymity has little to do with hacking skill.

True, however, you're ignoring a few basic facts here.

1) Proxies aren't modifications to the protocol. They simply forward packets to and from appropriate machines and ports. Using a proxy as an attempt to mask your identity is an old trick, and predates gnutella, napster, et al.

2) (and more importantly), If you're using a SOCKS proxy, chances are it's run by your isp, or some other place you have an account. While it would be difficult (and probably illegal) to trace you immediately, you are still traceable. Simply put, if someone sees copyrighted material being shared up through the proxy, and they've gone through the trouble of tracking it that far, they could just as easily register a complaint to the isp that provides access to the proxy machine. They, in turn, would be forced to kill your access to the proxy or face disconnection themselves. If the proxy is being run by your isp, then you're back where you started in the first place.

In short, if you're afraid of the big bad wolf, proxies are the equivalent of a straw house. I'm not sure what the brick house looks like, and I'm not sure it can be built legally and still allow all the piglets inside.

One thing is for certain: there is no privacy on the internet, and there is no annonymity. Anyone who claims otherwise is either trying to fool you or trying to fool themselves, or both. Or they're obtaining that annonymity illegally.

That said, I still think the biggest threat to the gnutella network is from within, rather than from outside the network. How we fix that is up for debate, as always. Maybe the GQ idea isn't the answer.

I still like my original answer of providing lists of shared files so that users can determine for themselves whether or not the folks downloading their material are legitimate contributers to the network. Include connection speed, if you think it's an issue. And mask it behind bogus usernames, proxies, or whatever makes you feel safer.

Just don't pretend that you have any privacy at all. If you're sharing copyrighted material on the net, the only thing that can save you is the law of large numbers.

In the end, all my suggestion does is make it easier to access the information. It's already out there as part of the gnutella search protocol, and anyone who really wants to find it (i.e. people who have a real interest in finding it, for example the big bad record companies or software developers) can get it. All I'm suggesting is that the coders make it available to the rest of us too with an easy to use interface.

Sorry, I should have made more explicit what I meant. Proxying is actually something very simple to implement (I once wrote a proxy server in Java in about 20 minutes). By proxy I didn't mean a proxy server, but some kind of proxy feature that could be implemented directly into every client (or just some clients) on the gnutella network. (A SOCKS is just a proxy that decouples protocol from software and has nothing inherently to do with ISPs.) The way that might work is by issuing a search for "prxy79" (or something singular like that) over gnutella from the computer that desires a proxy connection; establishing a proxy connection with one high bandwidth / high up-time host that responds. That's actually not hard to implement, and a SOCKS proxy could even support other protocols (such as http) and is pretty good anonimity, especially since (although not with gnutella) the proxying could use public key encryption.
For example, a host might issue an anonymous query through one computer, get back a result + a public key, and request an encrypted download through a second proxy. The serving computer has no idea who the downloader is, and the proxy has no idea what the content is. Imagine now that every client on this network supports at least one proxy connection and you have thousands of computers working for anonymity versus a small number of policing computers working against it. To even get significant enforcement results there would need to be more policing than contributing computers (because you would have to be so lucky as to own at least two computers participating in that transaction, and *then* have to deal with the encryption). Connect to a second proxy through the first, and it is so close to impossible to identify both the person and the content that we may just as well call it impossible. Furthermore, those users with no proxy connection even get some additional anonymity because it can't be proved (at least to the satisifaction of the courts) that they are the ones making the transaction and not someone else proxying through them.
I think the reason someone might say internet anonymity is impossible is because we're used to thinking about the internet in terms of a client and server. P2P changes all that, and it will change a lot more. Remember, if 30 years ago you tried to argue that public key encryption was possible, even intelligently, you would have been laughed right out of the cafeteria.

Interesting idea. It could very well work, but it could also crash and burn.

What troubles me is that you're asking these high-bandwidth machines to essentially act as shields for the rest of the gnutella network. The policing you're refering to doesn't even need to occur to bring the network down. Simply pay your employees a few bucks an hour to pull stuff down on gnutella through these proxies. Wham. You now have a situation where the proxy machines either have to ban every single person who's using their machine to share copyrighted data, or be disconnected by their isp through the typical complaint-and-ban process.

Certainly, it's quite easy to determine who your proxy is, even if it's not apparent in the gnutella interface itself. A simple netstat will be enough to determine that much. The encryption isn't an issue because the policing machines *are* the machines recieving the public key. And now you've made the network more centralized, with a handfull of high-bandwidth proxies to go after instead of thousands of users. Even if the proxy machines aren't sharing any files themselves, if they aid in the distribution of copyrighted material, they can be shut down if they persist in that behavior.

So, yeah, I guess you can provide anonymity for a good number of users, but only by putting a smaller, and therefore more vulnerable, group of users at greater risk. Under those conditions, I wonder who would be willing to act as a proxy in the first place.