Ya, I never thought about the shared files skewing the data. I have no videos, but just about everything else. So if one client is used mainly for videos then the numbers are going to be low in my data for that client. Other then that my files cover a broad range of file types that should give fairly good data.

Maybe I should download a few videos, just for testing purposes.

1) What (objectively) is PHEX allegedly doing to mess up the Gnutella network? Is it really behavior that needs to be corrected to bring it into compliance with Gnutella 0.4?

2) I am beginning to write my own client (which will be a plug-in to a front end which I'm designing). Is there somewhere I can go with technical questions about the "Bearshare trailer" to the QueryHit descriptor, how clients identify each other (I don't see it in the 0.4 spec anywhere!) and stuff like that?

BTW 1.5GB and 357 files are how many I am sharing, not how many were downloaded from me... this whole thing was done by hand so it was kind of hard to keep a track of that kind of information. I did notice BearShare users downloaded a lot more from me, but multiple downloads didnt add to a clients count, only unique IPs did. I also didnt count unknowns.

Well, I dont think its really on the content as much as the general size of the content. video tends to be larger then audio, and audio bigger then images... other kinds of files like zip files can range greatly... but anyway bigger files will likely be what broad band users look for and smaller files for others (there may be more 56k users that prefer Bearshare for all we know, actually that would be an interesting statistic to see in terms of download bandwidth of diffrent gnutella clients, I'll probably try that tomarrow night ). bigger files also mean a system is held up much longer, on certain occasions mine was held up, which can also have certain effects on results.

One thing I will say for sure, people sure like pron...

"this whole thing was done by hand" Man, you are hard-core. I would never think about doing that if I had to do it all by hand. I was counting multiple downloads in the file and GB counts, so one IP is downloading an average of 3.85 files from me.

Even though I do not have any videos I still have good range in file size, including a good number of larger zip files that are comparable to video sizes. I don't have any data about download bandwidth to clients. But I do get a good range of download bandwidths from other people. Since my connection is really fast I never get enough uploads at once to reach my max uploads, so they never get a busy signal from me, even though I have larger files that take people a long time to download. So my system is never held up.

I don't have any porn, but I don't think that is client specific.

"After some discussion with the developers of Limewire and Bearshare I had to change Phex´s automated searching behaviour. It causes less traffic now while increasing the high probability of sucessful downloads." -copied from the Phex home page.
http://www.konrad-haenel.de/phex/

Moved this post info to the top one, so it would be easier to see.

Isn't that just cool.
Good point! I totally agree with you, just have to say that I didn't start it.

I just thought about something else, this only gives you a percentage of users that download, this information does not include people who share. This could very well effect the numbers, partly because BearShare is by far the most advertised (I heard it mentioned on TechTV a number of times along side Napster and other users, sites like zeropaid put that as one of their favorites despite the user ratings), this can skew things by getting most new users who dont have much to share, are out there looking for things to download. Then there are those who download a few things and share a lot and have tried diffrent ones, have less things to download (experienced guntella users). So its more likely the newbie will be accounted for more then the experienced users by tracking what people are downloading.

I think a more well rounded statistic would both include uploads and downloads... I'll give it a try (I need to free up hard drive space though ), see what the results are...

How would you get data about what client people that are sharing are using? Or what client experienced people are using? Also I don't think that those two group are necessarily the same people.

The only way I know of to do this would be tracking who you personaly download from. Sorry but tracking who you download from is what Kirby thought that I was doing, and Kirby was right it is meaningless data.

I was not trying to track who is sharing, I was just looking at the number of people using each client in comparison to one another, i.e. the popularity of each client. And for this purpose my data is more accurate and larger then any I have even seen before. Mainly because I have only seen personal opitions or data from about 100 unique IP addesses or less.

You are right LimeWire used to be the popular one, but now BearShare is so older users most likely sharing more or going to be using LimeWire more, and the new users are going to be using BeaShare more. But since I am not trying to find old vs. new user or sharing vs. freeloading these don't matter. I am only looking at popular vs. not so popular with the users on the network today.

If you want try to tack something else please do, but this thread was started to track client popularity on the network, and the voting poll was for people to vote for their personal favorite client mainly so they could disagree with the client popularity data if they wanted to.

If you want to do something like sharing vs. freeloading I am sure that lots of people would be interested in it.

By downloading from them... Phex gives me this information... admittedly there is some issues involved in doing that, because in some cases someone sharing will disconnect and then phex will connect to someone else to download from, so I have to monitor who I am downloading. I will have to try to download a variety of files, from small to big, and diffrent formats and genre as a means of keeping diversity.

There is no exact way to do that, but someone who shares will more then likely have more experience and adding downloads counts from each client will grab a more diverse group of clients as oposed to monitoring just shared uploads alone.

How so? People who share make up a certain percentage of the entire network, they dont make up 50% of the network, but they do make up 25 to 10 percent of it. To not include that information is to be 25 to 10 percent off in our statistics.

That wasnt my point though, my point was that, that does affect our statistics. It does not need to be accounted for, but both users that share and dont share do affect our statistics.

I'm sorry you didnt quite understand this, maybe I worded it to wrong, but still we are talking about the same thing.