Talk:Gnutella

From Wikipedia, the free encyclopedia

Contents

[edit] List of Software

I find it a bit ridiculous that the list of Gnutella software makes up for about 30% of the article although it's pretty insignificant information. If nobody objects, I'll move the list to a secondary article List of Gnutella Software in a while. That way everybody can add his favourite and it won't clobber up the Gnutella article. --82.141.60.62 17:02, 28 October 2005 (UTC)

I don't see any reason to split it from the article. This is a relatively small article and it can easily handle the list provided. - Tεxτurε 17:03, 28 October 2005 (UTC)
Would you find a list of automobile vendors useful in an article about "Car"? I don't think lists itself are of encyclopedic value (except for directory entries or meta-articles maybe) and a screenful of such a list is beyond what I consider appropriate. Whether the list items here are really of significant value is quite questionable for example Xolox has been unmaintained and dead for years. In an article that provides a generic list that really doesn't matter but here it clobbers the article and distracts from the actual content. I even suspect that some entries exist only for publicity reasons. So to prevent an annoying discussion about which clients should be listed, I suggest to put the whole list elsewhere so that it doesn't disturb and may be excessive. Those vendors which provided sigificant contributions to the development of Gnutella are mentioned in the article anyway. --82.141.60.62 18:55, 28 October 2005 (UTC)
I have copied and updated the list in Comparison_of_Gnutella_software. However, I have removed older clients. Ie, if they have not been updated in three years, they will not connect to the current network. This article is also in the "See Also" section. I tend to agree that the list detracts from the article. It is fairly easy to click a link. Bpringlemeir 16:40, 10 May 2007 (UTC)

User:217.227.85.218 wanted to put Phex, gtk-gnutella and LimeWire at the top. I don't think we should deviate from alphabetical order. I agree that many of the clients are extraneous. See the Comparison of Gnutella software. I would recommend that clients unable to connect to the current network should be removed from the list, but they should be kept in alphabetic order to avoid bias. For instance, the most popular client will always be available for Windows due to user count. Please help with the Comparison of Gnutella software (and possibly remove antiquated software from the list). Please don't re-order the list without discussing a rational here first. Bpringlemeir 20:25, 18 May 2007 (UTC)

I have merged Comparison of Gnutella software (see history)and added a List of Historical Gnutella Clients wiki. No one discussed anything here about the merge tag since June. Some of the clients in the main list are historical and should be moved. Bpringlemeir 00:49, 17 September 2007 (UTC)

I think that the latest Release of Shareaza should be upped too 2.2.5.0. I'm a member of the Shareaza community, and in fact we only make a difference between debug and release. Releases are the one with at least one zero on the end, and debugs are the one without a zero at the end. This difference is made because releases do not contain a debug database and a bigger binary (which is useful for developers to debug, but will increase the installers size 2 or 3 times). Debug versions do contain them. This change should be made because 2.2.1.0 was actually a bad release, with a lot of bugs harming the networks. If you advertise 2.2.1.0, instead of 2.2.5.0, it'll keep on bugging the networks. —Preceding unsigned comment added by 81.83.226.84 (talk) 12:33, 8 October 2007 (UTC)

Then go to the Shareaza wiki and change the software info box to the right of the page. It has a latest release and a preview release. Click on the +/- and you can change the revision numbers (and dates). It is very nice if you provide references (URL with ChangeLog, etc) in a "<!-- Comment -->" or in the edit summary. Bpringlemeir 14:16, 8 October 2007 (UTC)

[edit] Wikisource editors want to delete GnuFU

The Gnutella For Users user guide hyperlinked to by this article has just come up for deletion at Wikisource. You may wish to consider hyperlinking to the original, wherever it is on the World Wide Web, instead. Uncle G 10:19, 13 November 2005 (UTC)

[edit] Number of users

The estimated number of concurrent Gnutella users on Slyck.com is nothing else but the value as provided by LimeWire's crawler: http://www.limewire.com/english/content/netsize.shtml

As you can see these values have not been updated for almost half a year. Rest assured that despite all FUD, the number of Gnutella users is still growing. I believe the methods and accuracy to determine these figures differ quite a lot between the networks. They are not neccessarily comparable. Thus, there isn't really any point in starring at those values, consider them as the undeniable truth and update the article constantly. Actually, I wonder whether it's a good idea to copy whatever given values from whatever website without knowning who and how they have been estimated. --82.141.60.248 15:06, 24 March 2006 (UTC)

The network is substantially larger than 2.2 million users, although it is not tivial to determine its true size. I have written a small crawler tool to count users (I'm somewhat hesitant to publish it) but I can prove that the average size is somewhere between 5 and 10 million unique hosts. --137.226.77.2 21:29, 6 October 2006 (UTC)

having done some independant research into this area with my own crawler I can speak to the number of hosts on the network, like all network services usage statistics are different depending on the time of day and week that you check...on a school night (sunday-thursday) I see between 200 and 500 thousand users on at any given time, that number spikes around 100 thousand users as the western hemisphere hits nightly prime time, on weekends (friday-saturday) that number easily spikes to around 1 to 1.5 million (although I've seen it as high as 2+ million at times)...if anyone is really interested in arguing this to the very specifics i can get actual numbers and post my methodology but it'll take me some time --Michael Lynn

I believe something the casual reader will miss is that this number is the amount of concurrent users online. The number of people using Gnutella in general is magnitudes larger. --82.141.58.164 21:31, 12 December 2006 (UTC)

this is a good point to make, I can also speak to this although to get more specific numbers I'll have to check my data again as I dont remember, but the number of unique hosts on the network (not concurrently logged on) is closer to 10 million, however that data is bound to be scewed by dyanmic ip addressing and the like, if its really of interest i have ways of fingerprinting individual users but again it would take some time --Michael Lynn
I agree these numbers are a bunch of BS. Bpringlemeir 16:45, 10 May 2007 (UTC)

[edit] Gene Kan

he was one of the first to create an application to use the gnutella network and he's not mentioned on this page.. he later committed suicide.. worth mentioning

Wired story, Wikipedia entry. The client wasn't that popular, but anyone can modify wiki pages. I think that there are many people who are important to the development of Gnutella. I think that people have a tendency to glorify those who died. An entry on Gene Kan by itself would not make sense. There are many other people who have contributed to the Gnutella protocol. They shouldn't be punished for not committing suicide. Bpringlemeir 13:31, 15 May 2007 (UTC)

[edit] Pronunciation of "Gnutella"

Since this information is repeatedly remove from the article: Non-English speakers often pronounce Gnutella with a non-silent "G", for example, Germans and French. Gnutella is not an English "word", thus there's no right or wrong pronunciation. This might be hard to imagine for people who know only English and those might actually be confused when people talk about something that sounds like "Nutella" (the hazelnut spread). In other languages there's a clearly audible difference between "Gnutella" and "Nutella". If the original authors thought of this a joke (most-likely), it only works for native English speakers.

Insofar as GNU is pronounced "guh-noo", I've always pronounced it "guh-noo-tella". I'm an english speaker, FWIW. - 74.112.174.10 19:22, 3 January 2007 (UTC)

[edit] Add/remove Acqlite?

An anon just removed teh Acqlite entry with the edit summary: "Acqlite is an illegal stolen version of Acquisition and should not be represented.".

I don't know anything about Acqlite, so I won't revert, but it's a strange edit summary, so can someone else have a look? [1]

The removed entry was just "| Acqlite || Mac OS X || GNU GPL". Gronky 04:31, 7 March 2007 (UTC)

I don't really know anything about acqlite, but I do know a little about acquisition, it reports its user agent as "LimeWire/4.9.30 (Acquisition/122)", LimeWire is GPL...I'll take a look at it a little closer and see if its just another in a long line of GPL rips, but that is my suspition, and if its true then I see no reason to remove another client GPL just because its a "stolen" version of what would basically be a stolen version of limewire... --Michael Lynn
I restored Acqlite last night, but I've only just noticed this discussion, so sorry if I was premature. The license situation with Acquisition is a bit subtle: basically the program's split into two processes -- one with GPL code and one without -- which communicate through Unix pipes, so the author believes that it's not a GPL violator. Other people disagree though. There's a discussion about this on the Acquisition talk page if you're interested. Anyway, Acqlite is a fork by one of the people who think Acquisition ought to be released under the GPL. Epimorph 16:23, 8 March 2007 (UTC)
hmm, thats an interesting argument to be made, did they have to modify the gpl code in order to make it use this IPC mechanism, if so then surely that change would have to fall under the GPL, but IANAL, so I can't comment further, with regards to removing a client because it has been aledged to be infrenging I think that any such contraversy is cause for posting it on wikipedia and perhaps even writing more about it, it's not our place to deside what is and isnt infrenging, but if its a current issue related to the protocol and the p2p network then i'd say its worth talking about --Michael Lynn

[edit] Additional "See also" link

I know this is about software and the links all relate to other computer related things, but should I add a link to Nutella, as in the spread?--Maier 03 01:18, 8 March 2007 (UTC)

[edit] MP3Rocket

Greetings... my addition of MP3Rocket to the clients list was undid due to it being a scam. While I'm not going to disagree on that point in the least, I will note that it is a Gnutella client, and it does have an orphaned article which could use a little bit of sunlight.

Given this, I will re-add the link and suggest anyone who has more experience with MP3Rocket or whatnot (e.g. 82.141.49.41) update the page accordingly. I will, however, list the license as unknown... Rtucker 21:22, 14 April 2007 (UTC)

[edit] Comparison to Kad

So what are the differences between Gnutella and the Kad Network? Shinobu 15:27, 30 July 2007 (UTC)

Read Kademlia and Structured versus Unstructured. Gnutella will have a DHT added; actually, I believe that LimeWire has already added it in their latest version. Bpringlemeir 14:21, 31 July 2007 (UTC)

[edit] Weasel Words

...it is still one of the most successful file-sharing protocols to date.

how do you measure success? The wording is a bit weasely to me. --Echosmoke 18:41, 29 September 2007 (UTC)

Yes, those are weasel words. However, it is still functioning seven years after its first deployment and it has been in the top five population counts most of the time. However, currently p2p networks have gotten so large it is hard to measure populations and many people find them suspect to begin with. You can remove those lines. However, I find it hard to believe that you would find this article full of weasel words, compared to the G2 article (or many other articles I have read ) and it is at least based on facts not cited in the article. Bpringlemeir 04:57, 30 September 2007 (UTC)
Please stop bashing the Gnutella2 article as "weasel words" and "biased". Yes, it needs a cleanup. Yes, it needs cites. But I made sure I made the article as neutral and factual as I could. Every single time you run around and insult that article... well gee, it's not nice. Also, at this point I'm beginning to consider it a personal attack on my integrity. Please stop. -FrYGuY 06:18, 30 September 2007 (UTC)
The article isn't entirely written by you, so how could this be a personal attack? The G2 article is not neutral or it would mention some of the facts above. For instance, the Gnutella network has five times (or more) peers than G2 in the network comparison. Personally, I am afraid to edit the G2 page as anything I write would seem to create an edit war. I am only trying to point other people to that page to assay whether it is neutral or not. WP:Avoid weasel words and WP:NPOV are part of Wikipedia. How could I not discuss this? However, I am sorry I mentioned G2. Bpringlemeir 14:06, 30 September 2007 (UTC)
Also, I question my own objectivity, which is another reason that I would like someone else to look at it. I hope you don't find that too sinister and assume good faith. Bpringlemeir 14:31, 30 September 2007 (UTC)
Actually, the gnutella2 article is, in fact, almost entirely written by me. But, if you feel an entry on average network size is encyclopedic, feel free to add it, but I'll share a few reasons why I didn't add network size into the Gnutella2 article... one, it's not a fixed number. Both Gnutella and Gnutella2 change size with factors such as time and season (Both the networks tend to get smaller during the summer, and is generally larger during US work hours). Neither have accurate crawlers which can tell you authoritatively the size of the network. Gnutella's populaton, for instance, was at one point vastly under-counted (To the tune of almost 4 out of 5 nodes being uncounted), while at another point, before it had a real crawler, Gnutella2's was vastly over-counted (to the tune of each node being counted as many as three times), and at least during one point, the counting differences lead to people parroting that Gnutella2, mere months old, was larger. The figures from the only somewhat accurate crawler for Gnutella that I know of is no longer running, or at least no longer publicly displays the data. Gnutella2's only somewhat accurate crawler is often down. Mostly, I figure that the sizes of self-organizing ad-hoc networks is kind of pointless since people leave and join constantly, faster than even the best of crawlers can keep track of them. Yes, Gnutella is anywhere from three to five times larger than Gnutella2. But I don't really see why that's really important since a) both actively try to keep you from contacting the network unless necessary, b) Both networks change in size pretty drastically depending on time, season, phase of the moon, and color of the pants you're wearing, c) The results of a crawl of either network can return fairly significant differences in terms of assumed network size, depending on your assumptions (Limewire's crawler and Aenea have some pretty decent work in trying to calculate the network size accurately, but at the end of the day, they're both really just educated guesses with some evidence to back it up) and d) The actual size difference is pretty hard to measure when you can't accurately or precisely determine the size of EITHER network!
Back on the topic of Gnutella, I would heartily agree with the statement that it is one of the most successful file sharing protocols to date for a few reasons. One, it was the first ad-hoc self-organizing file sharing network, meaning that it didn't rely on a central server. Second, its popularity has remained pretty significant even as other networks have waxed and waned... FastTrack, Overnet, WinMX. It still has a number of clients which are popular in their own right: Gnucleus, LimeWire, Shareaza, Morpheus... I wouldn't say that they're 'weasel words', as the point (At least, some possible meanings of the phrase) is verifiable, although it is vague. Perhaps defining "success" in this context? Network size? Longevity? Total network bandwidth? All of the above? Gnutella is 'successful' in all three metrics, but to varying degrees of success, as well as varying degrees of verifiability...
As far as assuming good faith, I don't believe you have any particular agenda you're trying to push, or intentionally meant to demean me. I was simply annoyed that you chose to use my article (I did, as I mentioned, write most of the current version in one go, with most of the changes since then style and organization, and as such I'm rather paternal over it. Character flaw of mine, I 'spose) as a throw away 'weasel-y' article after the discussion we had on NPOV on the Gnutella2 article... it's the way it is because it tries to represent both the Gnutella Developer Forum's view of "Mike's Protocol", as well as the Gnutella2 developer's view. It's not "some people say" or "critics say" without citations of developers of one of the two categories (wherever I was asked for citations, in fact), and an explicit notation of WHO was saying the opinions. I really did strive to make sure that I didn't paint too bleak OR too rosy a picture of the network... to show both of the important viewpoints of the network, AND to show both the advantages and disadvantages compared with the network which it forked from. I really pride myself on being as neutral as I can possibly be, to the extent that while during the Gnutella2/Mike's Protocol flame wars, on the now defunct BearShare forums, Vinnie complimented me (That's how I took it, at least) as being 'not a fanboy', during a time where he flamed anybody saying ANYTHING positive about ANYTHING related to Shareaza, despite having been defending the decision and technical merits, and I wore that as a badge of pride, and STILL do... so when people come along, and describe what I tried to make as neutral, factual, and useful as possible as "biased point of view", it struck deep. The quality of the article, sure. I'll be the first to admit it needs work, and can use fleshing out. It's been a long time since the article was written, and in the almost THREE YEARS (Will be that long in less than three weeks) since I wrote it, both networks have evolved, Wikipedia's style has changed, much of the surrounding environment has changed (For instance, much of the discussion on BearShare's forums is lost for the ages, as is everything that was on Shareaza's old forums). But I took particular offense when your first change was to mark what I took particular care to write neutrally with NPOV, and now again with "Weasel Words", when both of those are things I made sure was as minimal as I could make it.
Wow, this has turned into a really long note. Summary: I don't think you're a bad person or necessarily meaning offense. -FrYGuY 05:54, 1 October 2007 (UTC)
woops..you two, have a cup of tea ;) Now, i didnt even want to question the success of gnutella, also am not implying the article is "full of". I just would prefer if there was some reasoning added like
  • exists for 7(?) years (longer than X, Y)
  • still active and functioning
  • user statistics (3rd biggest?) (yes pretty questionable, no matter what source)
  • several clients (actually a load of ;) ) still developed further
  • still more in use than its (non-)successor G2 (if that can be established firmly - to me its obvious)

--Echosmoke 04:31, 1 October 2007 (UTC)

Agreed, it's certainly vague. -FrYGuY 05:54, 1 October 2007 (UTC)
That is the point of WP:Avoid weasel words. Ie, use facts and not generalized statements. I guess Weasel words sounds insulting. I thought it was clear from context that it was a Wikipedia guideline. Weasel words can be true, there are just better ways of saying it.
Regarding Gnutella being the third biggest, that might have been true at one point. It is very difficult to count after clients run in the millions. Nobody seems to do this recently as it is too problematic to get accurate results. There is no real way to determine the size of BitTorrent networks (or all BitTorrent clients running). Also no one can measure Dark Nets, private Http and Ftp sights, and Usenet binaries newsgroups as a file sharing mechanisms. Generally, the all end up being connected, either by multi-protocol clients or by people running multiple applications. However, Gnutella is one of the major file sharing networks, but that is a moving target as FryGuy has stated. Bpringlemeir 17:10, 1 October 2007 (UTC)
I find it ridiculous that FastTrack is still supposed to be the second largest P2P file-sharing network. Is there any evidence for this at all? You don't have trust LimeWire's crawler reports. Others, for example, Daniel Stutzbach have confirmed a peer population of over 2 million peers in 2006. Papers on Gnutella --217.87.79.182 01:34, 3 October 2007 (UTC)
If you reread the opening, it says, roughly, that Gnutella has stayed about the third most popular peer-to-peer file transfer method, as the decline in popularity of FastTrack roughly coincided with the rise in BitTorrent, meaning that as it passed one, another passed it... in other words, it's still third, but instead of being behind ED2k and FastTrack, it's now behind ED2k and BitTorrent. -FrYGuY 10:09, 3 October 2007 (UTC)
I read it carefully. BitTorrent is no network. Hence it implies the relative order is unchanged. --217.87.79.182 18:50, 3 October 2007 (UTC)
Technically, each torrent is its own network. However, it's often implied that the aggregate of these networks is considered BitTorrent's "Network size". Of course, with BitTorrent being vastly harder to count than Gnutella (Which, as I mentioned before, is already incredibly hard to count), the 'measured size' of all these networks is pretty much just a pissing contest. -FrYGuY 01:22, 4 October 2007 (UTC)
Well, then each FTP or web server is one network too. It's just not very P2Py. The values are not about the amount of people using some P2P network but about the amount of concurrent peers connected to it any time on average. Comparing BitTorrent with these other networks does not make much sense as you admit yourself because it's a different concept with a different topology. A BitTorrent network is not more than a swarm for a given file, Any Gnutella peer has usually multiple of these swarms but this is not the number given here. It is about peers connected to the Gnutella network itself and nobody ever counted the population by taking swarms into account. Technically, you don't have to be connected to Gnutella at all in order to be part of a Gnutella swarm. In fact, Shareaza, for example, allows the user to disconnect from Gnutella altogether but it may still appear as uploader or downloader in a Gnutella swarm because it does not clearly separate G2 swarm from Gnutella swarms. Actually, it's a bit worse in so far that it may leech from Gnutella sources but not upload to them. So if you think swarm population should be taken into account too, you may get different numbers anyway. You claim, it's hard to count Gnutella peers. That may be true but nobody counts them. They are estimated, That's a difference. If you had looked at some of the papers, you'd see that there's little doubt about the accuracy when using the methods shown in these papers. The estimations for Gnutella are in fact the most accurate of all the networks and the only ones that you can find references for. The others are simply claimed values with no credible evidence. You also seem to assume that all networks are mutually exclusive. This is nonsense. Gnutella and BitTorrent can be considered complementary because both have different characteristics. Many of those who have used Gnutella before are still using it, just to a lesser extend and maybe only if BitTorrent fails them. Thus BitTorrent may still be on the rise and has taken a large portion of the cake but people have not abandoned Gnutella at all. "People with cars don't stop walking." While I can't prove it, I'd even claim the impact of spamming and spoofing had much larger impact on Gnutella than people "converting" to BitTorrent. What I'm trying to say is, attempt to look at the big picture and don't jump to overly simple conclusions. Wikipedia is not the place to claim, assume or speculate. If you don't have the values, don't understand the methods, just don't claim things. Also don't underestimate the amount of lies spread for commercial purposes. If some business who sells BitTorrent technology claims 50% of all traffic is BitTorrent, you have to take that with a grain or better chunk of salt. It's also funny that people recently noticed HTTP traffic being on the rise but not a single word about Gnutella which has been using HTTP ever since and which is certainly pushing more data than 5 years ago. Instead everybody assumes that's YouTube. This may be true but it's nothing but speculation. --217.87.78.15 16:12, 7 October 2007 (UTC)

[edit] Network security

I think that this there is a big part left out, namely the security issues with the Gnutella protocol. There are several scientific articles which emphasise how the Gnutella network can be used (or could, I haven't studied the latest protocol implementations) for DDoS attacks, even on hosts outside of the network. I think it is an important part.

Articles:

(Note that the last two might not be available to everyone (I get access through my university network). 130.243.240.192 (talk) 07:28, 16 May 2008 (UTC)