Danse Macabre of Fasttrack - nominee for longest post competition (sorry)

[Smoketoomuch]

I thought and thought about Fasttrack and Morpheus, and here is what I came up with. Some points and statements here are open for debate, I would like to see what do you think about these problems.

Ok, here is a very promising statement from CEO.
„Lastly, we want to address some of the misinformation we've seen recently. There have been many comments that we caused these problems intentionally. Let me assure you that we would NEVER treat the Morpheus users in this fashion. Others have said we would re-launch with a paid subscription model, again, not true. Our commitment is to always provide you a free version of the Morpheus software product. Thanks again for making Morpheus the most widely downloaded p2p product in the world. We are hard at work to provide you with the products you have come to expect from our company. From our perspective p2p means people to people.”

Paid subscription model not true, but notice that there will alwayb >a< free version of Morph. No specific detail again on that one, and this can mean various things:
a) there will be a subscritption modell also, which means that must be a lot better then the free version – ergo free version will be a stripped down version or
b) there won’t be subscription, only licence fee that removes adds – it seems to me very unlikely – a small licence fee is agreeable to pay for no ad, but subsciption is not, and is agreeable only if the prog knows a lot more then the free version.

But then, my problems are much older then that. First and foremost, I saw not many posts about this, but the fact is, that connected to 450000 users sharing 642313 files or whatever is simply misleading. You are never connected to that many users. U are connected to a small segment of Fasttrack network, and multiple searches for the same stuff won’t let you out from that segment. As borgdrone1of2 (I hope it was him, if you remember differently, please correct me) said on old forum> there is a good chance you download from someone from another continent (which is true, since most people I download from is from US) but a very small chance that you would be able to see the shared files of the guy living across the street. Which means that from our (the user’s) point of view the “connectedness” (sorry) was never as good as many would like to think (or would like us to think). My estimate is, based on availability (music, soft, vids), and even taking into consideration the leech problem, which is quite serious on fasttrack, the fasttrack network you are actually connected to is no bigger then winmx connected to frontcode and 2-3 high capacity (like 10000 users or above) network – but if you disregard leeches, and base your estimate only on availability, then the size roughly equals winmx connected just to the frontcode server.

Technology

Expectations flew high back in november/december, when morph 2.0 was announced. Didn’t think of it then, but since everything that technologically is considered advanced (file resume, auto search, multiple source download) was already there. I honestly don’t know what we expected of new version back then. Leech control was a priority issue, but that is not a big step forward technologically speaking. The other expectation was gnutella. This still poses a problem for me. If you search back on Zeropaid for old articles, you can see that developers of Gnutella clients are not always on friendly terms, they want to make as much money as they can (exept a few like XoloX) – so the question is: what is their gain. This trails several question, and I hope you can answer some.
First, if Streamcast makes a new client that can use both fasttrack and gnutella network resources than the first problem is:
I didn’t see announcment from current gnutella client developers that promise access to Fasttrack network in there new version. This means, that if this is not a two way street, then those using new morph (if it does what it promises) become virtually leeches of gnutella client users. This means, that one has to be either stupid or fanatically attached to a LimeWire or XoloX to keep using it, but the majority will use new client from MusicCity. This would mean that for Bearshare, Limewire, whatever, its over. But this is hardly believable – aren’t they independent companies that keep their own interests in their sight? IF it would be a two way street, what would motivate one to use MC product, especially if it would be a stripped down free version, where s/he can use Limewire or any other next generation gnutella client? So what will happen to independent (independent from MC) gnutella client developers? If two way street: will they merge in one megacompany with MC or what? If not, how MC paid service would be successful?

Just as a reminder: only such drastic changes would warrant a version change from 1.3.3 to 2.0, that is to say, incorporation of gnutella network or a radically new technology, that would be as great jump as the switch from single source to multiple source download capability. If there is such technology, then what would it be? Some clue might be gained from Napho’s thread about Rebol > http://www.napsterites.net/undergrou...&threadid=9606

But then how would that new techno look like from our point of view. As far as technology goes, the most advanced p2p techno is used by eDonkey. The only addition to multiple source download is the ability to download incomplete parts, which is done by the donkey, and is a huge advantage. 80% of a large mov. I got from people who are still in the process of downloading is. At the end of the post you see an example for that. You see multiple sources for same file, black meaning in the colored bar the parts that a particular user has. For example if there are 3 people having full stuff, all that download that particular file from them is a source for you. It was funny to see once that there was someone with whom we got the file at the same time. Sometimes he (or she) managed to download a part, transfer immediately began from his puter till it was on „no needed parts” then when I got some parts from someone else, he downloaded those, then me again when he got new parts, and so and so. (oh, the pic is a freshly started donkey with few sources, but it should suffice – u can see – ok, ok, with terrible slow speed – but I’m actually downloading from someone who doesn’t have the complete file). You may have noticed that I became a big fan of donkey, but my only point here is to see which is the most advanced p2p prog today – that has something more in addition to multiple source dl, and the donkey is one – and excellent leech control, which, actually can’t be called forced sharing either, since it is a technology (that is the file u are downloading can be downloaded from u) that is simply makes file sharing more effective.

My point is, that except incorporation of another network (that is making it bigger) only technological improvements like adding an ability similar to donkey warrants a new version. I don’t say that only useful technology is improvement, or rather, I don’t mean useful in a narrow sense. That is to say, donkey needs servers – its decentralized, since you can download server version and make a server, but technologically it is more advanced if you don’t need a server at all (btw, Swamp is rumored to working on a new network that needs servers no more, but no one knows how he wants to do it…), since it grants a protection against lawsuits as you can claim there is nothing you can shut down. We just witnessed the collapse of this myth – it is not only possible to shut down fasttrack, but it is also possible to do it selectively, moreover, it is something that can be used by RIAA as well, since it works like a virus (you just need the ability to spread false supernode lists on the network). For further details on this read>http://www.zeropaid.com/news/article.../03012002b.php Its a long article, but I recommend everyone to read it, especially since the newest news on ZP is that it seems to be confirmed…(„On February 28, 2002 an anonymous post on Zeropaid set in motion a HUGE amount of discussion and speculation as to what had exactly happened between Morpheus and KaZaA. The post, seeming to come directly from an episode of the X-files, outlined a "paper" trail to a mysterious company called Sharman Networks. There were good guys and bad guys, back room deals alluded to with a previously UNMENTIONED company called Brilliant Digital, the plot definitely thickened. While some readers scoffed at the perceived paranoia, other followed the trail. Jorge, a regular contributor, continued the framing of the issue and let Zeropaid members see that maybe there was something sneaky going on in the world of P2P.”)

Something sneaky… Well, just think of what I wrote about gnutella/morpheus compatibility…


Huh. At the end of this extremely long winded post, I could continue but I’m tired of it. There are a lot of unanswered questions, please let me know if you have any… (I know Snow Man has a few). I have a few also (ok, I can’t stop this, sorry…). One is about Dat format – those who know more about it (AYB for instance) is there a way to hash those dat files in a way to make parts of it intelligible for the network? If not, and it needs an entirely different technology, then MC can’t implement a feature like that of the donkey except if they shut out every client using the old dat format. Another is concerned with privacy: was I correct (I don’t know, I’m quite a lamer in this respect) when I assumed reading AYBs attachment that it contains the IP addresses of those who had those files? Even if donkey is not that decentralized, still it has no such datas in their incomplete downloads. Ok, before I go „hey old Morpheus comrades, I think its time to move on onto something new and based on our high expectations, donkey is the way”, I shall finish this post, waiting for tons of ideas and questions I am too tired to ask (but this whole issue has the nature of a fractal, an endless vertigo of questions, hence the long post). Ok, I’m finished. For now. Please continue.

Best Wishes

ps. - no spell check, no check at all - terribly sorry - s2m in big hurry and writing like in frenzy.

[Smoketoomuch]

oh, the picture:

[TankGirl]

An interesting post, Smoke!

Originally posted by Smoketoomuch
But then, my problems are much older then that. First and foremost, I saw not many posts about this, but the fact is, that connected to 450000 users sharing 642313 files or whatever is simply misleading. You are never connected to that many users. U are connected to a small segment of Fasttrack network, and multiple searches for the same stuff won’t let you out from that segment. As borgdrone1of2 (I hope it was him, if you remember differently, please correct me) said on old forum> there is a good chance you download from someone from another continent (which is true, since most people I download from is from US) but a very small chance that you would be able to see the shared files of the guy living across the street. Which means that from our (the user’s) point of view the “connectedness” (sorry) was never as good as many would like to think (or would like us to think). My estimate is, based on availability (music, soft, vids), and even taking into consideration the leech problem, which is quite serious on fasttrack, the fasttrack network you are actually connected to is no bigger then winmx connected to frontcode and 2-3 high capacity (like 10000 users or above) network – but if you disregard leeches, and base your estimate only on availability, then the size roughly equals winmx connected just to the frontcode server.

Good technical point. None of the decentralized clients/networks has yet managed to tackle the global connectivity problem satisfactorily. Whether on Gnutella, FastTrack or WPNP the users end up to localized 'pockets' where they can see some other users and their stuff but miss the great majority of users and whatever treasures they share. At its best Napster managed to provide a full connectivity (with search, browse and hotlist services) to a million people or so - bringing the whole mp3 sharing world into same virtual space and thereby making a lot of rare stuff available. The number of p2p users and the average size of their libraries have multiplied since and doing the trick in a centralized way á la Napster might already be unfeasible. It needs to be done in a decentralized way and I also trust it will be done so within a year or two.

- tg

[JackSpratts]

Wow Smoketoomuch, what a post! And to think that was the fast version! Ok. Here's a couple of points about some things you addressed.

Yes, the node problem is a major weakness in that most searches (almost all actually) don't go much beyond the local supernode for reasons of network stability. This was to be addressed in later versions, which brings me to your comments about technology. There was, and is, huge room for improvement in almost every area of the Fasttrack system, from the client to the network. From bandwidth management to shutdown protocols, supernode voting and social issues, there is much that needs to be done to make the program more powerful and easier to use. Some of these changes that were to have taken place in v2 included improvements to scalability, chat and hotlisting. Other, more esoteric social functions would be added like neighborhoods (all Rebols' department) and then of course, two-way connectedness with Gnutella. A lot of little things were to be fixed like user/file estimating (always way off) and GUI improvements too (remembering settings, not losing info bars, adding color and so forth). As time went on things like the port problems would be looked at with the goal of dropping reliance on specific ports altogether (1214 issues). When one of your biggest constituents - college students - are forced to use a different client, you've got major problems right there. I'm not sure about downloading from partials but looking at the way the system lays out, you'd think it was designed with that in mind. I may have heard something about it but I can't say for certain.

There are of course other, extremely important things a p2p can have, and should have in my eyes. First and foremost obviously, a network can't survive todays' hostile legal climate unless it's truly decentralized, and that needs to be carved in granite. Next and almost as important is true IP address cloaking and security against attacks and virus protection. I could go on for pages about the weaknesses in various systems but why make it easy on Hillary and I'm out the door besides!

The day is long in the future when file sharing technology will be improved “beyond improvement.” And that's frustrating sure but really to me, that's what makes it so cool.

- js.

[Smoketoomuch]

s2m pretty messed up, messed up indeed. I have this damn tendency of self reflexive thoughts. So I'm I see this post as an attempt to sublimate the libido I used to cathect my object of desire, the energy that ran the risk on turning back upon my own self - resulting in anxiety. What is interesting is that I'm not sure if I'm in at all, not at all. So what I do is post long topics and off topic remarks.

Both of your replies were enlightening indeed - cleared up some issues like scalability, node problems, the complexity of the question of technology. An interesting solution for the problem is the extended search feature of donkey, which is automatized by bot. Bot has been updated 10x in last month, resolving the issues of overloading the network. They seem to have a different approach to the problem. The global search feature can be adjusted to make a thourough or fast search, but now it is also made to stop at 100 hit to avoid overloading servers with requests. This appears to me as a good solution for the problem, since it makes sure that your search goes trough the entire network without slowing it down. Now I have 28 sources for the same download, but it aqcuired those sources gradually, and every time its on looking mode it doesn't search those servers that it already had - that's how it manages availability of multiple sources. Still remains the problem of silence about two-way (two-way was my clue regarding my new understanding of my post) connection on the part of gnutella developers.

Other thing is that I have to write about p2p in very basic terms for Vulgo (a literary and philosophical journal). That is to say, the issue of decentralization (if I count p2p as one of my addictions, then there should be at least something productive coming out of it). It appears to me that it is the material manifestation of a tendency that gained speed in second half of last century both on the fields of science and economics and the field of philosophy. (in the science I have Mandelbrot in mind and Chaos theory, Richard Gleick wrote a good book about it). In philosophy, there is an approach to language, the self, reality that parallels this development spearheaded by french philospohers. I always thought that language partly determines the way you think. There are bilingual scientists (I have a french matematician in mind but don't remember the name) who think different aspects of the same problem in different languages, as if it would be easier to resolve practical questions in english, built a structure, a hypothesis in german, and see what's wrong with it in french. I always though of french language and words as having no body - I mean, especially the spoken form, is so fluid, they have a tendency to chop the margins of every word - so no wonder that everything important in phil. in 2nd half of last sentury came from french thinkers (Derrida, Foucault, Baudrillard, Lacan for instance). I mean problematization of hierarchical structures, canonization of works, the transcendental signified, absolute truth and objectivity, and so on. However, german language is like bricks, you can built large palaces with it. English is somewhere between. I just mention this because there is something genuinly french - if such analogy between internet development and especially the tendency of decentralization and language can be drawn - in the way donkey comes up with solutions to problems. Okey, this does not rest on firm grounds, my theory of seeing recent developments in communication, copyright issues and authorship and authority as one manifestation of "deconstruction" at work is more sure. I'm damn lazy to write it actually, especially the intro to p2p, but that is no wonder considering the readers of that journal - mainly philosophers and literature lovers, many of whom I know that are barely accustomed with email and are reluctant to use the internet, seeing it as something ezoteric. Ok, rant over, won't be online for a few days probably, sorting this question of love or not to love out. Or maybe not. Because I already thought trough it, which left me only one option - there is no option but to wait how things turn out. That can be a very long time, but then, this frenzy (like all frenzies) cannot be kept up for long - probably I need sleep, that's why I was in hurry, but if I come to think of it, its just the stupidest thing I ever come up with - being overexited and writing like a machine to be able to get to bed - in the same state??? Come on s2m, how can one sleep in such a state. Nonetheless, I'll give it a try. I think. See ya.

ps. no, I don't smoke that much, it just kept accumulating and it spilled out. Not only here btw, I feel pity for my enviroment sometimes.

[pod]

Originally posted by Smoketoomuch
You may have noticed that I became a big fan of donkey, but my only point here is to see which is the most advanced p2p prog today – that has something more in addition to multiple source dl, and the donkey is one – and excellent leech control, which, actually can’t be called forced sharing either, since it is a technology (that is the file u are downloading can be downloaded from u) that is simply makes file sharing more effective.

I've been saying this way back in the MC forums... Morpheus had NOTHING on eDonkey when it came to multiple source dowloads. It's beautiful how it works. And no leech control, though your effective download speed is largely determined by how fast you're uploading and how much you're sharing, which is pretty slow if you're just leeching.

WinMX promises to match eDonkey's multi-source, I've noticed my incomplete downloads being utilized, but not sure if this works WHILE downloading, or only when your download stops. Warrants some further testing at the least.