Summary of conclusions from a conversation with jrandom: 1. We should eventually use i2p's low level transport code (JMX) 2. They have a GMP wrapper that we could use NOW to speed up asymmetric crypto (bigint modPow especially). We should use it! We can include both the class and the linux .so and the win32 .dll in the jar, and put that into freenet-ext.jar: http://www.i2p.net/node/view/147
This would be useful even if we get freenet running on Kaffe as IIRC Kaffe doesn't run well on Windows, if at all. Look what we get when we cooperate :). Free software lives! BTW what happened to bundling the onion native code in freenet-ext.jar? 3. The big prize is not domination of the p2p "market". It is building something that will actually work in REALLY hostile regimes. I personally think we have an advantage in that, IF we can get routing and specialization to work well: An overall DHT is very vulnerable, because it can be spidered. This is a fundamental flaw in I2P even if you use wierd transports. Freenet has a similar fundamental flaw: reference harvesting. However we can jettison this once routing works well, at the cost of some performance. Hopefully our users will be interested in developing, testing, and funding this sort of tech post 1.0. I2P on the other hand would have to develop a whole new routing algorithm to remove this problem. I appreciate that this is science fiction at the moment because freenet routing does NOT work well, but its problems come down to: a) Plain old bugs. Many of them relatively simple. Many of them have MAJOR impact on performance. That is my main focus at the moment. Hopefully another month or two debugging and we'll have a reasonably solid 0.6 to put out. b) Major design flaws in routing and other areas. I believe these are solvable, and I will certainly try to solve them! We have some idea where some of them are: * The relationship between rate limiting and routing. Various ideas to solve this, hopefully after 0.6. The big prize here is for routing to actually work, even with very large amounts of content. * The NG Routing implementation is almost certainly woefully inaccurate, and probably has other difficulties. Ian had an idea for an alternative, there may be others. In both cases, major progress may well be contingent on some systematic testing framework - simulations MIGHT help, but IMHO the test network will probably be the way forward. This can all be dealt with post 0.6, hopefully. On Mon, May 24, 2004 at 06:03:13PM +0100, Toad wrote: > I think we need to talk to the I2P people, and to use some of their > technology now (IRC over I2P), and possibly some in Freenet 0.7 (for > premix routing). Or at least we should look into it. The technical > rationale: > > 1. We need an anonymous chat mechanism to talk to content authors, > arguably. IRC over I2P is the obvious choice. > 2. IMNSHO, we need premix routing before 1.0. Otherwise users who > actually use Freenet (mainly with large sites and splitfiles) are > vulnerable to all kinds of correlation attacks. We may well be able to > borrow some of I2P's code to do this. > 3. I2P's alternate transports infrastructure may be useful eventually. > As would the possibility of having I2P traffic indistinguishable from > Freenet traffic to an attacker. > Technical issues: > 1. Interface: > We can simply look at their code and reproduce it, or copy some of it, > since they are GPL. However we would then have to maintain it. I think > we may want to adapt I2P so that it can provide the level of > functionality we need, and then bundle it and run over it. > 2. Node discovery: > They use a DHT for discovering routes to nodes. We would probably want > to ignore that and do our own. We might have security issues with having > more (easier?) ways to harvest node references? I don't know exactly how > it works, perhaps it's not as bad as it looks. Our own harvesting > vulnerability is pretty damning. > > Political rationale and comments on DHT over I2P: > 1. It would be useful to be able to talk to content authors and so on > in real time. This is not such an issue since there's a gateway, but it > would be better to be able to connect directly. > 2. We may well be able to use some tech from I2P; they may be interested > in some of ours (e.g. anonymity filter). > 3. I2P with a storage DHT, which is being proposed, could conceivably > obsolete Freenet. However: > a) It won't necessarily be faster and better than Freenet. Each hop > would be onion routed, which is unnecessary (and not very useful) on > Freenet. There is no need for this; the only reason would be to protect > node owners if the DHT were illegal, but that would be pointless as I2P > would probably also be illegal. > b) Scaling the DHT, especially if it stores content, may not be as easy > as it sounds. According to Ian a lot of money has gone into such things > and often to no result. > c) More importantly, I2P is not the enemy. If you agree and find such > arguments boring, skip this paragraph and the next. The forces of > censorship are the enemy. In Monty Python's Life of Brian, set in 1st > century Palestine, there are two revolutionary groups: the People's Front > of Judea and the Judean People's Front. Both hate the Romans, but hate > each other even more. They both try to kidnap the consul's wife, and their > ineptitude and disunity helps the Romans greatly in arresting them. This > happens constantly in many areas of life (marxist groups are a good real > life example). What we do here is of some use. Right now Freenet (and > other software) can be used (and has been used) in China, and many other > places we wouldn't want to live. It may not be perfect but it's a hell of > a lot better than the obvious alternatives. Even in the West, right now, > web sites whose operators are constantly harrassed on the Web are > preserved on Freenet; the various Scientology sites are a good example, > as are the Diebold files. > > Personally I am of the opinion that we are looking at a major > struggle that will be won or lost by this generation. The overall battle > is of course the free software versus DRM thing. It comes down to > Microsoft, the MPAA, and software patents on one side, versus the Free > Software community and the several large corporations that support it on > the other side (their allegiance is very dubious in many cases). Admittedly > there is a lot of money on both sides now; however, we can still lose. If > we lose, IBM loses a few billion and has to port AIX to DRM. If we win, > Microsoft's share price collapses and it takes a lot of the US economy > (20% of the S&P index last I heard; a lot of pension funds are linked to > S&P) with it. How is freenet linked to this? Well, firstly, freenet is > Free Software. Secondly, until all PCs are *completely* locked down > (which will require the ability for the corporations to get a court > order to tell the DRM authority to remove all files of a given hash from > all PCs anywhere), Freenet and I2P will still be viable options for > online free speech (even once they have been banned). Imperfect though > they may be, they are a start. Both in the West and in places like Saudi > Arabia or China, where the current options are very limited, and the > authorities' enforcement efforts so far have been technically clumsy and > inept. > d) Freenet users benefit from us being close to I2P, whether we win or > lose, IF there is a clear technical rationale. This would have to be > investigated carefully; we may be able to use the I2P message > transports; we may be able to use their onion routing capabilities; it > may turn out that it's not appropriate to use it at all. > e) If Freenet ultimately turns out to be inferior for all conceivable > applications, then being close to I2P will be beneficial for Freenet > contributors. However for the above reasons, I don't think that Freenet > will necessarily be superseded. If we are not, then we have nothing to > lose by using some I2P technology. > f) It would probably be useful at some point to review the I2P > architecture and their statements on attacks; there are issues that can > be dealt with in Freenet. > g) My role in all this: I am paid by the Freenet Project non-profit > corporation. I therefore have a vested interest in getting Freenet to > work. If I2P undermines Freenet before we can get a reasonably good 1.0 > out, my future employment prospects are greatly reduced. But if we work > together, this is less likely, and will be less damaging if it DOES > happen - both to me and to our users. Finally, if and when we lose the > Copyright War (metaphorically), I may have to switch career. :) -- Matthew J Toseland - [EMAIL PROTECTED] Freenet Project Official Codemonkey - http://freenetproject.org/ ICTHUS - Nothing is impossible. Our Boss says so.
signature.asc
Description: Digital signature
_______________________________________________ chat mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.general
