On 14 Oct 2005, at 15:06, jrandom at i2p.net wrote: >> As such, it is rather disingenuous to argue that I2P is not >> harvestable just because a small subset of the peers in the network >> can remain hidden, the vast majority of the peers in I2P will and >> indeed, must remain harvestable for the network to work. > > That is a neat characteristic, and that does help clarify Freenet's > position - that it is striving to offer a hidden network that will > operate on small to medium scales in countries disconnected from > the Internet, or in a post-apocolyptic world where there is no more > West.
I wish I shared your optimism about western governments level of tolerance for systems like Freenet or I2P, but I don't. Already in the UK British Telecom is blocking websites based on content, initially those that contain child porn, but I don't doubt that it will expand from there. If Freenet or I2P was to become successful, it won't take long for them to block known I2P and Freenet nodes, and this will be much much easier than the kind of heuristic traffic analysis that would be needed to detect Freenet 0.7.0. Even if they were to resort to such traffic analysis, there are plenty of things we can do to make Freenet 0.7 harder to detect. In contrast, I2P will have no recourse other than a fundamental redesign of the network, you would find yourself in an architectural dead-end. > For everything else we've been discussing, I2P suffices, and > (arguably) performs better. I2P today may well perform better than Freenet 0.5 does today - although to some extent its an apples and oranges comparison as we offer different functionality, but I think Freenet 0.7 will offer functionality, such as realtime efficient broadcast, that could well leave I2P in the dust. I'm sure you disagree, time will tell who was right. > But performance and efficiency is > something I'm sure we'll discuss more about later, as I know you > disagree. In any case, its 'cheaper', as it already exists, and > additional things that a censorship resistant content distribution > network needs could be provided by Freenet, simply reusing the > anonymous communication layer offered by I2P. We had that debate, and we considered it carefully, and the general conclusion of Freenet's developers is that it would not be sensible to implement Freenet on top of I2P. I won't reopen that debate, but just for one simple example, an important feature of Freenet 0.7 will be that it doesn't require NATs to be reconfigured because it uses NAT traversal, from what I can tell from your website, I2P doesn't support that. More fundamentally, Freenet provides functionality similar to that of a DHT, how can it make sense to build on a platform that already contains a DHT? We want to explore a different part of the design space. > Competition is fine, but cooperation better. Agreed, where it makes sense. Just having these debates is co- operation of a sort, since it forces us to respond to a critical perspective on what we each of us is doing. Code resuse also makes sense on some occasions, and we do make use of a significant amount of third-party code, but only when it supports our intended functionality, not undermines it. Ian.
