> > I want to first thank everyone for all the feedback I have been getting. > However, I want to also make clear some things about DistribNet. > > 1) I want the system to be designed with scalability, availability of > data, and speed in mind not absolute privacy and anonymity.
Fine, but why must you blatently ignore possibilites for privacy that don't hugely affect your desgin? As was stated before, crypto does not radically change the speed of Freenet's routing, and there is no reason to believe that it will be any different on your network. A good block cipher (like AES) is pretty quick and won't noticably decrease your efficency while doing much to increase privacy (if done properly). > > 2) I want the network to be secure against attacks which can bring the > network down but not against packet snooping. > > 3) I want to design the network so that old data rarely falls off the > network. I want to make it possible to always find data on the network, > even if it is only sitting on a far far away nodes. I believe this is > entirely possible though careful indexing and routing. > > 4) It is not that I think that always routing data though other nodes is > inefficient, I just think it inflexible. > > The main reason I am starting this project is that in my view freenet, as > it now, is way to concerned with complete anonymity. And more > importantly, it is overly concerned about preventing people from > controlling what type of data is stored in there own node as demonstrated > by the recent removal of the delete command and the fact that freenet > stores all of its data in a single file with its own property indexing > methods instead of relaying on the file system or other well established > simple database system such as Berkeley DB. IIRC, one argument for using the datastorefs instead of the native file system was that you can immediately claim the space specified. Under Fred 0.3, if you specified your data store size to be 200 MB, but later on you only have 150 MB of free space on the disk, your node would run into problems. With the datastore, you specify 200 MB now, it creates a 200 MB store on the disk and thus other files won't take up the space Fred needs. Also, you open up the possibility of encrypting the entire store with a passphrase that must be typed in every time you start the node (this isn't done right now, but it could be in the future). I'm sure there are other good reasons for using a special datastorefs. > I am also not very happy > about the fact the Freenet 0.40 was pushed on the end user way before it > was ready as evident my the many "mandatory" updates posted and the > constant need to reset the data store. Freenet 0.40 may be stable *now* > but it sure wasn't in December and January. "It seemed like a good idea at the time." I think the argument was that 0.4 became more stable than 0.3, so there was no reason to encourage people to still use 0.3. Now, 0.3 totally sucked, while 0.4's only problem was a plauge of datastore problems (which is certainly bad, but it still manages to be better than 0.3). > > Someone also pointed out GNet to me which looks promising however it is > too designed around complete anonymity. > > I have nothing against complete anonymity, it is just that I am afraid > that both Freenet and GNet or more designed around the anonymity and > privacy issues then they are around the performance and scalability > issues. > > It may be that the Freenet network is able to perform very well however, > based on what I have not seen any evidence of that so far. It certainly *should* be able to, but an ill-connected network (0.3) and datastore problems (0.4) have prevented it from showing this ability thus far. _______________________________________________ freenet-tech mailing list [EMAIL PROTECTED] http://lists.freenetproject.org/mailman/listinfo/tech
