BitTorrent v2 (2020)

Post by sitkack » Mon, Jan 10, 2022, 4:06 PM UTC

banana_giraffe wrote: post #65

sitkack wrote: post #55

Earlier quoted context omitted.

Do you make visualizations of the traffic?

Some different views. For instance, to give some sense of a scale for the 250 or so v2 torrents seen every day is, here's a view of the total number of torrents it discovers each day [1]. (Or, more properly, this is the number of torrents it gathered complete metadata and validated against the infohash each day). One of these days I should find somewhere to put all these bits of metadata I've been downloading. [1] ht…

Post by noodlesUK » Mon, Jan 10, 2022, 4:11 PM UTC

cyphar wrote: post #56

noodlesUK wrote: post #53

Earlier quoted context omitted.

Generally speaking this is called “cryptographic agility” and the security community has decided that it’s an antipattern. The alternative is to have a specified suite of security mechanisms pinned to each version. You can still have backwards compatibility, but it helps avoid downgrade attacks.

Multiformats is not really cryptographic agility, it's just adding a small tag that indicates what type of data something is (what kind of encoding, what kind of hash, what kind of binary payload, etc). The benefit is that the tags are standardised rather than being tied to a protocol version or some other out-of-band information. Whether or not you start making negotiation decisions based on that header is what dete…

Post by makeworld » Mon, Jan 10, 2022, 4:18 PM UTC

moolcool wrote: post #43

BitTorrent slaps so hard as a protocol, and DHTs are absolutely brilliant. With all of the recent hype about decentralization with the rise of crypto, I'm surprised BitTorrent hasn't seen much hype or revitalization as well.

Post by aumerle » Mon, Jan 10, 2022, 4:21 PM UTC

marcan_42 wrote: post #21

elcomet wrote: post #16

Earlier quoted context omitted.

unlikely ? all methods are broken eventualy when compute becomes faster

This is not true. Modern cryptography is not breakable via brute force any more, nor will it ever be. You can prove that the amount of energy required assuming a thermodynamically ideal computer is more than will ever be available in our galaxy, with classical computing. Breaks of symmetric crypto or hashes that are practical require an actual cryptographic weakness, whose existence is not a given. Then there's quant…

Post by j16sdiz » Mon, Jan 10, 2022, 4:24 PM UTC

chasil wrote: post #68

xd1936 wrote: post #37

Earlier quoted context omitted.

Unless I'm mistaken, we'd need to settle on a "default" piece size[1] if we were to have automatic upgrading in a client. I learned recently[2] that there's no standard set by the protocol, so Client A might choose a 16KiB piece size, and Client B might take the same files and choose a 64KiB piece size, resulting in two different hashes (v1 or v2!) and thus two peer pools for the same file set. 1. https://stackoverfl…

I'm not sure if the the piece side continues to have bearing when the hashes are all in a Merkle tree on uniform 16k blocks. "Enforcing these encodings also help make it more likely that two people creating a torrent from the same files, end up with the same info-hash and the same torrent." In the discussion of backwards compatibility, it appears that v1 collections of identical content encoded with different piece s…