832 entries in 0.516s
![](/log-search-day-hist.png?q=gossipd)
mircea_popescu: and coupled with
gossipd, it becomes impossible to establish "whose chopper this is". owner reports his choppers stolen, good luck "enforcing".
mircea_popescu: as people (alf, mostly) pointed out in the intervening coupla years, there's absolutely no sane reason to marry
gossipd to extant-internet, or tcp/ip or etc, as the original draft was trying to
mircea_popescu: but honestly,
gossipd as a finished product can not possibly come other than with a mesh-over-wifi and mesh-over-radio ready made.
Framedragger: entirely unrelated: wouldn't it be somethin' to test out some
gossipd ideas over *actual* cheap long wave or whatever wave radio devices between these here people?
mircea_popescu: trinque i know i brought this on myself, but an unsubscribe will be useful so i don't get everything on
gossipd spec twice lol
mircea_popescu: moreover nothing prevents emergent/de facto specs. the job of the
gossipd spec is to keep people from doing world-breaking idiotic stuff. how they handle their own wives is their own problem.
scriba: Logged on 2016-09-12: [11:08:04] <Framedragger> mircea_popescu, asciilifeform: regarding
gossipd, aside from the central point of disagreement, regarding a "lighter" matter: what about subscribing/unsubscribing to "topics" (a kind of pubsub model)? because there's no discussion of multiparty chat as of now; or is there not to be, in
gossipd?
a111: Logged on 2016-09-12 11:08 Framedragger: mircea_popescu, asciilifeform: regarding
gossipd, aside from the central point of disagreement, regarding a "lighter" matter: what about subscribing/unsubscribing to "topics" (a kind of pubsub model)? because there's no discussion of multiparty chat as of now; or is there not to be, in
gossipd?
Framedragger: topic-based publish/subscribe has been sorta well researched, but i guess this problem is on another 'layer':
gossipd document would leave this for 'implementation'. even though it may not be trivial at all, to make decisions regarding such matters, choose best spec, or design it from ground zero. but of course makes sense to discuss the foundations first
Framedragger: mircea_popescu, asciilifeform: regarding
gossipd, aside from the central point of disagreement, regarding a "lighter" matter: what about subscribing/unsubscribing to "topics" (a kind of pubsub model)? because there's no discussion of multiparty chat as of now; or is there not to be, in
gossipd?
☟︎ mircea_popescu: of either kind in his production ; nevertheless he also claims dc is sufficient to sink any attempt to implemnet a
gossipd.
mircea_popescu: the matter in dispute is how to handle authentification. one solution is decrypt-challenge. the other solution is signed-hello. i reject signed-hello for a number of reasons, both practical and theoretical, which alf doesn't seem to think much of but which i am satisfied are sufficient to sink any attempt to implement a
gossipd on sh. alf purports to reject dc on similar grounds, except i can't find anything like an argument
trinque: more redundancy the better. these facilities end up being what's available on a particular
gossipd node, as I see it
jurov: but ben_vulpes is proll going for remote activation by rsa sig via
gossipd mircea_popescu: this is necessarily not
gossipd / bitcoin / p2p world.
a111: Logged on 2016-09-09 14:51 mod6: <+asciilifeform> mod6: the correct pill is to decouple (at least from enemy's pov) gossip station keys from royal keys. << so to clairify this a bit, at least for myself, the
gossipd node that I run and operate would verify transmissions sent to it with mod6-battlestation-key, but when I would actually go to send something out it would sign with mod6 "royal" or "personal" key?
a111: Logged on 2016-09-09 14:36 mod6: someday, when we have a tmsr cryptolib & rsa-o-matic, maybe we can starting building a
gossipd prototype.
phf: we had a thread about it two weeks ago, where the conclusion was that
gossipd as written in the only available spec has all kinds of problems and shouldn't be implemented/used
mod6: more than anything, i just want to be able to contribue a bit better to these
gossipd discussions - which will be a vital part of the infrastructure.
mircea_popescu: come to think of it, is there such a thing as a
gossipd spec even ? what's the current canonical version ?
mod6: i don't want to disagree with what I just said here: <+mod6> a lot of new questions and discussions will arise once someone starts to prototype
gossipd out.
mod6: <+asciilifeform> mod6: the correct pill is to decouple (at least from enemy's pov) gossip station keys from royal keys. << so to clairify this a bit, at least for myself, the
gossipd node that I run and operate would verify transmissions sent to it with mod6-battlestation-key, but when I would actually go to send something out it would sign with mod6 "royal" or "personal" key?
☟︎ mod6: a lot of new questions and discussions will arise once someone starts to prototype
gossipd out.
mod6: someday, when we have a tmsr cryptolib & rsa-o-matic, maybe we can starting building a
gossipd prototype.
☟︎ mod6: interesting convo this am re:
gossipd & v
Framedragger: asciilifeform: that's what i'm worried about, you may have to burn IP, too. and i'm all up for mesh networks and post nuclear radio, but kinda sucks that the whole internet backbone may be incompatible with proper
gossipd, gotta admit.
Framedragger: and
gossipd without any auth whatsoever wouldn't really be that? in all honesty, i should reread the spec, which is probably outdated, and log search sucks, fml
Framedragger: mircea_popescu: how about i (an nsa employee, say) just make a filter which grabs all observed
gossipd traffic (packet timing or w/e, and if it's an actually new transmission protocol, then supreme joy is me) and send it for further analysis. i shall assume that while it's not certain which messages are legit and which are not, the offending t3rr0rist group is too lazy to transmit proper false positives to provide noise,
Framedragger: (i am aware that proper
gossipd doesn't have to run over internet)
Framedragger: btw wouldn't "nothing signed"
gossipd reality actually be not "only among chosen clique" but rather "only chosen among clique [so, okay, not for all] plus whoever listens to internet backbone including all teh agencies"?
mircea_popescu:
gossipd does "for friends ears only". and the only way to make sure that can't happen, si by signing things. because once signed, they're definitionally for everyone equally well.
PeterL: what keeps you from moving a signed message thorugh
gossipd?
mircea_popescu: but all this stuff aside, back to the important point here : the "
gossipd-like" thing contemplated for moving signed material (ie, v stuff) around is very much a different beast from the actual
gossipd, which doesn't work on signed material ; presumably doesn't work on tcp etc.
mircea_popescu: if we have to leave freenode before
gossipd is ready we'll reimplement ircd.
mircea_popescu: what's more important is that
gossipd works like otp : "sure you can decrypt this text mr evil. WHAT would you like it to decrypt to ?"
mircea_popescu: it quickly becomes dubious. because a signature permits a certain relationship. whereas
gossipd rests on dubious relationships.
mircea_popescu: i'll just continue using my current system, because such a
gossipd would be an outrageous degradation of everything.
mircea_popescu: a
gossipd that requires me to sign anything is not to any degree interesting.
mircea_popescu: this is not only a ridiculous hope, but contrary to design principle. the point of
gossipd is that he sees.
mircea_popescu: well, the principal difference between
gossipd-for-v and
gossipd-proper is that v operates on signed matter exclusively ;
gossipd-proper operates on unsigned matter exclusively.
mircea_popescu: certainly it is not contemplated here thatg anyone would be forced to run v to get
gossipd. heck, i don't even believe such "forcing" would be in principle possible.
mircea_popescu: anyway. this "separation of concerns" objection would have a lot more meat on its imaginary bones if there actually eixsted any sort of
gossipd. as there doesn't, the objection has no legs to stand on.
mircea_popescu: yeah, as contemplated in
gossipd there is no difficulty. but mining, still exists.
mircea_popescu: in fact, mining is included in the
gossipd implementation of privacy ; re-read the example given there to phf.
mircea_popescu: for... all implementations ? like, it happened to you so now
gossipd-central updates my
gossipd ?
mircea_popescu: asciilifeform uh. so now you're married to hardware ? why even do it software then, make
gossipd a chip
a111: Logged on 2016-08-26 13:04 phf: i don't think there's a solution to drowned in sibyl's in general. there's a cost to validating counterparty (which is continuous in case of
gossipd, there's no "validate the ip, and then trust it" which is what i mean by "no trust in ip"), which can be exploited by attacker.
phf: i don't think there's a solution to drowned in sibyl's in general. there's a cost to validating counterparty (which is continuous in case of
gossipd, there's no "validate the ip, and then trust it" which is what i mean by "no trust in ip"), which can be exploited by attacker.
☟︎ phf: asciilifeform's objections was that
gossipd relies on ip addresses at all
adlai: so when do we officially acknowledge that the first attempt at addressing the
gossipd RFC was a failure, and solicit a second?
trinque sees how wot ratings propagating over
gossipd make more sense than anything taped together on this
gossipd-less side of things
trinque: aha, I'm not gonna bolt some sweet shit on the side of deedbot til
gossipd trinque: but then, asciilifeform could
gossipd from his water-tomb I'm sure, so there's that.
trinque: thing's written such that it would hop atop
gossipd and provide commands just as easily
Framedragger: but it was nice to understand where "gossip" comes from in
gossipd: it's all about *not* providing any authenticity of source (beyond "this is coming from my wot")
trinque: and hell, you want a proper blog^H^H^H^H command line interface to a bunch of articles gimmeh teh
gossipd mircea_popescu: anyway. uci is not
gossipd. there is a good reason to implement
gossipd OVER uci, but on the uci layer the thing shouldn't be concerned with this kind of security i don't thinl