log☇︎
273900+ entries in 0.172s
diana_coman: sounds ok to me at this stage, yes
mircea_popescu: (in this discussion, irc is gossipd 0.1, obviously)
mircea_popescu: with the possible exception of pm, which should be a separate function, and point-to-point. (ie, client encrypts to dest's key, not to server's key)
mircea_popescu: not a serious problem, i really intend to simply include a chat client. didn't mention this in the fls, but... all comms should be over irc anyway.
diana_coman: yes, this is without chat stuff indeed
mircea_popescu: diana_coman this of course includes ~no chat chatter.
mircea_popescu: ok. so the reasonable move here would be to spec this to transport 10kbps from client to server ; in chunks no longer than 2048 and no shorter than 32.
diana_coman: hm, kind of hard to say before having done at least some sort of concrete start on this; at this moment I'd say half more as a guess than an estimate
mircea_popescu: and by the revision we're doing we hope to cut this... in half ? by 90% ?
diana_coman: aha; might be worth mentioning too that average can be dubious, since you'll get more likely spikes when lots of things happen where you are and otherwise some very low regular exchanges
mircea_popescu: meh wait, i did the 8 conversion the wrong way. 50kbps ~= 10 messages / s if they average 600bytes each
mircea_popescu: so for an average message size of say 600 bytes, the server on average sends 800 messages to the client each second, and the client back about 1-200. that sound right ?
diana_coman: from the server to the client mainly; the other way around is much less, below 10
mircea_popescu: so what'd typical client do ? 10kbps ? 1 ? 100 ?
diana_coman: mircea_popescu, asciilifeform currently eulora messages exchanged between client and server are between 43 and 2048 bytes (really upper limit and rarely seen); this being said, the current system of messages/ exchanges between eulora client and server is quite a mess from many points of view, so it's set to be changed too
mircea_popescu: and then that "trash the dress" bullshit. looks exactly like a donna dolore set for the first two minutes, then it turns out the director isn't present and shit goes downhill from there.
mircea_popescu: "if i can't comfortably lift the girl in the air, maybe it's time to work out! or maybe it's time to find another girl!" thought no groom ever. "if my wedding threatens to look exactly like the wedding of every other bourgeois idiot, maybe it's time to take off the dress" thought every bride ever, AND THEN CHICKENED OUT ON IT. etc etc.
mircea_popescu: there's only so many ways you can anchor gauze in a tree so it appears windblown.
mircea_popescu: all of them are ooooobviously "speshul". except... all of them are obviously the same fucking thing, the fuck are you gonna do out of a dork who doesn't own suits and a rando girl in a white gauze marshmellow ? within an acceptable budget ?
mircea_popescu: se daddy's artistique. hopefully the wife fucks around with a non-beta. but anyway :
mircea_popescu: possibly one of the more depressing activities available in contemporaneity is checking out the portofolio of one of those "artist" photographers that in practice survives out of taking wedding pictures. i don't mean the friend-of-the-bride etsy idiot that DOESN'T survive. i mean the dude who keeps a steady stream of work enough to feed a family, provided that family somehow decided to do on half the income of a plumber becau
mircea_popescu: like, if ethereum-classic created some dilutive ethereum coins to pay off random idiots to run their chain instead of the ethereum-fork. or vice-versa.
mircea_popescu: anyway. greenmail is an antieconomic activity, in the sense that management group A divests capital goods in order to pay management group B to no longer pretend like it could do a better job of managing said capital goods.
ben_vulpes: even /i/ know investopedia is trash
mircea_popescu: also i find now that the webternet is retarded. investopedia definition for instance is pretty nonsensical. i guess pinoy everywhere.
mircea_popescu: that's the core of that concept.
mircea_popescu: but the "employees" are still paid chiefly to not go to the "competition".
ben_vulpes: this greenmail notion is entirely unrealistic, shares allocated to "enginering" are utterly trivial compared to investment tranches, and shartup operators routinely reneg on non-dilution clauses for staff.
mircea_popescu: which is also why start-up paying anything but warrants/options/stock is primo nonsense. but then again a no-nonsense approach would shut off most of the morti di fame who like to pretend they're participating in the economy, as they'd have to get proper jobs instead of aspie-ing all over town.
mircea_popescu: ie, labour is when you engage a stable of cowsies to die in your stable, circulating company scrip between the company store and the company workshop. then wages actually contribute to assets.
mircea_popescu: i've yet to see a start-up engage actual labour, in the classical sense of this term.
mircea_popescu: the sort of "salaries" start-ups pay have more in common with greenmail than with the proper concept of salary.
mircea_popescu: anyway. the scheme is both correct and very, VERY fucking scary to "start-ups", because within 12 to 18 months, ~100% of all their capital turns to "sweet parties we have thrown". which is a sad reality they dun wanna look in the face, especially not while trying to get someone/anyone to give them moar money.
mircea_popescu: anyway, the whole job of valuating a public corp reduces to adding the proper factors to those 3 categories.
mircea_popescu: seems a rather correct description of reality, too.
ben_vulpes: 3% of s.nsa's balance sheet amounts to "sweet parties we have thrown" by that reading
mircea_popescu: the difference being that the tangible you can turn back to cash whenever you want to ; whereas the intangible you gotta wait for it to decide on your own.
mircea_popescu: well, if you cash your paycheck and don't spend it, it's cash. if you spend it to buy a pair of shoes, you have a tangible asset. if you spend it to buy a round for all your friends, it's goodwill.
ben_vulpes: mircea_popescu: sure, this is likely a misunderstanding of "intangibles and goodwill" as an asset.
trinque: and the fucking bug is "mom my toy is too slow" not "HOLY FUCK THIS TOUCHES THE NETWORK WITHOUT BEING ASKED"
trinque: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=20015 << if the world ever rights "helpful" is going to be a euphemism for "tries to slip dick in when you turn your back" ☟︎☟︎
trinque: you'd like to think your machine will with default settings not take it upon itself to diddle the network
trinque: in other lol today I discover that emacs tries to ssh to host.does.not.exist at every start
ben_vulpes: mircea_popescu: how does the server rental fee make sense as "intangibles and goodwill"? ☟︎
BingoBoingo: http://btcbase.org/log/2016-08-05#1515648 << Half way to Step 1!!! ☝︎
mircea_popescu: profoundly, thank you.
shinohai: mircea_popescu: chick from earlier says to `profoundly thank you`
mircea_popescu: asciilifeform re glue, we'll be able to run a test server for it anyway.
asciilifeform: this kind of thing is a large part of why crypto implementation tends to show the attributes of religious ritual
asciilifeform: 'i cannot reverse it, thereby no one can!' etc.
asciilifeform: 'this appears to jumble the bits well, ergo cryptographically strong'
asciilifeform: the unfortunate thing about hashes is that they are every bit the voodoo that block ciphers are.
mircea_popescu: only in this context the decision to "add it" or "not add it" can be said to make any sort of fucking sense.
mircea_popescu: in this case : here is the security loss from adding a 13th step to the 12 step scheme : 79. here is the security loss from not adding it, because rng starvation : 45.
mircea_popescu: i didn't mean it that way. i meant, all items must be reduced to TWO scalars.
mircea_popescu: asciilifeform in order for management to make decisions, all items must be put in the form a > b.
asciilifeform: unfortunately crypto schemes are not really experimentally testable on 'human' timescales - in the sense that even the dumbest algo appears to 'work great!111' until one day, likely quietly, it does not.
mircea_popescu: asciilifeform no i'm not discussing here "a way to community consensus". the question is quite specific : given 12 passes, tell me how much strength i lose adding a 13th.
shinohai: has anyone checked the rng for the new blockchain.info `wallet`
mircea_popescu: (a large part of this project is me trying to field-evaluate crypto schemes in the first place)
asciilifeform: it is a common way to stream cipher.
mircea_popescu: anyway because messages are highly structured to begin with, there's a whole lotta attacking surface available.
asciilifeform: iirc bernstein's 'salsa' thing works on this method.
mircea_popescu: well, at least theoretically.
mircea_popescu: if he decyphers the 1st he gets 12 messages.
mircea_popescu: if he decyphers the 12th in a 12pass scheme he gets 1 message.
mircea_popescu: the next, but not the previous.
mircea_popescu: so it's impossible to make a rational decision here, from both ends. win-win, as they say.
asciilifeform: the one obvious gotcha is that if enemy can decipher (perhaps via known plaintext) ONE instance in the stream, he has the rest !
mircea_popescu: but anyway, it's ok that there isn't : there ALSO isn't a way to evaluate strength loss due to "rng" starvation.
mircea_popescu: basic cryptoquestion i do not expect an answer to : is there a way to evaluate strength loss by hash pass in a scheme where otp is hashed and the hash used as next pass otp, for N passes of this ?
asciilifeform: (this will determine how much optimizatory massage the item would need.)
asciilifeform: perhaps diana_coman will be able to give rough picture of the bit rate.
mircea_popescu: that part is not hard : a) you get a client ; b) you write it a drop-in wrapper for the messaging - this should be one function/class + the keygen. the same class should be used by server the same way, which we'll do the integration of.
a111: Logged on 2016-08-06 00:37 mircea_popescu: asciilifeform so im thinking here, without a byte/sec sort of idea of what messaging is liable to look like, we can't really spec anything anyway.
asciilifeform: http://btcbase.org/log/2016-08-06#1515769 << yes. also will need some detail re the desired glue (where subsystem plugs in? etc) ☝︎
mircea_popescu: asciilifeform so im thinking here, without a byte/sec sort of idea of what messaging is liable to look like, we can't really spec anything anyway. ☟︎
shinohai: mine is no better than phf 's
shinohai: dat metadata tho
asciilifeform brb, food, will come back to this thread.
asciilifeform: the one sticking point may be entropy starvation on certain systems (it is my understanding that a number of eulora players are stuck on winblowz)
asciilifeform: not really, given that mircea_popescu already solved this problem in his article by describing 'otp pool'
asciilifeform: and the constraints of the built & runtime.
asciilifeform: diana_coman will be able to describe necessaries, e.g., bit rate.
mircea_popescu: currently the "what happens past key generation" and "there's an otp" part is left pretty loophole-y
mircea_popescu: hm. im guessing we'll get together with diana_coman and write a spec ? unless you want to propose one ?
mircea_popescu: gotta admit the idea is promising. so lessee...
mircea_popescu: well... i guess specifics of this will be tied to specifics of that.
asciilifeform: answer would depend on what is specced, but if spec will substantially reflect linked trilema article, then, likely, not, and not.
mircea_popescu: would it cost a lot ? would it take very long ?
mircea_popescu: hm. like, s.mg to hire s.nsa as a subcontractor ?
asciilifeform proposes to implement this subsystem for s.mg. ☟︎☟︎
asciilifeform: soooo, http://trilema.com/2016/eulora-forward-looking-statement-august-2016 << 'From a technical perspective, the RSA migration is intended to work on the following scheme : RSA privkey is generated correctly ; this key and the server's key are used to pass OTPs between client and server' is fundamentally Right Thing.
mircea_popescu: como encontraste el articulo en trilema ?
mircea_popescu: mi gusto. decile a tus amigas tambien :)
mircea_popescu: medelin hm ? how's the weather there
mircea_popescu: one of the best parts of ba this time of years is the invasion of freesia. wunderbar.
mircea_popescu: fucking spanish imperative i never get it right. escribe y toma, is it shinohai ?
mircea_popescu: ok, escribir b68fe7aa sobre sus tetas y tome una photo. como lost otras chicas.
mircea_popescu: so tell her to speak in chan liek normal ppls.