log☇︎
144100+ entries in 0.515s
asciilifeform: ( see ancient metathread running back to 2013 )
asciilifeform: afaik even the shoddiest 1980s hash algo, produce ~perfect 'white noise'. hence the popularity of faux-rng via hashwhitener etc.
mircea_popescu: see their PRIME densities etc.
mircea_popescu: might actually be an interesting approach to the whole "hash evaluation" problem.
asciilifeform: ( suffers from the flaw of potentially never terminating tho )
asciilifeform: dun suffer, in principle, from the nextprime()-problem
mircea_popescu: in fact, the return 4; hash is the ~only hash function about which we can say with certainty that no inverse exists.
asciilifeform: (i.e. such that there is ~nothing~ that can be said in advance re what kind of prime will be chosen)
mircea_popescu: apeloyee great hash. always yields 4. unbreakable. what's the problem ?
asciilifeform: well if done with nextprime() then yes worst
apeloyee: return 4;//guaranteed to be random
mircea_popescu: i can'\t come up with a worse one on the spot.
mircea_popescu: in fact imo the "prime hash" is a textbook example of "worst hash ever".
asciilifeform: this is troo hm
apeloyee: some primes have many composites before them and thus are more likely
asciilifeform: its only flaw is that it leaks via time
mircea_popescu: a hash where output x is y times more likely than output z is not a hash. it's a fash.
asciilifeform: and how the hell does it compare to 'add the digits' -- there you get $base possible outputs
apeloyee: I didn't claim prime constructer, just an improved method to generate candidates for miller-rabin
mircea_popescu: that you'll get some values more likely than others.
mircea_popescu: "add the digits" is a better hash.
asciilifeform: at any rate, prime-constructor ( see apeloyee's thread re subj ) is entirely separate problem from 'prime hash'
mircea_popescu: we're still not clear on "The needs this wrench helps". so far very much blender-toilet. "maybe someone somewhere needs to make turd batidos"
asciilifeform: you get a handful of primes this way, before the arithmetic becomes unwieldy
mircea_popescu: ie, the 5th prime down.
mircea_popescu: nevertheless, the correct solution to this "quickly, prime this many bits long" is a n, k tuple which contains n as the bitsize and k as the "oddness". if you want the n - 396 k = 5 prime you get 2^396- 1229
asciilifeform: it suffers from all of the limitations of other hashes.
asciilifeform: ( nothing to stop you from imposing other types of hash in between, incidentally . as i did in the pubexp example )
asciilifeform: call it hash, yes, that outputs a prime.
mircea_popescu: well then stfu and read alf's haranguing re hashes.
mircea_popescu: there is no such correspondence. there's more n than log n
asciilifeform: i don't want ~an~ n-bit prime, i want THE n-bit prime corresponding to arbitrary n-bit input I.
mircea_popescu: if you want an n bit prime calculate 2^n and substract the correct small integer.
asciilifeform: that doesn't auto-map a n-bit integer to an n-bit prime.
mircea_popescu: asciilifeform there's much faster ways, such as the 2^x - y method.
mircea_popescu: apeloyee there is no such thing in tmsr rsa.
asciilifeform: one or the other, per packet
asciilifeform: apeloyee: with 4096b rsa troo
mircea_popescu: enlighten me, what does this wrench perfectly legitimately ever do ?
mircea_popescu: "but mp, rats don't reproduce by eggs" "And why do you have to thank for that ?"
asciilifeform: 'where is next prime' is perfectly legit wrench in any respectable numeric toolbox, wat
mircea_popescu: there are some primitives you don't wanrt to keep around,
asciilifeform: apeloyee: sorta why i suggested making pub-exp nextprime(keccak(commentstring)). satisfies the basic req of e being 1) long 2) nonstandardized
asciilifeform would rather pick up in person, when he finally gets around to visiting mircea_popesculandia
asciilifeform: and one of these is the unfragged ip packet.
asciilifeform: the sad thing is that ( at least until we get, e.g., the shortwave net, going ) there are some boxes of fixed size, that in practice gotta be sat down into if at all possible
mircea_popescu: and i just appointed a new crown provider of chocolate. you should see this thing, so endearingly evidently hand-made copy of "how a chocolate looks" when hershey makes it...
apeloyee: the amount of computation that you must do ,<< the same as for any packet: 1 mod-exp and check padding. >> and bits you must buffer, to do friend-or-foe, is considerably larger. << twice as much. might be acceptable, depending on circumstances.
mircea_popescu: in other domestic tranquility news, just finished stuffing a mason jar with baked peppers. they have some FABULOUS kapja peppers here.
asciilifeform: the 'everyone has same e' thing was slipped in under clinton , when his nsa invested in 'acre of asics' with e=3|e=65537 presumption baked in
jurov: a111: strings are sequences, not lists. cons/car/cdr does not apply, there's different set of functions for these ☟︎
mircea_popescu: for exactly asciilifeform 's reason : shorter this way
mircea_popescu: aha. yes, this is exactly what we do not
apeloyee: of course, must check that e/=1
mircea_popescu: so your model is, i have your pubkey, and encounter an item signed by that N, with a specific e that's included in the signed text ?
a111: Logged on 2017-11-08 23:10 asciilifeform: returning to the exponent thing, seems that mircea_popescu is right, nothing particularly interesting can be done by distributing a pub with e' . ( other than 'believe me , his e is 3' and then messages ~to that pub~ are breakable if padding is broken. but that' sit . )
asciilifeform: mircea_popescu: apeloyee was extending the http://btcbase.org/log/2017-11-08#1734664 argument, as i understand ☝︎
apeloyee: I thought you said modulus _is_ the identity, and here I presume the modulus is known
mircea_popescu: but a signature signed by a pubexp i didn't have PRIOR to the receiving of the signature is definitionally worthless.
apeloyee: can attach your pubexp in plaintext, to the signature
asciilifeform: the amount of computation that you must do, and bits you must buffer, to do friend-or-foe, is considerably larger.
asciilifeform: right, but to do it, you first have to receive proposed exp 'on faith' at that endpoint
apeloyee: u can use these to verify that purported pub-exp is validly signed
asciilifeform: not if i have '9000' entry points to my gossip net, each of which rejects malformed rsa packets in O(1) at line speed
apeloyee: well, that means the enemy can drown you even if we kave pub.exps, by simply flooding
apeloyee: I don't see how you can extend to this case
asciilifeform: apeloyee: by same argument, '1 bit of the modulus is sufficient to init conversation, after it we'll agree on next bit' etc
apeloyee: http://btcbase.org/log/2017-11-14#1737542 << that's just DoS. but, if you have computational capacity, you check the padding. may also require that it's signed with my key, with the pubexp attached if you don't know it. Thus, the modulus is in principle sufficient to _initiate_ the converstaion ☝︎
phf: for example your Symbol_EqualP should just be a pointer comparison, rather than string comparison. (the whole point of a ~symbol~ over a string is that it's interned, i.e. same sequence of characters always map to the identical Symbol object)
deedbot: http://trilema.com/2017/il-merlo-maschio/ << Trilema - Il merlo maschio
a111: Logged on 2017-11-14 11:25 spyked: http://btcbase.org/log/2017-11-13#1737264 <-- strings are (lisp) lists-of-characters. which, as it is, unfortunately makes parsing and evaluating builtin functions (e.g. cons, car, cdr) a pain in the ass. can be structured cleanly though. also, this makes it not a simple matter of find+replace in shithub scheme.adb.
phf: http://btcbase.org/log/2017-11-14#1737529 << that doesn't sound right, read and eval are distinct phases, by the time you get to eval you shouldn't be operating with strings when but instead with interned symbols (i.e. things that can be eq'd in lisp and pointer equivalent on c machine level) ☝︎☟︎☟︎
mircea_popescu: mmm 2005. no, there was some 1970s item
mircea_popescu: http://btcbase.org/log/2017-11-14#1737917 < o right! not a bad film, at that. ☝︎
BingoBoingo: In other news, apparently Tower 4 of the world trade center is not actually inside the world trade center free zone.
a111: Logged on 2017-11-14 15:30 asciilifeform: there was a '90s american film where there is a scene, where an airplane lands in africa and in fast motion gets stripped for parts, like elephant carcass by hyenas
davout: http://btcbase.org/log/2017-11-14#1737688 <<< ty, will read ☝︎
a111: Logged on 2017-11-14 16:54 BingoBoingo: mircea_popescu: I asked him for the "how do we get a corporation fast" answer. His answer is off the shelf. This isn't the first time I've heard "bank reference" being bandied aboutwith respect to opening a corporate account.
davout: http://btcbase.org/log/2017-11-14#1737789 <<< needed the same when opening a bank account in .mu, reference letter didn't need any particular judgement about whether i'd be able to meet a financial committment, but just something along the lines of "had bizns with this gentleman for X years, didn't leave with unpaid debts, isn't a fucking gypsy" ☝︎
mircea_popescu: "oh the numbers are wrong". yeah, im sure they are. and the tractors invented, and the working the fields with oxen and horses and wives pre 1940 calumny, perhaps. hurr.
mircea_popescu: and he has the numbers : right here, where we happen to be, i ~know the name of the local lord~, and his numbers : 600 to 2209, he says. and what is teh retort ?
mircea_popescu: dude has the audacity to ask for "a ballet on the topic of having finished the kolohoz-isation". you know ? at least asks the right fucking questions, even if nobody in audience has the mettle to make answer.
mircea_popescu: the 1965 and 1968 items absent, but pars pro toto this shall have to do.
mircea_popescu: im totally finding this, motherfucker, im sure i saved copies on trilema somehwere PRECISELY for this conversation
asciilifeform: when asciilifeform wanted to read lysenko in the original , he had to visit the rarest-rarities reading room in american national lib of medicine
mircea_popescu: i read piles of ceausescu transcripts, they're all exactly nowhere. 1968 meeting with the writer's union ? really, NOT EVEN THAT ? 1965 may 19th "who wants a seat in helicopter" one ? nope ?
asciilifeform: i found last yr, somewhat surprisingly, that the works of stalin are not sitting in plain txt anywhere
mircea_popescu: dude it's incredible how fucking useless the internet is.
mircea_popescu: humiliation is definitionally "in the terms of the idiot"
mircea_popescu: see, this is the problem : selection and education are different.
mircea_popescu: not according to the ant.
asciilifeform: maybe ants also have complicated requirements for how to humiliate. but ! ddt still worx!
mircea_popescu: asciilifeform cuz that's what the mediocre intellect has decided upon.
asciilifeform: why does it have to be a paper humiliation.
mircea_popescu: BingoBoingo hung with ~nobody. didn't even fuck around, married a 25yo whore 9-year whore, stuck to her 50 years.
asciilifeform: mebbe this shows that asciilifeform is same sort of peasant as the shoemaker, but finding it hard to see why not counts. having pantsuit hoisted on a public stake would even moar count imho
mircea_popescu: ideally, it's a "hey, check out how your stupidity #1 contradicts your stupidity #2, what do you have to say for your stupidity that isn't throwing a hissy fit ?"
BingoBoingo: <mircea_popescu> entirely oral thinker. << Ah, hung out with too many Frenchies did he
mircea_popescu: it has to be "mrs pantsuit, where's your having won the presidency nao ?" at the lowest.
mircea_popescu: yes, but those "don't count"