log☇︎
146800+ entries in 0.086s
asciilifeform: what ought to do instead, is to rngize the bottom Q bits, where Q is log2(estimated prime gap at the current bitness, times severalfold engineering margin) ☟︎
asciilifeform: ( the koch method, of taking R and adding 2 to it until m-r says yes, trivially leaks )
asciilifeform: but i have not yet tried this.
asciilifeform: incidentally it can probably be shown that you don't have to throw out ALL of the bits in a candidate R, when it is not found to be prime, and still not leak anything
asciilifeform: i am probably doomed to make a pci version, because of this, at some point. but so far not urgent itch.
asciilifeform: theoretically one can connect an unlimited # of'em to pc (in practice linux kernel starts behaving oddly when more than 7 serial devices, i found )
asciilifeform: 7-8 kB/sec, depending on room temperature.
asciilifeform: this costs time, yes.
apeloyee: "pick random" will usually yield a composite, which must then be filtered out.
apeloyee: aha, so you're ok with the method produces junk, so long as the primes in range are equidistributed?
asciilifeform: and since i built the rng, i can also be quite certain that there is not such a prime that it will forever avoid generating.
asciilifeform: i can show that no 2 inputs will produce the same prime, trivially
asciilifeform: the random-rubbish followed by probabilistic test , satisfies this demand within the obvious limits of the test.
asciilifeform: i want a method, as stated above, where inputs map one-to-one and onto, the primes , up to the register width.
apeloyee: plox to qualify. you want a method to generate Nth prime, but accept the "pick random number until prime" , despite in not being the Nth prime generator, but no other methods?
asciilifeform: but as i currently understand, my q reduces to 'find The Formula For Primes' (tm)(r), a millenium-long headache for mathematical folx , to this day unsolved. ☟︎
asciilifeform: if apeloyee knows a proof for this, i'll read
asciilifeform: i believe in the arithmetic, yes. but how to show that the two conditions i stated, hold ? ☟︎
asciilifeform: also gotta show that no prime exists , in the given bitness width, that CANNOT be chosen. ☟︎
asciilifeform: in particular, gotta show that no two inputs will produce the same prime ☟︎
apeloyee: "pick random" is obviously not this
asciilifeform: to have a satisfactory constructor, it would have to be shown to reduce to this and exactly this.
asciilifeform: i.e. a 1:1 pipe from random ints to large primes.
asciilifeform: let's model the ideal prime-shitter. it would be an item that takes integer N , of whatever bitness, and produce the Nth prime ( or eggog if the Nth prime is bigger than the register bitness permitted. ) ☟︎☟︎☟︎
apeloyee: at that point might as well give up
apeloyee: perharps running rabin-miller "amplifies small/temporary imperfections in the rng, into fatal"
asciilifeform: the other problem is that prime constructor proof may or may not fit-in-head. infineon's method probably seemed correct to various folx.
asciilifeform: the use of constructed primes, potentially amplifies small/temporary imperfections in the rng, into fatal
asciilifeform: 1 problem is that perfect uniform distrib and perfectly independent rng bits, cannot be shown to exist physically.
a111: Logged on 2017-11-07 14:34 a111: Logged on 2017-10-07 21:48 apeloyee: http://btcbase.org/log/2017-10-05#1721485 << alternatively, can *construct* numbers which don't have very small factors. pick a nonzero remainder mod 2, mod 3, ... mod largest-prime-fit-in-your-primorial and find what number of primorial is congruent to it using chinese remainder theorem
apeloyee: the http://btcbase.org/log/2017-11-07#1733339 method, given uniformly distributed and independent remainders, generates uniformly distributed numbers modulo the primorial ☝︎
asciilifeform: it is not the same exact algo, no. but it is same from my pov, in that it is NOT the 'find b-bit random R and test for primality, potentially forever, until found prime'
a111: Logged on 2017-11-07 14:34 a111: Logged on 2017-10-07 21:48 apeloyee: http://btcbase.org/log/2017-10-05#1721485 << alternatively, can *construct* numbers which don't have very small factors. pick a nonzero remainder mod 2, mod 3, ... mod largest-prime-fit-in-your-primorial and find what number of primorial is congruent to it using chinese remainder theorem
apeloyee: it's not the same method as http://btcbase.org/log/2017-11-07#1733339 ☝︎
asciilifeform: how is this not 'constructed prime' ?
apeloyee: I understood that much
asciilifeform: apeloyee: q was re the effect of composite exponent
apeloyee: http://btcbase.org/log/2017-11-07#1733338 << per bernstein, they didn't. ☝︎
apeloyee: I must say that I dont understand how "exponent 131074 reduces to exponent 2 somehow", and what CRT has to do with it ☟︎
asciilifeform: no reason to. not on ffa, at any rate.
mircea_popescu: asciilifeform http://trilema.com/2017/tmsr-rsa-spec-extremely-early-draft/#comment-123474 so there.
asciilifeform: 'Further computations were consistent with the guess that p and q were being generated as powers of 65537 modulo L, where L was either the product of all primes through 691, or the product of all primes through 701.'
asciilifeform: any method of 'constructing' primes , rather than hunting for wild, is guilty until proven innocent of 1) constriction 2) bias
a111: Logged on 2017-10-08 00:16 asciilifeform: http://btcbase.org/log/2017-10-07#1722402 << this is a fundamentally wrong way to generate cryptographic primes. we had a thread about it, http://btcbase.org/log/2017-08-14#1697562
asciilifeform: 'This could even be a deliberate weakness with plausible deniability: "Oh, sorry, nobody warned us that 2 mod 11 was okay and 10 mod 11 wasn't."'
a111: Logged on 2017-10-07 21:48 apeloyee: http://btcbase.org/log/2017-10-05#1721485 << alternatively, can *construct* numbers which don't have very small factors. pick a nonzero remainder mod 2, mod 3, ... mod largest-prime-fit-in-your-primorial and find what number of primorial is congruent to it using chinese remainder theorem ☟︎☟︎
a111: Logged on 2017-10-17 05:59 jurov: "The flaw resides in the Infineon-developed RSA Library version v1.02.013, specifically within an algorithm it implements for RSA primes generation. "
asciilifeform: in other lullies, bernstein exposes the item the original 'authors' sat on in http://btcbase.org/log/2017-10-17#1725945 >>> https://blog.cr.yp.to/20171105-infineon.html ☝︎
mircea_popescu: wait, wrong decade. it's not the 1990s anymore!
mircea_popescu: heh. i think quantum computing was abandoned because it was discovered it leaked CFC gasses which are bad for the ozone layer
shinohai: You may remember Vitalik from other great books such as "How to run a quantum computing scam to raise capital for a crypto scam" and "Dress like a furfag and cook your own meth!"
shinohai: "How to fail and make people liek it - By Vitalik Butterin"
mircea_popescu: it's important to be creative in any case. they can produce more eth at any time, but the experience was valuable.
davout: according to unverified hearsay it seems that ~1% of the total mEth supply got accidentally'd
shinohai: Shit has more rollbacks than Walmart as it is.
davout: or maybe they'll grow used to having Ctrl+Z as a thing
davout: who knows, maybe that leads to an "Ethereum Classic 2"
davout: "This means that currently no funds can be moved out of the multi-sig wallets."
davout: buttfinex also has such a market with roughly the same volume figures
davout: trade volume seems pretty low on all these ~100 btc/24h
deedbot: http://trilema.com/2017/vedo-nudo/ << Trilema - Vedo nudo
mircea_popescu: iirc the confederate republic scored this low after lee surrended and not at any point before
mircea_popescu: well, 10 day futures at 10% is terrible in and of itself.
mircea_popescu: when's teh stuff supposed to come online again ?
mircea_popescu: shinohai that dun look too good.
ben_vulpes: was shitty for a bit there
asciilifeform: ( at trilema, that is )
asciilifeform: diana_coman: see thread
diana_coman: hmm, I suppose I could run a trial test on a batch of generated public exponents with co-prime rather than strict prime requirements, to see what small factors are there but not sure if this will say a lot really in itself; fwiw I don't feel particularly comfortable with the idea of a non-prime public exponent but I don't have mathematical proof for weakness introduced, hence my question
jhvh1: mircea_popescu: The operation succeeded.
mircea_popescu: !~later tell apeloyee http://trilema.com/2017/tmsr-rsa-spec-extremely-early-draft/#comment-123449 let's get teh guy involved as well.
shinohai rushes to set up "betterbeastforum.com" and offer unlimited messaging and access for 5 bux a month .....
mircea_popescu: turns out it's not the brightest idea to deny people access to the body of a message if you're gonna send that subject line via email.
mircea_popescu: shinohai part 2 of the lulz is that the userbase is now revolting. "Well screw that then. This website is a big hoax."
deedbot: http://trilema.com/2017/alexandru-osvald-pastorel-teodoreanu/ << Trilema - Alexandru Osvald "Pastorel" Teodoreanu
a111: Logged on 2017-10-08 14:18 mircea_popescu: " “The Saudi’s will not be able to link the S-400 with Saudi’s current (US and Europe-sourced) infrastructure, nor will they be able to connect the S-400 with US systems." << says who the everloving fuck.
asciilifeform: as described on trilema and elsewhere, usd ain't money, it's 'money[*]', and the * typically includes 'cannot be spend on defense from usg'
mircea_popescu: idjits don't know when to open fire, keep waiting for "a better day".
mircea_popescu: "richest guy on planet" could not afford tanks and missile launchers enough to dispose of "new anti-corruption unit" ?
mircea_popescu: kinda weakly defended, them bluebloods.
asciilifeform: possibly gotta 1938 the wrong-chirality bluebloods who want to spend it in usg-unapproved ways (e.g. btc) ..?
mircea_popescu: hey, the saudis have a lot of paper moneyz.
mircea_popescu: "we're running out of money, tax the provinces" ?
asciilifeform: afaik 'anticorruption' only ever means 1 thing
mircea_popescu: oh i see, it's the even friendlier us ?
asciilifeform: 'The conference was spearheaded by Crown Prince Mohammed bin Salman, who is pushing an ambitious reform plan that is aimed at reducing the country's dependence on oil over the next decade. The prince is also leading the anti-corruption committee that has the authority to investigate, arrest, issue travel bans and freeze the assets of those it finds corrupt.' << lol!
mircea_popescu: so what's the rumous mill, the friendly iranians ?
asciilifeform: the moar, the merrier.
asciilifeform: they can have a 1938 if they feel like, wainot
hanbot: <asciilifeform> in other lulz, https://archive.is/UVcTG >> 'reports of death of yet another Saudi prince' << meanwhile i was reading this morning "riadyh ritz-carlton now detention center for saudi royals", (http://archive.is/p7KPc), but tuned out when "evidence" seemed to consist of "booking websites" not having rooms available. ☟︎
shinohai: I'm just happy to know mircea_popescu isn't building a new kind of stable.
deedbot: http://trilema.com/2017/how-the-beastforumcom-private-messaging-function-became-a-paid-user-only-item/ << Trilema - How the beastforum.com private messaging function became a paid-user-only item ☟︎
mircea_popescu: "but there are no alternatives111!!!"
ben_vulpes: lookie, usg agents teaching the world that depositing into their banks is a fucking terrible deal
a111: Logged on 2017-11-06 16:21 asciilifeform: https://archive.is/ZuzvO << from same rag, 'They accused him of “structuring” — depositing money in increments of less than $10,000'
davout: http://btcbase.org/log/2017-11-06#1733218 <<< "The IRS wouldn’t comment on Kwon’s specific case, but a spokesman noted that he pleaded guilty to the structuring charge." ☝︎
BingoBoingo: !!up TomServo
asciilifeform: may be in the logs.
asciilifeform: but i never tried
TomServo: Ah, bit more than I can chew, trying to fashion your crapolade masks into a workable Gentoo laptop.