asciilifeform: upstack re pow item : i'ma leave an observation for mircea_popescu ( and others, but he's prolly the one to crack this nut properly ) . consider how in classical bitcoin , pow is used for 2 entirely different and quite orthogonal items : initialcoindistribution, and rewindprotection (i.e. preventing history-rewriting) . it's a bulldog-rhinoceros. so q is, suppose that coin distribution somehow took place at t==0. do you still need po

asciilifeform: ( sores -- http://shaov.kulichki.com/texts/volshebstvo.htm )

asciilifeform sings silently '...Говорили мне ребята: "Виски ты не пей", А я как только виски тресну, Так мне чудится фигня...'

asciilifeform has yet to taste this mythical item, 'good whisky'

asciilifeform: '...there's whiskey in the jar...'

asciilifeform: dunwanna diff it -- dun put it where things are gettin'diffed

asciilifeform: that's what cleaning's for

asciilifeform: ( e.g., gprclean actually cleans ffa . )

asciilifeform: i have scripts that ruthlessly thermonuke the kind of rubbish that doesn't get diffed, instead.

asciilifeform: i've never done it

asciilifeform: why wouldja do this, phf ?

asciilifeform: no srsly what's it do

asciilifeform: what does 'exclude' do

asciilifeform: which is this, phf

asciilifeform wonders whether he gotta commit another warcrime in the spirit of http://btcbase.org/log/2017-02-28#1619719 , or to simply lay the http://btcbase.org/log/2017-12-18#1753876 item out in the l0gz, reasonably compactly. ☝︎☝︎

asciilifeform: but i'ma bbl: meat.

asciilifeform: ( 'pow in tx not in block' )

asciilifeform: it uses an algo from an ancient mircea_popescu thread

asciilifeform: possibly.

asciilifeform: ( it is specifically a pill against 'can recognize work done by itself as opposed to work done by others' aspect ) ☟︎

asciilifeform: upstack in re 'mining is a bug' , asciilifeform has a theoretical pill, potentially interesting , but will chew on it for some weeks -- there's gotta be a catch

asciilifeform was scratchinghead, puzzled

asciilifeform: lol

asciilifeform: aaaaa

asciilifeform: this is in cr ?

asciilifeform: that's, what, 100m in modern printolade neh

asciilifeform: ( sucks for the reason, incidentally, that mircea_popescu already fingered , http://btcbase.org/log/2017-12-02#1745524 and elsewhere ) ☝︎

asciilifeform: mining intrinsically, i suspect, suxx, and fiddling with pow won't make any dent in it

asciilifeform: after 5yr of thinking about it daily, asciilifeform still has nfi what even ought to be asked of a pow !! ☟︎

asciilifeform: i was about to say

asciilifeform: it's a bad thing if you want 'asic-resistant' or even fpga-resistant

asciilifeform: modexp in turn can be cut into B mod-mults, where B is bitness

asciilifeform: ( multiplication parallelizes 3 ways , with karatsuba, and N ways with toom-cook )

asciilifeform: it's not necessarily good choice, in that it is very amenable to parallelization

asciilifeform: in applications where people want 'expensive', they ask for 1e6 or whatever iterated hashes

asciilifeform: sha2 (and for that matter other hashes) weren't built to be expensive, quite opposite

asciilifeform: what's to stop him

asciilifeform: so miner waltzed his nonce until it's a composite and trivial

asciilifeform: ok..

asciilifeform: i'ma allears

asciilifeform: mircea_popescu: humour me, how wouldja turn this into a pow

asciilifeform: ( though even there , you'd shed 1 or 2 from the exponent , not earth-shaking necessarily )

asciilifeform: this is one of those things that really begs for dedicated iron

asciilifeform: it was aha

asciilifeform: supposing anyone can be aroused to do it ( g++ aks.cpp -lgmp -o aks )

asciilifeform: i'ma let sumbodyelse alsotry, to make sure.

asciilifeform: i was hoping to find that i made a mistake, but the numberz check out.

asciilifeform: if somebody made a practical aks, it is a deep, dark seekrit from asciilifeform . ☟︎

asciilifeform: and the order is log(n^6)

asciilifeform: for a 14 BIT NUMBER

asciilifeform: btw that french aks is geological. n==11701 has been running for 10+ minutes, and is on ~9600; author's code indicates that answer appears at iteration 1000000007. that'd be 723 days on this box. ☟︎

asciilifeform: and we don't know how the shit-dust comes back together to form what gpg thought was primes

asciilifeform: it'd be one of the shit-dust ones

asciilifeform: rather than '2 fat factors'

asciilifeform: that's what it'd actually look like, if this animal exists

asciilifeform: thinkaboutit, stronger m-r found both to be prime

asciilifeform: but it isn't clear to me why weaker koch test would have different answer than stronger m-r

asciilifeform: not specifically; was tested only using gmp's m-r. really oughta do both.

asciilifeform: the first passed m-r . it's the only primality test currently in battlefield use.

asciilifeform: makes sense ?

asciilifeform: and the bottom item was reported as a factor on account of each, taken separately, having passed gmp's m-r after the 1st was found.

asciilifeform: good example.

asciilifeform: https://archive.is/EJSOD

asciilifeform: only to a numbers that are factors of 2 or moar moduli. and they dun come with any certificate of primality, except for the smallest 'dust' ones

asciilifeform: so from that pov i have no access to the original p,q that gpg may have generated, for any of the keys

asciilifeform: phuctor finds a divisor. if it is toobig for sieving, i'm left with m-r, when asking whether it is a ~prime~ divisor.

asciilifeform: to say if 'there's moar' or not.

asciilifeform: m-r

asciilifeform: grrrr

asciilifeform: i'm left with... m-4

asciilifeform: say i have 2 2048b pieces on the table.

asciilifeform: mircea_popescu: consider how phuctor decided which numbers are 'completely' phuctored , to begin with

asciilifeform loox at http://btcbase.org/log/2017-12-18#1753697 ... ☝︎

asciilifeform: place of 'cryptographic community' is the lime pit.

asciilifeform currently suspects that none will, but nobody suspects worth a damn until -- found

asciilifeform: walk the known prime-divisors and see how many pass gpg-1.4.10's litmus

asciilifeform: and i'ma test it as soon as dulap-III is up ( btw crate of raid cards just came in last hour )

asciilifeform: y'know , this hypothesis is testable

asciilifeform: the only 1 with factors sized somewhere near half the bitness of the modulus, other than the khadeer key ( which had P == nextprime(Q) , which dun fit the bill )

asciilifeform: mircea_popescu: the only 1 that might fit the shape of this item is the german 'research key' one

asciilifeform: esthlos: point is , there exists a guaranteed-correct algo. using it, supposing it were practical, is The One Right Thing

asciilifeform: even something with, e.g., 6 month, run time, can be useful. ( esp. if parallelizable )

asciilifeform: (err, candidate-primes)

asciilifeform: btw if esthlos ( or anyone else, anywhere ) knows how to aks on 2048b primes in nongeological time, i'll happily port their algo to ffa, even if original is in cobol

asciilifeform recommends that n00bz write own, from scratch, without looking even at mine

asciilifeform: ben_vulpes had something resembling a CL one

asciilifeform: material will make 9000x moarsense

asciilifeform: i very muchly recommend that all n00bs, regardless of skill, follow the www tutorial

asciilifeform: nope, and the complete barrett is unreleased , it is not up to releasable standard yet

asciilifeform: there was a version given to mod6 , should be in the log somewhere , that had all of the basic arithmetic (+,-,*,/)

asciilifeform: and no i have no intention of dropping it out of schedule

asciilifeform: this is tricky business because i changed a great many things, ch1-3 is in many ways incompatible with the old item

asciilifeform: or mircea_popescu was asking for other-than-ch1-3 ?

asciilifeform: and on phf's

asciilifeform: it's on my www, what

asciilifeform: fed with fg naturally.

asciilifeform: ( where asciilifeform observed that, for serious battlefield key, a month of pre-use m-r, plus a box doing it continuously ~4evah , is not excessive )

asciilifeform: ( we even had thread, not long ago )

asciilifeform: m-r with programmable iteration knob is however worth making imho

asciilifeform: matter of fact, it is. unless there's been a serious advance that i slept through