log☇︎
144300+ entries in 0.28s
mircea_popescu: also, polynomial time isn't THAT great. depends a little on the polynomial involved :D
mircea_popescu: how useful something is is of no consequence in the classification of things. it still gets a label.
asciilifeform: 1024 is ten bits of info, readily brutable
mircea_popescu: let teh enemy fight with the army it has!
mircea_popescu: http://btcbase.org/log/2017-11-14#1737600 << oh, i see now what you were talking about. right5 you are, yes, i was focused on the edges not the middle. ☝︎
a111: Logged on 2014-05-10 20:40 mircea_popescu: fluffypony that's right, this is the channel of the minority of people into bitcoin that aren't also pedos.
mircea_popescu: oh i thought that couldn't happen!
mircea_popescu: there's prolly some stuff on trilema too but too lazy to dig moar. shoulds suffice i guess.
a111: Logged on 2015-02-24 21:00 mircea_popescu: the gays got their way with marriage prior to the freeze, but that's it. pedos aren't getting theirs.
a111: Logged on 2015-02-24 21:00 mircea_popescu: who cares that pedo today ==== faggot 1965.
a111: Logged on 2015-03-06 01:01 mircea_popescu: the very notion that the us has any sort of purpose, not even conscious or expressible, but any sort whatsoever is ridiculous on its face. it has all the purpose of a shambling headless zombie. which is why they're stuck trying to invent nonsense, "women and gayz! anti pedo!!" etc.
a111: Logged on 2015-04-30 15:40 mircea_popescu: incidentally, the original (mostly politically driven) explanation for violence against homosexuals - some sort of "homophobia" suffers a lot seeing how ALL SORTS of other sexual behaviours elicit the same violent response. such as transsexualism, or pedophilia.
a111: Logged on 2017-02-21 22:18 mircea_popescu: hence all the discussions re "fags got their weddings, pedos are next".
a111: Logged on 2017-05-03 19:26 mircea_popescu: Framedragger "have you tried not wanting to fuck guys???" whadda ya know, next they'll "cure" pedos. because they're so totally unlike gays and everything.
a111: Logged on 2017-05-07 05:09 mircea_popescu: BingoBoingo i'm preparing popcorn for when they finally unearth the pedophilia issue. so far swimming in the brackish water of imaginary problems that are only problems ot the problemizers.
a111: Logged on 2017-06-30 23:00 mircea_popescu: much saner to put all gays in jail because "they offend baby jezuz" than to put all pedos in jail because "they harm little kids"
asciilifeform: sorta lulzy to learn that a day of bulldozering costs moar than a used bulldozer.
asciilifeform: supposing i had where to keep one, and weren't gearing up for escape
asciilifeform: re 30k, it's moar than enuff for own motherfucking microscope
a111: Logged on 2017-11-14 01:09 asciilifeform: the way i'd implement the whole shebang, is simply to reject both primes if the highest bit of pq is not 1 .
mircea_popescu: http://btcbase.org/log/2017-11-14#1737387 << this is alternatively a perfectly acceptable approach ; expensive as all fuck though. prolly should be the standard for homemade keys. ☝︎☟︎
asciilifeform: i can't picture it appreciating to the point where i can spare 30k worth of it
asciilifeform: because the talent pool aint any bigger
mircea_popescu: same or moar in today's dollah.
asciilifeform: aha and the 40 remaining % gets to charge same or moar
asciilifeform: it's a surgery, d00d strips carefully the layers , takes pics
mircea_popescu: it will sadden you to find surgery is highly roboticized, in the sense of ~60% of surgeon man-hours put in by 1967 surgeons are now done by machine.
asciilifeform: the microscope per se is 1940s tech.
mircea_popescu: (there's a reason imperials have that problem -- it's called "technological progress".)
mircea_popescu: this is the fundamental, and perhaps only result to date of cryptology.
mircea_popescu: this is the problem with "complexify the code machine" tendency. somehow it appears intuitively evident that having a portion of the code INSIDE the machine is "a more complex, therefore a more secure system". it is not. 100% of the key belongs in the key. ☟︎
asciilifeform: in sadder noose, cheapest bolix ivory microscopy quote, chinese, that covers all the layers ( metal and passivation ) is ~30,000 usd. ☟︎
diana_coman: mircea_popescu, there is that
a111: Logged on 2017-11-04 14:22 mircea_popescu: consider the following point : the recent las vegas shooter had bought a room ; and had ferried a quarter ton of materiel up there ; and then the concert started, and he sat.
mircea_popescu: the fundamental imperial problem (see eg http://btcbase.org/log/2017-11-04#1732837 ) is that good imperial can never act for fear waiting may pay more. ☝︎
a111: Logged on 2017-11-14 11:46 diana_coman: that being said, I can't quite see it being any *worse* than this affair with fixed size AND fixed top bits
mircea_popescu: http://btcbase.org/log/2017-11-14#1737538 << it's worse in that the per-unit-effort benefit it delivers is smaller than the benefit we would accrue by simply keeping the current system and increasing the bitsize. ☝︎
asciilifeform: i'ma half a mind to build that antenna , btw, and see how long before confiscated
mircea_popescu: though /me is consoled to see bright minds have the same intuitive inclination.
a111: Logged on 2017-11-14 11:33 apeloyee: diana_coman: if keeping the minimum of 2^2047 for primes, you can, for example, generate primes between 2^2047 and 2^2049, and start over if the modulus is unacceptable. not sure what minimum for p and q makes sense.
mircea_popescu: http://btcbase.org/log/2017-11-14#1737536 << this was part of the original speccing of "unequal length p, q" etc. that line of thought withered and died, see the whole thing in the log it's worth the read. ☝︎
mircea_popescu: the problem here is structure, you have to have the "emitter" pretty well identified to make useful pattern measurements.
asciilifeform: but this is a thread for another time.
asciilifeform: reminds me, not long before the beginning of the time of dulap troubles, and the isp winter, asciilifeform was experimenting with statistical tests of the ~moduli~ a la dieharder
asciilifeform: or mircea_popescu was thinking of the particular one illustrated earlier
mircea_popescu: in the scheme discussed above, periodicity is specifically the point.
mircea_popescu: nobody cares for as long as they're used for what they're intended to be used, such as reproducible scientific simulation.
asciilifeform: generally there are cheaper ways to make use of 'key' to chump rng
mircea_popescu: you now look for the signature blocks of a in b, and have the whole show.
mircea_popescu: afaik "shadow brokers" didn't release this principal rsa exploitation tool of "teh equation group", but : imagine you have a machine a) working on your own special-purpose made prng ; and b) generating rsa keys all the time and on call.
asciilifeform: ( bottom or top consec. bits )
mircea_popescu: "all you need to know is where the signature 1blocks come in the "rng", as they will be preserved by binary mult, can be seen in modulus."
mircea_popescu: kinda why faux rng calibration is done on 1blocks in the first place.
mircea_popescu: (on proper rng. on bs prng / kochgpg etc, they don't.)
mircea_popescu: there ~are~ extension attacks, but those typically require more than a dozen bits known.
asciilifeform doesn't expect to see a pill against this, other than he already obvious engineering margin of using respectable number of bits of entropy for whole thing
mircea_popescu: which is to say, small.
mircea_popescu: of course, the odds of getting a FF FF q and a FF FF p are 1/2^32
mircea_popescu: and it will be larger than this surface.
mircea_popescu: however you "cut" the problem out, the surface of the cut becomes the problem ☟︎
mircea_popescu: anyway, there's entirely no way out of this. for purely mathematical reasons.
asciilifeform: this is sloppy thinking
asciilifeform: where's the ... in there
asciilifeform: srsly try it
mircea_popescu: the only way to have an N = 11111111.... is if both q and p are 1111111......
asciilifeform: problem only properly exists in the tail ('youngest') bit, where carry is 0
mircea_popescu: anyway, the same problem exists in the tail : if N is 000000001 at the end, i know BOTH p and q were 11111111 last 8 bit
mircea_popescu: left to right. most significant bit first.
mircea_popescu: literally, what the machine bytes.
mircea_popescu: ftr, octet is the fucking right word for 8 bits. a byte should be 64 bits these days.
mircea_popescu: i have a serious issue with bit/byte confusion trying to get myself out of it somehow.
diana_coman: asciilifeform, myeah, hence my "what does it even do for me simply making this interval larger"
asciilifeform: the other thing, diana_coman , is that if enemy knows that you will never use a p or q below limit l -- he can start bruting from l
a111: Logged on 2017-11-14 11:05 apeloyee: if M is say, 1.999999 * 2^2047, then ~20 first bits of p and q are known
mircea_popescu: http://btcbase.org/log/2017-11-14#1737519 << let's keep this binary. are you saying that if the first 2 octets of N are 1 therefore the first 2 octets of BOTH p and q must also be 1 ? ☝︎
davout: prime things first
asciilifeform: diana_coman: generally speaking, anything one could conceivably walk over, is unsafe - i.e. primes smaller than the number of femtoseconds in a millenium, if i had to give a heuristic
mircea_popescu: davout ima deal with this prime thing then fish out the convos if you'd like.
mircea_popescu: yes, you can expect that your modulus will lead 10 more often than 11. but this is in the book already.
davout: mircea_popescu: i remembered mostly the "pedos as afraid of women who'll identify them as beta cux"
a111: Logged on 2017-11-14 11:02 apeloyee: if the modulus is M, then p,q>M/2^2048 (because p,q<2^2048)
mircea_popescu: http://btcbase.org/log/2017-11-14#1737516 << this is no leak ; it is a restatement of "first two bits set to 11, so you only use 2045 bit keys" ☝︎☟︎
mircea_popescu: fortunately for everyone, the state of the needy is croaking before managing to complete its self-defined mission.
asciilifeform: understand, setting the top bit won't help you, i can just as easily say 'but what if the middle 2000 bits in my prime end up zeros!'
mircea_popescu: http://btcbase.org/log/2017-11-14#1737511 << amply discussed here, "pedos the next gay-like oppressed community". ☝︎
asciilifeform: diana_coman: calculate the probability of fatally small p or q.
asciilifeform: spyked: i use generic, which (quite unlike in cpp) results in a fully static structure that is created at runtime
a111: Logged on 2017-11-14 11:29 spyked: http://btcbase.org/log/2017-11-13#1737294 <-- not sure if possible with ffatronic ada subset, though, because of "no dynamic objects" restriction. in my (yet-unpublished) prototype, lisp memory size is a static knob.
a111: Logged on 2017-11-14 11:22 spyked: http://btcbase.org/log/2017-11-13#1737268 <-- this. ftr, current adalisp prototype (not-yet-published and thus yet-vapourware!) represents "pointers" as indices in a statically-allocated array.
asciilifeform: http://btcbase.org/log/2017-11-14#1737525 << this is therightthing. but note that not only is http://btcbase.org/log/2017-11-14#1737533 not a problem, but the behaviour is fundamental to ffa. in ada a structure is considered nondynamic if its size doesn't change at run time. not if 'magic number' size, like in overflowlang. ☝︎☝︎
diana_coman: asciilifeform, I meant in the final pair; i.e. you get p=3 and q=2^4095+1 sort of thing; ofc throw both in a pair if product not right size; but if not enforcing any size condition at all on p and q then you can end up with any small prime too
a111: Logged on 2017-11-14 06:15 hubud: stumbled on trilema blog last week, been reading nearly nonstop since..
mircea_popescu: http://btcbase.org/log/2017-11-14#1737481 << lol. don't forget to come out for air, trilema is larger than what can be braved in one go. you'll break yourself. ☝︎
asciilifeform: and diana_coman , there is no 'trouble if one is too small', you throw out ~both~ if the test fails
mircea_popescu: kinda curious how this bet works out irl.
asciilifeform: imho ^ is The Right Thing, at least for folx who aren't generating keys in a burning hurry
a111: Logged on 2017-11-14 03:58 lobbes: http://btcbase.org/log/2017-11-13#1737253 << time will tell. Depends on how much the logs-to-date worth of archives end up being in drive space. Bandwidth also a factor. Many things left to be sussed out.
mircea_popescu: http://btcbase.org/log/2017-11-14#1737465 << i'd expect a few thousand items at 15kb each or so, so maybe 100mb total ? ☝︎
a111: Logged on 2017-11-14 01:09 asciilifeform: the way i'd implement the whole shebang, is simply to reject both primes if the highest bit of pq is not 1 .
a111: Logged on 2017-11-14 11:25 diana_coman: apeloyee, if I understand that correctly basically the only way to plug that leak would be to give up on diddling p and q, including setting size; which would mean just get random pairs of primes until their product fits the desired number of bits for the key; obv this lands into the trouble of having one of them too small