18900+ entries in 0.02s
mircea_popescu: but the main issue is that this reads like an approximately sorted succession of conclusions. painfully absent a) the logical justification for the various sentences and b) a very clear indication of structure, how they all work together ; from this b there would then flow a much better a.
mircea_popescu: also, dubious expressions, a collection : "to understand that why they want" ; "if your words actual words are avaible" ; "manufacter a pretext" ; "lack anyway to know" ; "the manufactered insecurity" ; "maladapted in ways invite bullying."
mircea_popescu: on the other hand, entirely unclear to me why the fuck we're discussing some [evidently buggy/misimplemented] program or my own inept handcalculus. either we agree a system of n equations with n unknowns is determinate or we don't and that's the end of the matter.
mircea_popescu: if we manage to fixate on a specific datapoint we'd prolly benefit here.
mircea_popescu: i don't actually follow your proggy any ; regardless, the values it spits dun seem to agree with what i expect.
mircea_popescu: by now, covered all the possible combinations of 4 bits (ain't that many) -- but never fell on the mp.py 1010.
mircea_popescu: K = 0101, P = 1110. RB = (1110), (1101), (1011), (0111). E1 = 1 x 1 x 0 = 0 ; E2 = 1 x 1 x 1 = 1 ; E3 = 1 x 0 x 1 = 0 ; E4 = 0 x 1 x 1 = 0 ; E = 0100.
☟︎ mircea_popescu: K = 0101, P = 1110. RB = (1110), (1101), (1011), (0111). E1 = 1 x 1 x 0 = 0 ; E2 = 1 x 1 x 1 = 1 ; E3 = 1 x 0 x 1 = 0 ; E4 = 0 x 1 x 1 = 1 ; E = 0101 ?
mircea_popescu: lmao ima have to redo this. apparently im terrible at handmaffs
mircea_popescu: unless you're willing do debug the .py, it's spitting out wrong values.
mircea_popescu: asciilifeform K = 0101, P = 1110. RB = (1110), (1101), (1011), (0111). E1 = 1 x 1 x 0 = 1 ; E2 = 1 x 1 x 1 = 0 ; E3 = 1 x 0 x 1 = 0 ; E4 = 0 x 1 x 1 = 0 ; E = 1000.
mircea_popescu: asciilifeform you agree that if i give you 5 equations with 5 unknowns, this is in fact resolvable ?
mircea_popescu: to be clear : it's exactly the same scheme ; showing you how P ~must~ be obtainable from known K and E.
mircea_popescu: so if K[q] = 0, then that line's skipped ; otherwise it's applied.
mircea_popescu: no no, x is xor * is multiplication (in the sense that if the key is 0 at that offset, the rb dun get applied)
mircea_popescu: E[1] = P[1] x P[2] x P[4] ; P[2] = P[2] x P[3] x P[5] is what i meant.
mircea_popescu: for a 5 bit key you only have to try 2^3 permutations or so, it's true. but anyways.
mircea_popescu: so : with K it's a system of 5 equations with 5 unknowns ; without K it's a system of 5 equations with 10 unknowns.
mircea_popescu: To obtain P back from E without K : P[1] = P[1] x K[1] * P[1] x K[2] * P[2] x K[3] * P[3] x K[4] * P[4] x K[5] * P[5] ; P[2] = P[2] x K[1] * P[2] x K[2] * P[3] x K[3] * P[4] x K[4] * P[5] x K[5] * P[6] ; and so following all the way down.
mircea_popescu: To now obtain P back from E and K : P[1] = P[1] x P[2] x P[4] ; P[2] = P[2] x P[3] x P[5] ; P[3] = P[3] x P[4] x P[1] ; P[4] = P[4] x P[5] x P[2] ; P[5] = P[5] x P[0] x P[3].
mircea_popescu: E therefore is : E1 = 0 x 0 x 1 = 0 ; E2 = 0 x 0 x 1 = 0 ; E3 = 1 x 1 x 0 = 0 ; E4 = 1 x 1 x 0 = 1 ; E5 = 1 x 1 x 1 = 0 thus E = 00010.
mircea_popescu: Consider K = 01010 ; P = 00111 ; RB = (00111), (01110), (11100), (11001), (10011).
mircea_popescu: as long as P doesn't have more bits than K, this is a determinate system.
mircea_popescu: so this is, in fact, a system of (here) 512 xor-equations, with ak nown result (e[x]) and a known parameter matrix (K)
mircea_popescu: (provided of course k>=n, which yes, it's a block cipher)
mircea_popescu: asciilifeform check this out : as per the "chipher", let there be a plaintext P of n bits ; and a key K of k bits. given a ciphertext E of n bits, it is a fact that any one bit of P is the result of xoring of up to k bits of P. if you know K you know ~which ones~, and as you have E you know ~what they must xor to~. this results in a message-wide system of k equations which is determinate.
mircea_popescu: asciilifeform this is so much better than having to correct the title.
mircea_popescu: "our block cipher has to have backdoor because we built a paper shredder and well..."
mircea_popescu: and no, "we have pre-padded ring buffers with THIS particular message that's worth hardwiring because it's ever so magical" ain't a reasoning.
mircea_popescu: asciilifeform like it or not, this is the ur-blockchipher. ALL OTHERS, not only casually but NECESSARILY, are mini-clipped versions of this. "competition or no competition", if it got shitboxes it's this and naught else.
mircea_popescu: ow shit, i mystypoed in the title, wtf is a "chipher".
mircea_popescu: i'd have said more, but the editor showed wordcount 777, and i deem this a very good omen re cipher quality.
mircea_popescu: bright kid walks in with a convincing story as to how he's gonna walk in with item != bright kid walks into here one day with that item.
mircea_popescu: asciilifeform likely explanation being "whisperer", ie, technical quimby sinking it behind the doors because "it's fucking stupid".
mircea_popescu: asciilifeform consider also that if a, b, c, d, e, f are rng words, then (P xor a) xor b) etc does not in fact substract anything.
mircea_popescu: the reason being that it wasn't gonna move till i baked it in some pie.
mircea_popescu: right. anyway, back to the practicals -- are sbox outputs ~equiprobable~ ?
mircea_popescu: asciilifeform the objection you had, you know, "this is trivially true but is not what we want" goes very much to the core of the issue -- cryptographic notions of strength are very hard to meaningfully describe mathematically because htey're always "obviously this has ONE trivial solution, but does it have more than the one ?!" math deals in 0 and 1s, it's ill equipped to deal with this batshit nonsense.
mircea_popescu: the function which takes all numbers <4 to themselves and all numbers >4 to 4.