asciilifeform: and to this day afaik nobody has anything like a sybilproof way to glue 1024 unreliable chickens into a simulacrum of a pc

asciilifeform: plugging the funnel at wrong end, so to speak.

asciilifeform: possibly we had this thread, but asciilifeform's unpublished, failed stab 'at bitcoin' decade+ ago was , more or less, 'uci running martianbank'

asciilifeform: ( usaschwitz inmate could live 100yrs without once seeing a live chicken )

asciilifeform: modern equiv prolly 'bike thief'

asciilifeform: 'first, steal a chicken..'

asciilifeform: this uci thing, is one of those items that is wholly trivial but for ONE unsolved boojum, quite like, e.g., roomtempsupercon

asciilifeform: upclose an' personal

asciilifeform: i fully expect that most 'uci' install would be by hand.

asciilifeform: and during this, thought 'you could easily get folx installing 'virus' wholly voluntarily, if it eliminated this & other crapola in a silent-to-brass way'

asciilifeform: when asciilifeform toiled in usg salt mine, on some days the Official shitbox would be ~unusable: ~no cycles left over from 'antivirus'

asciilifeform: mircea_popescu: right but they sell it at, what , 1000x margin.

asciilifeform: aha

asciilifeform: ( 'lease space inside nostril' aws et al also exists, uninterestingly )

asciilifeform: lol, that happens as often as buying 100 $ of waterfront

asciilifeform: what means 'by-capital' ?

asciilifeform: just that neither asciilifeform nor anybody, afaik, else, knows how.

asciilifeform: not that it's necessarily physical impossibility.

asciilifeform: ( 1024 chickens!11 )

asciilifeform: the ancient dream of the fungible cpu cycle keeps coming back. but sadly still - a dream

asciilifeform: that's the thing with sybils: what you thought were independents, turn out to be an aggregate

asciilifeform: depends in which way unreliable. e.g. say they all , on command, simultaneously reveal to be d00dz and pick up rifles. then battalion, whereas yesterday - brothel

asciilifeform: ( even to the extent of whether it happens )

asciilifeform: hey this worx great so long as the secrecy of the contents, or the reliability of the computation, dun matter at all

asciilifeform: http://btcbase.org/log/2017-07-12#1681717 is how you get http://btcbase.org/log/2017-07-12#1681713 , every time ☝︎☝︎

asciilifeform: http://btcbase.org/log/2017-07-12#1681720 << unfortunately-- wrong; sure does pay. hence the few thou. sybils currently masquerading as btc nodes ☝︎

asciilifeform: ( 'but there was a protected type!' alert reader will ask. the pill is the obvious one, i will reveal it later. )

asciilifeform: interestingly this gives ~0 performance boost. gcc optimizer, turns out, is notbad.jpg.

asciilifeform: mod6 et al : in other noose, i have a working experimental build with No_Implicit_Conditionals, No_Implicit_Heap_Allocations, No_Implicit_Dynamic_Code, No_Secondary_Stack, No_Exception_Propagation, No_Tasking, No_Protected_Types, No_Delay, No_Allocators, No_Dispatch restrictions. ☟︎

asciilifeform: sect. 3 in particular.

asciilifeform: mod6 et al : http://docs.adacore.com/gnathie_ug-docs/html/gnathie_ug/gnathie_ug.html is useful

asciilifeform: oooh upstack, http://btcbase.org/log/2017-07-11#1681525 >> '-gnatD' gives MUCH moar readable per-file ast ☝︎

asciilifeform: ftr!

asciilifeform: and 71 of 72, not 63/64 lol

asciilifeform: 9536 not 32 lol

asciilifeform: deeply nfi

asciilifeform: ( well-known crackpot chix's www btw )

asciilifeform: full text, with englischization

asciilifeform: http://freya.theladyofthelabyrinth.com/?page_id=512 << heeeere we go.

asciilifeform: '... and the man who cannot use a crow like this, does not deserve to find one'

asciilifeform: (unless you speak .no)

asciilifeform: well, 2nd link. but really you want the text, of this old folk song.

asciilifeform: grr

asciilifeform: !~google krakevisa

asciilifeform: ( it is normally ~impossible to actually use whole thing, because of routing constraints )

asciilifeform: we use 63 of the 64 available LUTs.

asciilifeform: initially targeted the '9532' but it dun even fit the uart. so it had to be the next size up.

asciilifeform: i deliberately wrote FG for the smallest commercially available cpld that would hold vonneumann + uart

asciilifeform: you can devise others, in principle, that use however many bits of flipflop you got to work with.

asciilifeform: a 'value of pi' needs a couplea dozen bits, at least

asciilifeform: von neumann's is the smallest that's worth anything

asciilifeform: *big

asciilifeform: q is , how bit

asciilifeform: well they're all statemachines

asciilifeform: even ~counting to 100 per se~ needs 7+ bits...

asciilifeform: mircea_popescu: for just about any statistical test ( other than von neumann in/out proportion ) you need a buffer.

asciilifeform: ( how the constant here follows from the circuit -- is an exercise for the reader )

asciilifeform: we ain't got any tests in FG, other than 'von neumann filter has not emitted 8 bits in past 0.5msec -- lamp is red'

asciilifeform: but rather tests a proposed test of entropy.

asciilifeform: as i understand, this doesn't test entropy

asciilifeform: so naturally it's moar 'a'

asciilifeform: mircea_popescu: b, as i understand, isn't even physically possible without a buffer

asciilifeform: ( for the intrepid: see http://btcbase.org/patches/fg-genesis#L276 , http://btcbase.org/patches/fg-genesis#L378 , http://btcbase.org/patches/fg-genesis#L475 . )

asciilifeform: ( recall the sheet of paper that came with your FG. )

asciilifeform: so we actually have it, in there, nao.

asciilifeform: eventually i derived that this is equivalent to the equation seen in the 'watchdog' circuit in fg.v

asciilifeform: i.e. a lamp whose brightness is proportional to how close (bits emitted from vonneumann filter per sec) / ( bits entering same, per sec) is to 1

asciilifeform: mircea_popescu: i actually considered something that is arguably a variant of your test, when designing FG,

asciilifeform: e.g., aes(0, 1, 2, ...., maxint) is perfectly uniform by all known tests.

asciilifeform: passfail or 1024bit number, dun matter.

asciilifeform: any test whatsoever, past or future.

asciilifeform: ( statistical tests only reveal broken rng , but never a working one )

asciilifeform: dijkstra's 'testing reveals presence of bugs, but not absence' applies just the same to rng.

asciilifeform: aha.

asciilifeform: even moar outrageously, adding the 100 1s does not in fact 'turn it bad', the P of these 100 1s appearing in a uniform distrib is >0

asciilifeform: there is no mechanical test that, e.g., 'bitflipped digits of pi' does not also pass.

asciilifeform: the possibly paradoxical answer, afaik, is there does not actually exist such a thing as good entropy. only bad and worse.

asciilifeform: traced operand flow

asciilifeform: iirc the d00d who found the koch whitening lulzgem used a proggy that worked quite like 'barium enema'

asciilifeform: btw, fun project for FG owners -- connect the analogue hoses to something other than the included boards, and see what you get

asciilifeform: incidentally if there's a vonneuman filter in the pipe, setting consecutive bits does 0

asciilifeform: so this's a 'meta' test of a given dh-style test ?

asciilifeform: aite, and what to do with M and M' ?

asciilifeform: what's that do ?

asciilifeform: and then we take Pth to P+100th bits and.. what then

asciilifeform: fills 1048576 bits with rngolade?

asciilifeform: what's FG[1Mb] do ?

asciilifeform: i have nfi what 'rnd(0, 1Mb)' does

asciilifeform unable to parse an implementable algo out of the above

asciilifeform: ( for what does this operation use the original values that came out of the rng ? )

asciilifeform: what means 'flip a number of consecutive bits to 1' ?

asciilifeform: 'take a FG string... 100bits'

asciilifeform: mircea_popescu: i'd like to see an expanded algo for ^

asciilifeform: ...but confesses to tape-painting bot

asciilifeform: elsewhere, 'karpeles pleads notguilty'

asciilifeform: meanwhile in the land of toy crypto, https://eprint.iacr.org/2017/655 >> '...Inmarsat satellite phones... the 64-bit encryption-key... recovered in around 0.02s on average'

asciilifeform has not tried $subj

asciilifeform: mod6 et al ^

asciilifeform: http://gcc-melt.org << of potential trbological interest.

asciilifeform: !!up fromdeedbot