755400+ entries in 0.494s
Neil: <mircea_popescu> Neil since you mentioned jtd, you ever read
the original mpoe prospectus ? << I have, probably
twice. What's
the connection?
Neil: Damn
these blocks getting scarce
dub: fuck you bitcoin, fuck you in
the goat ass
nubbins`: so
there's
this escrow guy on
the forums, saltyspitoon
davout: ;;later
tell mircea_popescu deposits pl0x
MisterE: looks like Stamp is
the new king
BingoBoingo: Like Blockchain.info when from being useful, cultured manure
to straight up salmonella liquishits
MisterE: the start of
this down
trend
MisterE: last week
they were
the big news on Bloomberg
VanCleef: i'll remember
that for next
time mats
Mats_cd03: whatever
the
tx fee in
the ref client is, double it, problem solved
kakobrekla: meh, blockchain buys rtbtc, everything goes
to shit.
cads: VanCleef: a
test network holds no market value so it'll confirm quick but mean zip
VanCleef: i sat with a guy for 8 hours at starbucks once and i ran out of
things
to
talk about about an hour
VanCleef: its just awkward when doing a face
trade with someone and
they ask if
they can sleep over while waiting for it
to confirm
cads: VanCleef: sounds like you want
to set up a kind of
test network?
cads: whatever
that means :)
cads: I want
to see a cryptocurrency backed in automated service provider contracts
Duffer1: riecoin's pow attempts
to find prime sextuplets
to generate absurdly long prime numbers
VanCleef: is it possible
to have a alt coin
that doesn't require any miners
to verify it? once its sent its instantly confirmed?
BingoBoingo: NMC was
thought
to do useful shit, but miners won't commit
transactions with useful data
cads: same here but I have
to say it's intriguing
Duffer1: ya, i must disagree with
that position
cads: MP's conjecture is
that it must
trade off utility between Pow and
the secondary function
Duffer1: primecoin has already proven
that hashing can produce utility beyond PoW
cads: for a really weaksauce example, say I
take
the square of a number x. In
the result I've also computed
the square of
the number -x, and
this for 'free'.
cads: Duffer1: I do feel like we're arguing more from philosophical/economical principle
that computation shouldn't give us
two
things for free - but
this seems false.
Duffer1: cads read
the Reicoin whitepaper you'd like it ^.^
Duffer1: MP read
the Primecoin whitepapaer you'd like it
cads: mircea_popescu: we'll have an epsilon for each
target problem, and maybe
the cost of solving al
the
target problems starting from A must be greater
than or equal
to solving B, C, D... seperately.
MisterE: not
that I have ever done
that
to a cat
MisterE: Whale is swinging it around like a cat by its
tail
cads: suppose we have algorithm A, and we have
that a solution of A reduces polynomially solutions B, C, D, E, F, G...
mircea_popescu: "turn
this wheel and an egg will fall out every revolution,
then look if any eggs have feathers attached" is not okay
mircea_popescu: "turn
this wheel and an egg will fall out every revolution and also a feather every
ten revolutions" is okay, whereas
mircea_popescu: byt
the idea here is
that we promise some sort of result in
the future.
mircea_popescu: you can in principle sift
through
the history and find convenient cases for any particular problem
cads: that's
the only place for
the e
to come from, isn't it?
cads: mircea_popescu:
to formalize
this completely we'll have
to assume something like a polynomial
time algorithm
that converts
the output of algorithm A into a solution for problem B.
cads: "Hypothesis: if an algorithm's output provides proof of 1 unit of work, and
that output is also usable
to provide
the equivalent of epsilon units of useful work for another problem,
then
the energy expenditure of
the algorithm must be greater or equal
to 1 + epsilon units of work."
mircea_popescu: i suppose actually my lemma may end up
the equivalent of quantum indeterminacy, but for maths.
Duffer1: afaik
the outcome of either cannot be guaranteed
to yield a useful result beyond PoW
BingoBoingo: mircea_popescu: Well, most minds have
their head splints set
to promote
that nowadays.
Duffer1: Riecoin is
the one for large primes
BingoBoingo: cads: Primecoin isn't even
that cool at finding useful primes
cads: mircea_popescu: I feel
the same here, for
the opposite side
BingoBoingo: mircea_popescu: Your lemma is sensical so long as it is restricted
to hard POW's
cads: mircea_popescu: I
think we should be able
to
test
the hypothesis directly against
the primecoin if we
try hard enough
mircea_popescu: to me it's quite straightforwardly obvious but im much
too wasted
to be able
to show it.
BingoBoingo: In a world with Auroa coin and Mazacoin...
THere is a good chance random graphs might mine better in
tard efforts at implementation
cads: mircea_popescu: I don't feel it mathematically yet but I feel why we'd like
this
to hold, economically
mircea_popescu: ie, if on average your useless-pow block costs 1,
the useful-pow will cost 1+ε, where ε >=
the useful
BingoBoingo: I mean for
the first set you could probably encode information much as RSA does. I don't have enough vodka on hand
to speculate if
that would be a good idea. Most of Elliptic curve cryptography works on spaces
that can be played with in graph
theory world
though.
mircea_popescu: should an algorithm exist
to resolve
the problem of
ts on randomly distributed graphs, an application of
that algorithm on a selected set of graphs with a useful application will cost more
than
the normal by an ammount at least equal
to
the cost of computing
the useful application.
BingoBoingo: For most naive attempts at implementations
the second set would likely be simpler and cheaper
than
the first
cads: BingoBoingo: right, I'm
to
tired
to do
that rigorously or even
tell if why I said even makes sense.
mircea_popescu: cads not at all. my lemma is more general, in your proposed
terms it would be :
BingoBoingo: Also ironic, how pissed would
the whales be if
they knew we used
the same word for our losers and
their dicks.
cads: well lets see a better example. Problem 1 makes us solve
the
travelling salesman problem on random graphs. Problem 2 makes us solve
travelling salesman on maps
that appear indistinguishable from
the first ones, but in fact encode valuable information in
their solutions. For your hypothesis
to be
true, it should be impossible for
the second problem
to cost less energy
than
the first?
BingoBoingo: Indeed. Most of
the really big breaks in computational
theory cam from
the women before
the dorks realized
they could make
their own fantasy worlds.
mircea_popescu: a bit ironic because
the poor woman suffered plenty
through being a woman. ended up
teaching under hilbert's name because
the sort of boneheads roaming about end of 19th century couldn't have a woman colleague
BingoBoingo: mircea_popescu: Once
they hit
the crystal... It's just safest
to assign
them
the pronoun
that inspires
the least sexual interest.
mircea_popescu: cads no but
the key is your "or".
turn it
to "and" and see.
cads: i don't see it because regardless of
the supposed usefullness,
the algorithm still uses 2^k energy.
BingoBoingo: The rejected hashes might already contain
that play.
cads: mircea_popescu: it's without question
that anyone can look at
the outputs of
the
two algorithms and verify
that ~2^k energy was expended. Perhaps
the problem happens when we require
the second algorithm
to also guarantee a useful output relative
to
the
third party.
mircea_popescu: i imagine after enough btc blocks, we'll actually get shakespeare's merchant of venice in
the hashes, so why not
mircea_popescu: cads but if it's purely coincidental
then you don't have a pow
that does protein folding, you just have a pow
that occasionally folds a protein
mircea_popescu: anyway, it's clearly interesting. or as
the romanian expression goes, "un prostovan arunca o piatra in apa si zece filosofi se chinuie s-o scoata"
cads: I feel
the value supposition must be completely independent and coincidental
to
the energy requirements. You could even suppose it is pure coincidence
the output of
the second algorithm is structured such
that it's useful
to
the
third party.
cads: Now suppose both solutions are verifiably by a
third party in polynomial
time. But suppose additionally
that
the answer for
the second problem also has a value of $100
to
the
third party.
The question is should
the second process
take more energy
to compute?
mircea_popescu: BingoBoingo no we want like emmy noether for
this shit
cads: let me offer a counter "you can solve a partial hash collision problem (something like an O(2^k) expected running
time), or you can solve a np-complete problem requiring brute searching a binary
tree of height k (assume similar expected chance of success)."
BingoBoingo: Shame Erdos and his bog bag of meth are out of
the picture
mircea_popescu: and actually
this could probably be expanded into interesting results for ppl
trying
to understand various higher level maths. "define a comutative algebra in which a ring can exist for which any addition also calculates a multiplication and vice-versa"
mircea_popescu: im probably restating well known and anciently established mathematical fact, but here, mp's lemma : any computation of
two results can be reduced
to
two computations of
the respective results, and
the sum of
the energy consumed by
the latter will be at most equal
to
the energy consumed by
the former process.
mircea_popescu: now. does
this mean your key will simply have
to be
twice as long for equal security ?
cads: this would be neat
to model, and may well be
true in a very
theormodynamic sense
cads: mircea_popescu>
this could be understood on a macro level as "if
there are outside constraints on your pow process, you'll need
that many more iterations"
mircea_popescu: no but you're math-literate, how would
this work ? you compute
two
things with
the same stone ? what's
the idea here ?
mircea_popescu: im not debating whether it's interesting, it's definitely interesting. i'm just saying it;'s stupid. stupid
things are usually interesting, at least as amusements
cads: the author does provide me with say 30 bits of novel information (I had not known of prime coin -
the paper is interesting)
mircea_popescu: this could be understood on a macro level as "if
there are outside constraints on your pow process, you'll need
that many more iterations"