778400+ entries in 0.496s
cads: So
that if an adversary has captured
the state vector
they can only continue
to simulate it faithfully
cads: I came up with
the idea of encrypted computing when I was
thinking how
to design an autonomous AI agent
that cannot ever have its state vector interrogated or partially simulated.
cads: you may have
to excuse my ignorance of
the commercial or political side of
this work
cads: right, currently it's amazingly inefficient and not general, and I also believe
there are unfortunate
theoretical limits
to
the power of any such approach
cads: Which could give us some padding between
the software agent and untrusted underlying hardware.
cads: asciilifeform:
that brings us
to study methods of invalidating
the security properties by changing aspects of
the substrate system
that
the security properties gloss over or assume by fiat.
mircea_popescu: asciilifeform your dream of software
that's unauditable by
the user is getting closer and closer huh
cads: fair enough,
then even
the most correct formally proven software should expect some faults due
to underlying hardware.
cads: hmm,
the verified software I've read about uses security properties which are proven via a formal proof assistant in a standard logic.
cads: It does feel like you're concentrating
the bugs into a very rare and amazingly lucrative bug class
cads: But I'm optimistic since we can study
the
theory of what should constitute a correct and effective security property
mircea_popescu: you can now have no bugs except
that sort of bugs you'll never find
cads: And
that's
the part
that hits my stomach with dread.
cads: And I appreciate
that
this makes it /easier/
to avoid bugs -
the only place a bug can now live is in
the security properties you prove about your access model.
cads: I appreciate
that we can create a kernel of security properties and
then prove it about our access control model, and
then build
the system on
that.
cads: for example UC's Quark verified browser kernel or
the NICTA's l4.verified project.
BingoBoingo: asciilifeform: Most Land Grants had
them back in
the day.
cads: mircea_popescu: sell some Coq nerds on
the importance of implementing a formally verified blockchain algorithm and further specialization
to a btc implementation :)
benkay: sure but
then you're not really putting
the barrel
through its paces are you?
mircea_popescu: a howitzer can fire blanks on
the football field just fine
mircea_popescu: benkay you don't have
to make
the shells actually do anything you know ?
benkay: u want shells
too sir?
cads: I'd love
to see a haskell or agda implementation
mircea_popescu: asciilifeform maybe i'm blinded by zeal but seems
to me sed is actually harder.
mircea_popescu: cs professors who failed
to have
the kids implement bitcoin in 2013
mircea_popescu: "humanities" "professors"
that have no idea what a college is and can not speak latin,
mircea_popescu: so
they can go around pretending like
they're
things
they could never be.
mircea_popescu: we're fortunate
to live in a world where nobody has any clear measuring stick
to be able
to measure exactly how
tall
the pile of shameful excrement
they find
themselves under is.
benkay: most us unis i've been
to are similarly oriented around hoovering up research + dod monies asciilifeform
mircea_popescu: "what, you have not implemented bitcoin as a
term project ? you're defunded. 100%"
mircea_popescu: and i seriously do not understand what a rector must be
thinking
to not close down
their cs department after
this
Bugpowder: How
to blow 1000BTC in under 5 minutes.
benkay: mircea_popescu: cs departments are for research. mechanical and ee departments are apparently where you learn
to actually weld shit
together.
mircea_popescu: this'd have been anyone's
term project if i were
teaching.
benkay: mircea_popescu: it's a goddamn shame
that everyone's leaving it
to
the self-taught hacks
Bugpowder: man, gox back down
to where it started... what a derp
mircea_popescu: as in a sane world no uni with a cs department could have done anything else in 2013 as a
term paper
mircea_popescu: there are 5k universities proposing
they have cs departments, which is a fraud
mircea_popescu: yet nobody went "ok,
the ideas are quite clear, let's do
this
then"
mircea_popescu: one
thing i can't for
the fucking life of me understand is why on earth has nobody made an actual competing implementation.
mircea_popescu: well yeah in ashell environment. not quite
that much flexibility in bitcoin scrupting
mircea_popescu: asciilifeform
the diference here being of course
that you're not expected
to run
the whole
turd
copumpkin: I'm a big fan of
the
tune in sonmi-451 meets chang, and
the various
times it recurs
throughout
the rest of
the soundtrack
cads: copumpkin: you're right,
this is good
BingoBoingo: Mark's face is shaped in such a way, I imagine, he imagines he can just hug away
the scam.
BingoBoingo: Well,
then
there is
the
third mistake, scam.
BingoBoingo: mircea_popescu: Well so far it is at least
two mistakes on Gox's part leading zeros and
txid as database primary key.
cads: if a soundtrack is good I usually watch
the movie
mircea_popescu: but apparently
the php lolcat interpreter adds leading 0's
to O RLY?/OIC blocks or something
cads: copumpkin: You've got me piqued on
the soundtrack
BingoBoingo: Cutting
those zeros changed
the
txid
though.
mircea_popescu: mtgox was at least 5
times
told
to stop with
the idiocy.
BingoBoingo: Anyways,
the problem in Gox's case is
they kept leading zeros in
their signatures. Everyone else was like STFU
that's dumb. Eventually some nodes started correcting
this mistake.
mircea_popescu: watch it abd blog about
the experience, i never saw it.
copumpkin: it's not ideal and
the devs are
trying
to adjust it
mircea_popescu: like
tits,
they get cold and flop arouind and your nipples hurt and everything,
benkay: the flaw is in people using
the
txid as a bookkeeping device which satoshi explicitly warned against iirc
benkay: anyways,
this malleability
thing as mentioned above is not really a flaw, cads.