Also regarding bribery: if an organization does an economy of scale bribe, such as say if the mafia bribes the police commissioner instead of each individual shady type bribing individual policemen, that wouldn't work. Competition would rear its ugly head. Even though the whole force was looking the other way by illicit influence from the top, individual policemen could risk being unusually upstanding and by-the-book--unless given an individual bribe as well. Lower rung criminals who were part of the larger enterprise would compete with each other to bribe lower level law enforcement more than the next guy as well.
]]>Does an ATHENA-style state of non-self-identifying/selfhood projection occur in humans, possibly as the result of brain damage?
If so, is there a name for the condition?
Sounds like something Oliver Sacks would have written about.
]]>Early in the history of Bitcoin, a very large transaction occurred and immediately after this, portions of this whopper of a transfer got repeatedly passed back and forth between a few of the participants, in a manner scarily like that employed to obfuscate who owns what in a conventional financial scam.
Bitcoin isn't conventional, though, so we can conclude that whoever was doing this didn't know that the actions could be traced and that they would not obfuscate the initial transaction. This then strongly suggests criminal and not-quite-smart criminal involvement in the whole thing. The net result is that for anyone reading that article, Bitcoins tend to seem irrevocably tainted.
]]>What I would expect the thing to look like is a huge number of execution threads, each of which spends pretty much all of its time waiting for the glacially-slow humans to react to the previous nudge, or just to wake up. This pretty much means that ATHENA has to have a truly staggering amount of disk-like storage, a rather lesser amount of working RAM and probably fairly modest CPU requirements.
It would operate by having monitoring processes periodically examining each thread of execution to see if anything has changed, and only waking these threads if there has been a material change. Once woken, the full power of the AI concentrates briefly only on the two or three threads that need this attention at any one time.
ATHENA is therefore scarily believable. It doesn't need god-like levels of CPU to function, merely rather large amounts of disk and much ingenuity to construct.
]]>Erlang has very efficient thread- and event management mechanisms and relies on message-passing instead of locks so it fits well with soft-realtime distributed computing that I think one would want in ATHENA (The language was originally developed for 1980's digital telephone exchanges).
The hard part of ATHENA is the tweaking mechanism - it requires super-human understanding of biochemistry and human intelligence to tweak the psychopath's medication into something that will also make him psychotic.
OTOH - such mechanisms can perhaps be evolved, partly by accident. There exists several parasites and fungi in nature which modify the brain of their victims to get a behavior that suits the parasite's purpose.
Stuart Kaufmann's research suggests that one would need a different "computing fabric" than pure software to evolve "code"/"algorithms" because computer programs are too brittle to be "evolvable" (However, I suspect that a cat-brain-simulation that someone uploaded by mistake would perform adequately ;-)
]]>The number of things one needs to figure out from first principles is inversely related to the power of one's ability to search. ",)
]]>