Quote from: Bayes on April 11, 2013, 09:25:51 PM
Isn't that basically what your "Tool" and Gizmo's "gizmo" is all about?
Indeed. Automating the process is all it is about. (and becoming more efficient at processing of course).
A human is limited as to the amount of real-time processing "power" he can exercise in between spins. An AI agent is only limited by the amount of resources allocated to it.
A human might for instance calculate a set of 4, 5 wheel-based triggers, maybe more (I doubt realistically someone could go beyond calculating double-digit triggers right, spin after spin after spin; the more demanding the more "ground" the brain becomes to perform over spins)... an AI agent on the other hand can be aware of hundreds or thousands of triggers and make calculations (i.e. SD and others). It does make a difference in spotting where the current opportunities lie. Assuming you grant there is a timeline which can be attributed to the random stream and accept the triggers/selections have cycles of activity and inactivity, for which the bettor's work is identifying the profitable phases as accurate as possibly and "piggybacking" on them while they last. The more triggers/selections processed, the more possibilities. A human can only see (process) a small "chunk" of the actual possibilities developing at all times on the many triggers. An AI agent can see more and hence has more to grab from (and jump to) due to the many timelines derived from the triggers being tracked.
Quote from: Bayes on April 11, 2013, 09:25:51 PMSpot-on dear friend.
IMO any method can be implemented on a computer, no matter how complex. Even intuition can be coded as long as you can express how you come to your decisions even somewhat fuzzily.
I guess in the end, all that matters is whether AI is able to deliver a better hit rate than random.
Doesn't it?