This question refers to Both versions of the type of
chess 'Rasputin' can play.
The worst-case scenario, I guess, would be for your
computer-program to keep on playing the same "losing-
line" {esp. in regular chess} ad infinitum, which can
easily be determined from the "Results" page, and from
which a number of Caissans' have in times past, padded
to greatly padded their online-rating, as a result.
 
I suppose that a randomly-inserted change {Re. a losing
line-of-play} is better than nothing, but ideally, IMHO,
Rasputin should eventually evolve into as strong a pro-
gram as possible, in Both versions of chess!

How do you propose to make Rasputin learn? How would that be programmed in? Are you referring purely to opening lines or would the code change? Do you know how difficult it is to program a chess engine? In the case of Chess960, once a certain way of winning is discovered from a certain starting position, how would Rasputin learn not to go down that road?

..Since I'm far from being any sort of computer-versed "geek,"
I must've grossly under-estimated the amount of effort re-
quired in re-programming most game-playing computers, which
given the present level of A.I. {artificial intelligence} finding
the corrections to innumerable inferior-to-losing, lines-of-play
{especially, in Chess-960} it would ultimately require a Whole
Lot of 'hand-holding' {I mean, cooperation} between a superior
chess-savvy team of experts, and computer-code "geeks"!
 
The fact though, that in the world of top-level computer-chess-
machines, that specialized news-outlets like "Chessbase.com"
seem to imply in their reportage that every one or two years
the top, or favored-to-win chess-engines become progressively
'stronger' in their search-ply-depth, and/or more narrow and
harder-to-gauge 'chess sense', perhaps adds to this misperception!

There's no way to make the program "learn" so-to-speak, but we will be extending the functionality of the computer opponents in the future to allow for you to choose their rating level. Also options to play certain openings etc. Any suggestions on how to improve the computer opponents are welcome.

I thought that was already what you were trying to do by having two computer oppenents, AgentSmith and Rasputin. Are we talking simply of adding more to cover a wider range of ratings, or is this something else you're planning to do?