that is absurd.mbabigian wrote:IDeA stands for Interactive DEep Analysis. Emphasis on "Interactive." The whole idea (pardon the pun) is that the player can guide the search by inputting moves for analysis. I routinely steer it away from dumb lines and into better ones. That is the whole point of IDeA. It then min/maxes the score from the best variation up to the root.
I can input 60 moves if you provide them and all the sub-variations you think are important. It will then examine the entire line and see if it finds any better alternatives. You can even specify a series of key positions along the line that you think are important. I can make those root moves and it will look to examine options from those specific positions. Sometimes if the position is particularly difficult for engines, you must manually refute many variations before the trees scores are correct. The KID is one such opening where I had to do just that after losing in a Freestyle tournament in that line.
In your g3 line are you suggesting 6. bxc3 Ne7 7. g3 then what are you suggesting for black?
To give you an idea of what I see for the moves, see the following image after White's move 6.
You'll note that the scores for Ne7 and Qc7 have changed since my last post as improvements where found.
After black's move 6 ... Ne7, you see:
As you can see there are not many positions analyzed (N column) beyond g3 at the moment. Although we can examine this line, I might suggest you stick with the strongest moves for white.
Remember, even if the majority of the variations lead to a win for one side, but a single line can draw, then the score will reflect the best possible moves by black giving the variation a draw score. We are not scoring "chances," but the best line found so far if neither side makes any mistakes.
If you'd like we can continue this via email where it would be easier to attach PGNs, images, etc. Send me a PM if interested and I'll reply with my email address.
Mike
g3 getting a negative score and placing only 15th or so.
it is at least the 3rd-best move, maybe even better.
similarly for f4.
I understand what the program is doing, but the problem is, how you know when to input alternative moves to the engine, and when not?
inputting just couple of alternatives here and there will not be sufficient, as this line involves some 30 plies of accurate playing, where most probably you should correct the engine on each and every ply.
concerning email correspondence, thanks a lot for the suggestion, but I will pass, I am really quite busy at the moment.
we can definitely though continue here for a while with some of the most important issues.
But its line includes Nf3 and also allows Black to exchange the light-squared bishops, which shows that the engine lacks your understanding of the position. Perhaps you will propose a reply to 7... Qc7. Or maybe one or two very short lines following 7... Qc7.