I didn't stop analysis too early. I understood, as did the quoted poster, that you are indeed winning. The comment was that you lost material in the process where the b4 suggestion keeps you with the same material and the same winning chances.
Ok, guess it was a misunderstanding. White does technically win material at the end, so a loss of material was only temporary.
Yes sometimes
I didn't stop analysis too early. I understood, as did the quoted poster, that you are indeed winning. The comment was that you lost material in the process where the b4 suggestion keeps you with the same material and the same winning chances.
Ok, guess it was a misunderstanding. White does technically win material at the end, so a loss of material was only temporary.
Yes, white is pretty much winning in any line. The main thing is that black could have postponed the inevitable longer. This is just from a quick engine analysis after 27. bxa4, with better play from black.
Again, you are still up material and you will end up with even more material. It just will drag on a bit longer. The b4 line might have forced the issue more quickly (assuming no mistakes were made following that 
)
(not sure what happened to the numbering there 
)
Sometimes they do, but very rarely. For example in a Billion Tactic Boy game I analyzed I was confused at how Fritz 12 missed Rxf6, then I fed it to the engine (have to computer check to make sure you aren't advising bull tactics) and it crawled up from 0.10 to .80 and eventually over two pawns.
That's because BTB has a higher tactics rating than an engine, duh.
try to let fritz play black and try it on him see how he defends against it
Maybe this is the horizon effect, the engine chooses a move but it evaluates another one better after it has already been made because it can "see" now one ply deeper.
I had one game where Stockfish suggested a move but after making my move the evaluation suddenly became much higher. After taking back, it preferred my move because the final evaluation was already in the hash tables.
Such examples are rare, the stronger the engine, the harder it is to find them. I think, most likely it is for a program to miss a tactical exchange which gives a strategical advantage in endgame. Humans evaluate some endgames at first glance but engines need huge depths (40 or more) to come to the proper conlusions.
I think engines pass up some tactics for better tactics and mating sequences...
In that case, the evaluation would not bump up after making another move than the first line.
Here, we are discussing rare cases when engines think one move is the best (giving evaluation "x") but after seeing another move they "change their mind" and evaluate the resulting position higher than "x"
I didn't stop analysis too early. I understood, as did the quoted poster, that you are indeed winning. The comment was that you lost material in the process where the b4 suggestion keeps you with the same material and the same winning chances.