Sicilian Defense: French Variation - a question

Hi all. In my latest game we played the Sicilian Defense: French Variation (B40), after 1. e4 c5, 2. Nf3 e6, and things were fine. I, as White, played 3. d4, and Black answered with the rather unusual 3...b6. I responded with 4. c4. This move is, according to Chess.com's Computer Analysis a mistake, and looking in the Game Explorer, I see this is true, since the Game Explorer shows 4. c4 to be the move that gives Black the best chances of winning (except for the moves played only once, and lost by Black). But here is the strange thing: according to the Computer Analysis, 4. d5 would have been much better, but in the Game Explorer there is only one game with this reply, and that game was won by Black. The most common replies to 3...b6 are 4. b3 and 4. c3. So, now I'm utterly confused! Should I trust the Computer Analysis, or is 4. d5 really not such a good move? Is 4. d5 really better than 4. c5? If 4. d5 isn't a good move, then Chess.com's Computer Analysis is highly suspect. Please enlighten me!

I'm too lazy to boot up Fritz/Houdini and check out the line, but many have complained about the Computer Analysis before as doing a less than optimum job.  The engine is apparently not all that strong.  Houdini 1.5 is a free download and a much stronger engine for analysis.  There are a bunch of free programs to use it as well, you don't need to buy Fritz.  Try searching the forum archives (upper right hand corner) for "free software" and get plenty of advice.

Thanks. :) But I still hope someone can explain why 4. c5 is such a bad move, and if 4. d5 really is better.

I don't think 4.c4 is a bad move ; it's just that 4.d5 gives a Ben-Oni pawn structure which classical players assess as slightly favourable for white (the d5 pawn confers a space advantage).

With 4.c4, you'll probably end up playing some Hedgehog structure with chances for both sides - ie. 4...cxd4 5.Nxd4 Bb7 6.Nc3 d6 etc.