I have seen many obvious errors in the computer analysis on this site. Tactics trainer is rife with them. I am talking about things that are obviously irrational, like giving an analysis of the position as +1.03 with a best line that ends in checkmate. Since the unlimited computer analysis of my games was one of the features that swayed me in to a premium membership, I am frustrated when it tells me my move 7. 0-0?! is an mistake with an evaluation of 0.73 and it suggests the alternative 7. 0-0 is much better with a 1.14 evaluation. I have come upon these kinds of head-scratchers over and over with this site's analysis. If I can't trust it to not be blatantly irrational, how can I trust the accuracy of its evaluations? 

Yes, I believe your observations have some truth to them.  You'd be better off to download one of the widely available free chess programs and have Houdini 1.5a or Stockfish analyze your games.  The example move of "7. O-O" might be an anomaly.  Computer engines don't always evaluate openings properly which is why many come equiped with an opening book.

Hey cburress who was in the checkmate position because +1.03 means white is winning.

You think that's bad, try looking at this:

http://www.chess.com/forum/view/help-support/inaccurate-chesscom-analysis

Apparently, the Caro Kann defense is "inaccurate" which I find to be ridiculous.

@Koala8 White had the checkmate, but +1.03 evalutation means white has an advantage worth approximately an extra pawn. The convention for the engine on this site was Mat(X) when there was a forced mate. To make matters worse, it wasn't even the highest rated move, it was just the only one whose analysis line ended in checkmate.

These examples just show that there are fundamental logic errors in the analysis engines this site uses. 

Please rest assured that there is already, as we speak, a team of highly trained specialists/strategic advisors on the case.  These things take time you know.

The OP was making a serious point (with which I agree); why did you feel the need to post an inane picture that wasn't even funny?

There is an "off topic" forum for that kind of nonsense.

Those were just the errors that don't require any chess knowledge to point out. The engine frequently has some bad horizon effects. For instance in my last game I played a non-checking move that threatened a mate in 2. The only thing white can do is check for a bit and then throw away all his pieces. It was obviously winning, the engine calls it a blunder and drops my evaluation to -0.8. On the very next move it says the best line for my opponent is evaluated at -95.45. Earlier in the game there was a forced sequence with nothing but what it called errors on my part that started at an evaluation of 0 and went to -4.4 (in my favor) after 7 ply. There is always something fishy with the game evaluations.

That image may not be that far off, I have seen complaints about the engine analysis that run three years back for this site. 

@Fekajxo Thanks for pointing out those free evaluation engines. I will look in to those for the future.

Boy, talk about grousy...and anyway, I thought the picture was pretty damn funny. 

Well there is a surprise... easily amused must be a very comforting trait.

So must being an eternal grumblebunny.