Is Chess A Practice?

I'm still a beginner at chess, won only 1 game, against the machine, 3 draws, and lost my first game here, though I don't know how she did it. Also some of the posts I've read here suggest that it certainly is not something fixed. You get good, then read some books, only to start loosing. It seems to be like a practice, like a doctor practicing medicine. Every game is so different from each other. There was a story of the recent world chess 2016 players on TV, the interviewer asked one of the top guy's, have you ever lost. He said the other day he lost to some one on line, anonymously. Didn't even know who he or she was, and he is a chess master. Si I suppose that no matter how long you've been playing, you can always get 'better'?

Free software running on an average personal desktop computer could beat Deep Blue even though DB calculated much faster. Just saying.

Yeah, there's always room for improvement. Also, just like with any skill, people's performance fluctuates. If you're within a few 100 points of someone, it's common to be able to win a single game off of them, even if they would easily win a 10 game match.

Regarding the depth of chess, sometimes at the very end of a game, if you could change the position so that just 1 piece or pawn were on a slightly different square, or if you changed it so it's the other player's turn to move, it can sometimes change the result from a win to a draw or loss. No person or machine can see/plan that far into the game. So essentially what you have when two players of nearly equal skill play, is they're making their best guess. Yes even world champions and super computers can lose games because sometimes their highly educated or highly calculated guess is incorrect in the end.

Since it's impossible to know all the consequences of a move, it's played with a mix of deliberate and emergent strategy. Usually mini-plans consisting of things like improving a piece, gaining space, weakening a square, weakening the pawn structure are carried out. Humans frame the mini plans in the context of general strategy and endgames (so we generally know they are reasonable moves many dozens of moves deep without even having to calculate). Then, as the game progresses, we can react to new developments and change plans as necessary.

The more types of positions you learn, the more games you've played and analyzed, in short: the larger your personal database of patterns and ideas is, the better you can play. In the beginning, you might learn basics like "isolated pawns are weak" then later you learn some exceptions, then later even exceptions to the exceptions and so on. Also, elements can compensate for each other for example you accept a pawn weakness for a lead in development, then later trade the development for material, and maybe later still sacrifice the material for initiative.

One interesting aspect is the idea of practical decisions. For example a position may be objectively equal, but it may be much easier for one side to play. This is an aspect of competitive chess that computers are oblivious to.

I could be wrong but I'm pretty sure that today's software running on an average desktop is now many many times faster than deep blue was. It's certainly true that DB was purely brute force with very limited positional knowledge.

I thought were were up to about 100 million per second on personal computers. That is what the first deep blue did. 2nd match was 200 million per second.

I may be wrong about modern computers though. I suppose it does change every year. I'd be surprised if it's much higher though. The playing strength is definitely much higher.

There are all sorts of positional values given to the engine. For example things like for each pawn that comes off the board decrease the value of knights +0.05 (I made that number up just for an example).  I don't know about stockfish, but for example IM Kaufman worked closely with the Rybka development team. Then they run huge batches of games, change the numbers a little, and repeat to look for any improvements.

Also the search function has improved... it ignores bad moves faster. Chess is still far too difficult to calculate raw. Efficiently ignoring bad branches and cutting them has improved.

Which is why some engines are better at e.g. finding tactics. Some deep puzzles (that involve an early queen sac for example) may have an engine totally ignoring the queen sac after it doesn't see any point after looking a few moves deep. You could let it run an hour and it may still totally ignore the winning line.

I'm amazed at how I'm check mated. I literally don't even see it coming, even though I have the lap top chess and an actual board to play out moves ahead to see what could happen. Also amazed at how it backs away from trading pieces, and others, gives up the Queen, though it didn't have to. As for me it works beings how I don't have an actual person in front of me playing. But for now I find it enjoyably fascinating, and a good mental workout. 

yes,yes