When will they design Fritz to play like a human?...

They should design the next Fritz to blunder at different levels and in different ways...some tactical, some positional, at the proper frequency.  That would be a great new feature. The way it is now is not realistic.  That would be the best way to improve Fritz.

You can put Fritz on the handicap levels and it will blunder at different times..

Yeah but it does not blunder like a human...

I have read somewhere that it is very hard for programmers to code chess engines so they emulate human errors. One obvious problem is the absence of psychology in computer chess, and when we are aware of this it probably contributes to our feeling of playing something "non-human". Another problem is to define a typical human player. In what way does a 1500 player blunder compared to a 1700 player, and what is a typical human blunder (positional, tactical, picking the wrong plan, lack of specific chess principals, loss of tempos etc)?   

How does Fritz not "blunder" like a human being when set on lower levels and/or on low ply searches? Let's see some examples here. And for starters, pls describe the differences between "human blunders" and "Fritz blunders." Definitions of "human blunders" and "Fritz blunders" would also be very useful, since this entire discussion is  minus any facts about how humans and computers play chess or how they are different (or similar)

Don't carbon-based units blunder erratically from time to time? I sure do!

I think the Fritz team needs to assemble a huge team and survey many many games of people of all sorts of ratings.  For instance look at say 1000 games of players of all levels...from 1200 to 2400 and compile hard data on the kinds of blunders (and just "weak or pointless moves)..the depth of the blunder, the nature of the blunderetc.  For example after a team looks at 1000 games of players rated near 1700  they might find that they blunder every 27 moves, make weak positional moves of a specific nature every 15 moves, overlook serious kingside threats consisting of 3 move combos 12% of the time etc.  This info could help the Fritz programmers to program an artificial 1700 opponent that makes oversights when searching just 4 moves deep etc.  They could also program 3 or4 different styles of 1700 players.  I think it would be an interesting project that would take a lot of time and effort but would be very rewarding for chess training programs.  All the programs I have played where you set the "rating" to a specific number have always been dissapointing...they just are too fake.  For instance look at your last 20 games and count how many blunders you have made and average that per move...even if you opponent did not see it.  Say you played 20 games averaging 40 moves each...that is 800 moves.  How many times did you blunder in those 40 games?  Depends partly on your rating right?  Say you are rated 2000...maybe it is something like 17.  That is a blunder every 47 moves.  So the computer would have to keep track of its artificial 200 player and over the long run average a blunder every 47 moves...and a 200 player blunder is not usually as obvious as a 1200 players blunder...I think it is the only place for Fritz to go now..they could do it...just takes lots of work!

Fritz is human, but he is German.

Not just blunders but weak moves...risky moves...daring moves...careless...and at different depths etc.

I believe it is very easy to have Fritz produce blunders. The program should just not take into account the existence of one or more of the enemy pieces when calculating his best move. I honed this technique for years and it gives great blunders,

I think why we feel engine is lack of human play style is that:

Much of the engine algorithms are only designed to find the "best" move well , so if you want it to find the second best move , it may change the whole design of that engine. 

But to people 's style of play, the human games are full of 2nd 3rd etc best moves of each position based on our rating of chess. 

So what make the engine be like low rating people?  

The designer usually do sth like : engine calculate the best moves as usually, but it does not play that move everytime, randomly it plays a random move(much easier than play a second best move or 5th best move) ,  and if a game is 40moves long,  20moves random and 20moves best may create a AI rated about 1200 etc

So no 1200 human will do like this , half a game be kasparov and half a game be a fool to make your rating about 1200...

The problem is that people's blunders occur, because they overlook a threat, which happens randomly. A computer, however, cannot do anything randomly, it can only simulate randomness. You can implement an engine, that blunders statistically after every twentieth move, but you cannot do much better.

On top of this, a computer does not have personality. Again it can only simulate an aggressive player or a positional player. Such a simulation is never going to be as versatile as anybody's personality - at least regarding most people.

Furthermore, most designers of chess engines want them to be as good as possible, so they do not even bother to make them human.

The author of Fritz is Franz Morsch, who is Dutch.  I don't why this line gets repeated about Fritz being German.  The author of Shredder is Stefan Meyer-Kahlen, who is German.  Wouldn't Shredder be the German program?

The company selling "Fritz" is ChessBase, which is located in Hamburg.

Frans Morsch and Matias Feist work for the german company chessbase. Both are of German origin. Fritz' is as German as can be....If it was Dutch it would be Frits and not Fritz.

They did come out with a chess engine designed to imitate humans, it didn't play very well, but it blamed its losses on outside conditions!

If you set it to Sparring mode, Fritz will purposely blunder to set up tactical shots for you.

I've found that the old type dedicated chess machines play a fairly human like game as they typically only look ahead 4-6 move.  Although they make mistakes they usually aren't the horrible blunder type.  Trying to dumb down Chessmaster or Fritz was frustrating as they'd  stupidly blunder a piece and then proceed to play like a grandmaster.

Lol.

That's how i play! Blunder à piece and then proceed like à GM :-)