But what I amaze is, when a same chip is rejected by the programmer yesterday, it is getting programmed today. I thought the problem may be due to, as you all said, the programmer, the programmer software, the ZIF socket, but I tried different programmers also

My BB programmers happily program a chip even if you specify another which has the same algorithm after you say "yes" to override. Are you saying that your programmer will not program a chip if it can not read the chip ID.

Anyhow e.g. the Winbonds I use only about half will read the correct chip ID and I guess this is due to the current used to read the chip ID being so low that it can not penetrate the bit of oxidation that is always there.

Erik