On the subject of computer analysis, Anand pointed out the need to classify positions into three broad categories. In the first category, the computers find the best moves quickly. But a good human player who has seen the analysis can reproduce it under tournament conditions.
The second category involves positions where some key moves may be impossible to find under tournament constraints. So those key moves need to be remembered. The third category consists of positions where very deeply calculated) key moves are required and cannot be reproduced unaided.
The second category requires conscious memorisation while the third category is all about memory. Anand says he will play through positions in the third category multiple times to thoroughly familiarise himself. He also says that after thorough familiarisation, irrational positions eventually start making sense and the application of pure memory is reduced.
There was no need for this sort of classification and meta-analysis in the pre-computer era. Oddly, the irrational moves that must simply be remembered are as likely to pop up in simplified endgames, though most people think in terms of opening analysis only.
One thing is clear. Despite computers, improving your chess skills is still all about riyaaz. The grandmasters do their 10,000 hours. The super GMs probably do the same kind of time but they do it smarter or they have greater aptitude. The work has expanded of course because the machines allow a much broader and deeper scope of work. Engine analysis of many positions with reduced material have thrown up new insights.
The diagram, WHITE TO PLAY (Kramnik Vs Tomashevsky, Tal Memorial 2012) is an instance of an ending, which can be won but it requires very sharp tactics. Even Kramnik can’t find everything over the board.
42.h5+ Kxh5 43.Nf5 Nd2+ 44.Ke2 Rd7 45.gxf6 gxf6 46.Rxf6 Kg4 47.Nh6+ Kg3 48.Rf8 Ne4 49.Rg8+ Kh3 50.Ng4 Kh4. So far so good. White has forced united passers and cut off the king. He goes wrong in the next stage.
51.Ne5 Ra7 52.Nf3+ Kh5 53.Kd3 Nf2+ 54.Kd4 Ra4+ 55.Kd5 Ra5+ 56.Ke6 Ng4 57.e4 Ra6+ 58.Ke7 Ra7+ 59.Kd6 Ra6+ 60.Kc7 h6.
It could have been better to march the king on the f-file protecting it from checks via the pawns.
61.e5 Nf6! 62.Rd8 Kg4 63.Nd4 Ra7+ 64.Kd6 Ne4+ 65.Kd5 Nc3+ 66.Kc4 Kxf4. Now it's close to drawn. Play continued 67.e6 Rc7+ 68.Kd3 Ke5? Something like 68. --Na4 draws. This falls into a horrible trap after 69.Rd7! Nd5 70.e7 Rc3+ 71.Kd2 Rc8 72.Nc6+ Ke6 73.Rxd5 Rxc6 74.e8Q+ Kxd5 75.Kd3 Re6 76.Qb5+ Kd6 77.Kd4 Ke7 78.Qf5 Rf6 79.Qh7+ Kf8 80.Ke5 Ra6 81.Qb7 Rg6 82.Qh7 Ra6 83.Qd3 (1-0). Mate in 23 according to the computer - white will grab the h6 pawn and then the rook, etc.
Devangshu Datta is an internationally rated chess and correspondence chess player