The Petroff Debate:

Oct 6, 2012, 4:55 PM |

While it's been pretty well known the Petroff is an effective defense for holding the draw, I've long held the stance that it is even more effective in computer assisted or 'centaur' games. I didn't just decide this out of the blue, but rather I came to the conclusion after logging many hours in ICCF games, pulling my hair out over trying to find an opening advantage against it. Unlike the Sicilian where both sides try to win and you get complicated positions from the chaos, the Petroff sends the message "I want to draw" and there's little white can do to avoid it when black is properly booked out.

My first demonstration of the drawing power of the Petroff was against PrawnEatsPrawn's 'beast' computer a year ago. He has put together a 4Ghz overclocked 6-core CPU PC with something like 12 gigs of ram at the time, and had gone on a winning streak of several games in a row against other centaurs. I informed him that on ICCF, merely having a beast computer wouldn't get the job done nearly as well. I also stated I was quite confident he wouldn't beat me in a centaur match, even though his computer was a full 4 times faster than my old PC was. At this point, he'd won every centaur game he'd been in, so his swelled head couldn't resist the challenge. The moment I played the Petroff, he immediately became sour and complained in chat during the game. I reminded him the point of my claim was that his computer wouldn't beat me, and that's exactly what my gameplan was based around. The draw indeed ended up quite easy to hold and was uneventful. His beast computer couldn't out-calculate established theory. Here's the game:


Next up is a game (with analysis notes this time) that came about after my stance on the Petroff was challenged by an OTB master named Francisco Guadalupe II. Although he and I have never faced over the board, I have attended tournaments in the same area as he does, so I already knew who he was and his background story. Francisco did not know me personally, and assumed I was speaking without knowledge or experience. As such, I was treated with the usual reaction I get from OTB masters: That I'm not skilled enough to know what I'm talking about; that a master will easily outplay a novice even with computers; that a master can make use of human concepts in key middlegame or endgame positions that computers fail to grasp, and so on. You know, the usual pre-conceived notions that most masters still believe about computer chess, which in fact are not anywhere near as potent concepts in this day and age. Multi-core PCs running the latest Houdini or Rybka to ply depths pushing 40 in complex positions have significantly reduced anything a master can contribute to a centaur game that a reasonably strong club player cannot also find. I've written quite extensively on the subject in threads here on, and I may devote a blog to reposting that work.

Continuing on, Francisco felt that he could use his OTB mastery to overcome any stagnant nature of the Petroff in a centaur game. I accepted the challenge and we played it here on While the result was a draw in the end as I predicted, I have to praise Francisco's efforts. Had it not been for my prior experience on ICCF, I very well could have found myself losing the game. Below is the game with notes on the key moments: