We're already reaching the end of our Arkell Interview. In Part I we heard from Keith about his life in chess and Part II covered his playing style and the chess press. Today to conclude, appropriately enough, we have Keith Arkell on the endgame.
As ever, this is me and everything else will be Keith's own words.
Not long before Keith agreed to chat with us I’d played an email game where I stumbled into defending a KR v KRB ending. Although I didn’t have the slightest clue how to play these positions I did at least know that for the most part such positions are theoretically drawn. Necessity being the mother of getting off my lazy arse, I settled down for some endgame study – and one of the first things I learned was that theoretical draw notwithstanding, Keith wins this endgame time and again.
… yes it's true. I have won the ending of R+B v R 17/17 times, but I have yet to play it against a GM; although I have beaten IM Lawrence Cooper in it twice.
When I mentioned Keith's response to my fellow bloggers EJH immediately responded that getting KRB v KR on the board 17 times in a single lifetime is almost as impressive an achievement as going on to win every one of those games. He inspired me to ask a question that otherwise simply wouldn’t have occurred to me - how does Keith get the ending so much more frequently than everybody else?
I've wondered that myself. I guess that the chances are increased because I am not averse to exchanging pieces in order to maintain or play for an edge. Rather a lot of my 6000 games have wound up in an endgame. The crucial explanation may be this though: it became clear during many of my post mortems that both my opponent and I were playing for the same ending - R+B v R! This was certainly the case for example in both of the games against Lawrence Cooper, and the games v Gayson, Lewyk and Daly.
To my eyes that answer makes Keith’s perfect record in KRB v KR even more remarkable. Not only is he cocking a snook at theoretical evaluations he’s also managing to outplay his opponents in positions they are deliberately heading for, hoping (expecting?) to be able to secure the draw. How is that possible? I can understand a GM notching some positive results against weaker (relatively speaking) opponents through sheer persistence but to do so as often as Keith manages it? How to explain that?
Keith began by talking how he developed his taste for endgames in general.
As far as I can remember, I first began to realise that I enjoyed playing, and had a good feel for positions in which there are only a handful of pieces on the board, when I managed to win the ending of B+N v N+P against England International Carey Groves at Jersey in 1985.
I do also recall though that when I won an ending of two Bishops v B+N with a few pawns on the board a couple of years earlier against a player called Zak rated 2310, Jim Plaskett said to me afterwards “Arkell, you would be an IM easily if you didn't play so many stupid games”. A typical Jim comment! However, it was during the game against Panzer at Hastings 1990 that I discovered I had an ability to accurately calculate long variations in simplified endings.
When I started outplaying strong GMs from these types of endings I began to realise that this kind of thing really was my chess strength, and compensated me for a lack of interest in opening theory, and having no special skills in using or defending against the initiative, and also my aversion to incalculably complicated positions.
At the Watson Farley Williams Grandmaster tournament in 1991, I played indifferently in the openings and middlegames, but won a very nice ending of R+R+g pawn v R+R v Mihai Suba, beat Danny King in an ending with R+3 pawns each, and also beat Robert Byrne with my R+B+ h pawn v his R+g+h pawns. There was a moment in that last ending when he turned down the chance to defend R+B v R, and later remarked that he invented the “2nd rank defence”, and would have drawn easily had he chosen to! Then there was the game against Tony Kosten in Montpellier 2002 which I won with R+B v R+N without pawns! I saw some long and pretty variations in this endgame, one of which was about 15 moves deep; although again there were moments when he could have tried to defend a difficult R+B v R position, but instead went down my main line, in which my Rook won against his Knight.
A long list of Arkell endgame greatest hits. Before talking with Keith I had assumed that these battles must have been the consequence of, or possibly the inspiration for, studying this phase of the game so deeply that he had theoretical positions, particularly KRB v KR, dripping out of his ears. As it turns out, that’s not the case at all.
Recently I was looking at R+B v R with Jonathan Hawkins, and it became clear that he was better versed than me in some of the very long and precise winning variations. I am also often not sure whether I am in a drawn or a won position, but I calculate very well when there is reduced material, and have a good feel for how to improve my position.
I have even wondered whether this for me is a kind of chess autism. What I mean is that typical autism traits include repetitive behaviour and a strong preference for a familiar environment. Chess itself is in any case attractive to people with autism spectrum Disorders (such as Aspergers syndrome) because of its well defined rules and familiar patterns, but these simple endings take the principle even further, and remove a lot of the uncertainty and randomness associated with complex middlegame positions.
Though just speculation, it may well be that my skill in these simple endings was triggered by a liking for them on an emotional level – i.e. by what I'm terming “chess autism”.
So it's back once again to Keith's approach at the board. Remembering his comments about his proposed endgame book (see bottom of Part I) this probably shouldn't have surprised me. I recalled a passage from Excelling at Chess - Aagaard has been looking at a number of examples of Ulf Andersson's endgame play then says, "Andersson does not do anything special at any time, nor does he show skills which cannot be understood. But every move is good - not great, just good. This is the skill one should aim for when studying the endgame."
Did that sum up what Keith was trying to do when he's got R+B v R (or any other kind of ending come to that)?
Yes, I just try to make progress and present my opponent with practical problems to solve. Of course, there are some winning positions and some drawing positions that I recognise, but in general it is about trying to increase my advantage until my opponent can no longer defend. This is both in general endgame play, and specifically in R+B v R. Even in such a simplified ending as this there are plenty of opportunities to just play chess and hope to play better than your opponent. Only computers can see everything in such endings.
I've played alongside Ulf in team events, and he once told me that he just tries to avoid mistakes, but this, together with Aagaard's simple descriptions of the way he plays belies his deep understanding of chess. I think it's too simplistic to say that he doesn't do anything special, and his skills are easily understood. For a start he knows way in advance exactly where he would like to put his pieces, and which ones he would like to exchange. Ulf very often makes progress in the endgame with finely tuned precision. All of this looks deceptively easy IN RETROSPECT for the very reason that that it all fits so logically and artistically, but to fully reproduce an Andersson game would require his level of understanding.
And there our interview with a Grandmaster ends. Thanks to all my fellow bloggers for assistance in the preparation of these posts. Most of all, many thanks to Keith for his time and sharing his thoughts with us.