Wednesday 28 November 2018

Lucky Thirteen

I think on reflection I agree with myself.

On principle I'm not against tie-breaks, as a system it seems - in and of itself - better than the alternatives, but as with penalty kicks in football, the problem arises when the tie-break system begins to affect the match itself, to such an extent that the match almost becomes subordinate to the tie-breaker.

Whether we've reached that point or not is a matter of debate. For all I know the next match is going to finish 5-3 with four draws and we'll wonder what all the fuss was about. What isn't a matter of debate, I don't think is that if there wasn't a tie-break coming up, this

doesn't happen.

It's the moment we've all been waiting for - the last game, the match level, one player pressing, one player fighting for his life, and really there shouldn't be anywhere to go from here but play.

So maybe there shouldn't be anywhere to go from here, no escape route, no play-offs.

I wouldn't mind a thirteenth game. (Or a fifteenth, or a seventeenth.) Tie odds to the player who has an extra Black. Who should that be? Maybe let the champion choose, that can be the nature of their edge.

Ideally - and no system is ideal - a system should

  • be as simple as possible
  • be as fair as possible
  • be as little disruptive as possible (in this context, affect how the players play as little as possible).

Seems to me it's simple enough to understand. Whether it's as fair as possible - I think we all know that the only fair way is to play to a given number of wins, but that isn't going to happen, for all kinds of good reasons. And we ought to have much longer matches than a dozen games, but that probably isn't going to happen either. Whether it would affect the way the players play - I'm not seeing it, any more than it has affected matches in the past when a player has gone one up.

Yes, it allows for a winner when the scores are level at the end. But that's what happened in 1951 and 1954. And, for that matter, in 2016 and 2018, because we had a winner then and we're going to have a winner today.

But the scores were level at the end.


AngusF said...

Suspect I'm less keen for change than others are because I've enjoyed the match and feel it's been a contest (despite the fact that all games have been drawn).

That said I'd prefer a longer match - say of 16 games - to give, I think, a greater chance of decisive results.

Perhaps an option would be to have fewer rest days. Three games in three days days and then a rest day - as in the Candidates tournament - would allow for 14 games with the same match duration.

Another idea: how about resolving tied matches by game count - like board count in tied team matches? For each player, sum the game numbers of won games; the winner is the player with the lower score. If the result is still a tie then eliminate the most recent decisive result and count again.

ejh said...

It strikes me that my tweet may give the impression I think every match will "always" go like this one has - for clarification, "always", refers to the situation in which Carlsen found himself. (I should probably have said "always likely to happen, but that's the problem with instant takes...)

kieran said...

I couldn't cope with the quadruple negative of "What isn't a matter of debate..", but guessed what you meant.

Why not give the champion draw odds, and force the challenger to win outright? I expect the organisers like the current set-up though: blitz games /are/ pretty exciting. Choosing a champion by armageddon wouldn't feel right, though.

ejh said...

Why not give the champion draw odds, and force the challenger to win outright

Perfectly viable plan: it's just with an odd number of games you'd force them to hold another game with Black to achieve that, which might strike some people as a slight inprovement.

Anonymous said...

I wouldn't mind a thirteenth game. (Or a fifteenth, or a seventeenth.) Tie odds to the player who has an extra Black. Who should that be? Maybe let the champion choose, that can be the nature of their edge.

This idea was suggested in 2016. The seemingly fatal flaw (your #2) is that person (undoubtedly the champion if you let them pick) with one extra Black has in fact a LARGE advantage (only slightly less than draw odds in the match). An 8-8 draw is about a 15% chance in a 16-game match (so overall 57.5 vs 42.5%). You system maybe gives the challenger a 44% chance or so.

To make it "fair" you need to give FIVE extra Whites to one side (10-5 or 11-6 split), and that's assuming rather low draw rates by modern standards. (This is independent of the total number of games, as one assume the other games with equal black/whites are split if the players are equal).

Still, FIDE seems to just randomly pick its systems anyway (the WCOC and/or Agon just rubber-stamp whatever is proposed, occasionally nodding to the Smartest Person in the Room for whatever pseudo-explanation is needed). Occasionally they ask math boffins (like the Polish guy Bartosz Socko), etc., but usually it's just by "feel" from FIDE insiders.

Whether it's easier (your #1) to explain a 5-White edge is unclear. OTOH, I think it's clear one extra White *would* affect play (one side essentially has draw odds), and of course five more Whites would undoubtedly modify play too, but in a different way.

Anonymous said...

A different calculation for the same concept. Having draw odds for Black in a single game is about a 75-80% rate, so 200-250 Elo. Averaging over the other 12 games, the challenger would need to be 15-20 Elo better (20 x 12 = 240) to equalize the match chances.

Another application of the same theory is that instead of having teams-of-4 play via split colours, you could instead give one team (either with teams of 4 or 5 players) all the same colour, and the other draw odds (cf. China vs USA in Olympiad).