David Smerdon | October 15, 2012 16:41

“Fixing Flaws and Stopping Draws”

A lot of my friends often ask me what I do in my studies.  Unfortunately, for the most part, it’s difficult to make what I study sound interesting to a non-economist.  Economics as a field is fascinating and incredibly useful, but the core skills and courses required of a PhD in economics are themselves largely based around complex mathematical models that could send even the more erudite of dinner guests to sleep.

Recently, however, I got the opportunity to tailor one of my assessment pieces to a chessy theme, for a change.  The subject was called “Social and Economic Networks”, a course focussed on a new and emerging branch of mathematics based around network theory.  I decided to write my final research proposal on how the field might be used to develop a new tie-break system for Swiss and round-robin chess events.

I should stress that I’m not really suggesting it as a serious alternative for tournament chess, but it’s definitely true that the current systems each have their flaws – which I discuss in the paper.  My system is based on tallying up a network of direct and indirect wins over all participants in the competition, and has an added perk in that it encourages players to fight for decisive results rather than settle for quick draws.  Mind you, it has its own flaws, but at least it made for a more interesting project than what I’m used to.

You can check it out on the link below, if  you’re interested, though I should add some sort of warning for the mathematical derivations and prosaic language.  As an example, consider the end of the introduction:

This proposal seeks to address both the inadequacies of current tie-break systems and the issue of ever- increasing draws in chess by introducing a new tie-break system for large tournaments. The system uses a measure based on both direct and indirect wins, and is a generalisation of standard centrality values stemming from the directed network of wins throughout an event.

Yawn.

Still, it was kind of a fun application of some complex math, and I managed to wrangle a bit of text on recent tie-break controversies at the European Championships and also the Commonwealth Championships.  Just try not to be too scared by all the Greek symbols.

(You can check it out here: 01072012 SMERDON – Social Networks Research Proposal)

Share |
David Smerdon's picture
Author: David Smerdon

David Smerdon is a chess grandmaster from Brisbane, Australia. David attended Anglican Church Grammar School and Melbourne University. To qualify for the title of Grandmaster, a player must achieve three Grandmaster norm performances, and a FIDE Elo rating over 2500. Late in 2007, Smerdon achieved his third and final Grandmaster norm. In the July 2009 FIDE rating list his rating passed 2500, so he qualified for the title of Grandmaster. He is the fourth Australian to become a Grandmaster, after Ian Rogers, Darryl Johansen and Zhao Zong-Yuan. In 2009, Smerdon won the Queenstown Chess Classic tournament.

Source: Wikipedia

DAVID SMERDON OFFICIAL BLOG

Chess.com

Comments

Matt Phelps's picture

Is there really an "ever-increasing" amount of draws in chess? I get the feeling there are many, many ways to "fix" a "draw problem" that doesn't really exist.

Anonymous's picture

I agree completely. No reason to fix anything.

Horscht's picture

You are missing the point, guys! It's about a fairer tie break system, making a better trade-off between fairness and combativeness (and remaining practical.) It's not about "fixing" the "drawing problem", the catchy title of the article is perhaps slightly misleading.
On first glance, I think the proposal is great! How come nobody has thought of it before?!

brabo's picture

Who will do the calibration of such system? This sounds to me harder work than writing an article.
If you want to use such tiebrakesystem then you will need an app for the calculations. This makes it less practical than a tiebrakesystem which you can calculate manually too. I also prefer that participants know in advance of the last round, how the final tiebrake more or less will look. This doesn't sound possible with your new tiebrakesystem. You will get situations like in the olympiad that long after the games are finished, nobody knows who actually won.

Anonymous's picture

He stressed that it's not a real alternative. It's a paper, that is all.

brabo's picture

The blogarticle says indeed that is not a serious alternative. However the paper itself has a complete other tone and pretends to give a real solution. I assume the professor only saw the paper and not the blogarticle as I can imagine that it was the purpose to use the network theory to find a serious solution on a problem.

boardgame's picture

Happy birthday !!! :)

boardgame's picture

a month late though ;)

Bartleby's picture

On a quick first read, I see two massive objections:
1. The two claimed chief differences with Buchholz don't have to do with Directed Network vs. Buchholz, but with the arbitrary choice to ignore draws for tiebreak. I suspect if you calculate Buchholz ignoring draws you come pretty much to the same result.
2. Basic fairness: If you win against Kamsky and draw against Nakamura, it should give you the same tiebreak as if you win against Nakamura and draw against Kamsky. If I understood the proposal right, it won't. Again, this imo undesirable effect has nothing to do with directed networks, but with the choice to ignore draws.

Scott's picture

David's paper has left me wanting to now see more, to see what happens when it is applied to differing sized swiss and round robin events (will this stage 2 actually happen?). If a 'good' delta figure can be determined, then it should be a relatively simple matter to incorporate it into the pairing software systems. Then tournament organisers in advance could state what delta value will apply, if this tie break system is being used.

luzin's picture

your blog is very interesting, thank you and keep it up!

Your comment

Comments HTML

  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.
  • Web page addresses and e-mail addresses turn into links automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
By submitting this form, you accept the Mollom privacy policy.
By posting a comment you are agreeing to abide our Terms & Conditions