This was originally written as a Q&A about the proposed changes. Now that those changes are in place, this is here to help explain some of the improvements that have been made.
Q. What method do you propose to use?
A. At every race, every boat is given a ranking score between 0 and 1. It's determined by:
best score of race (rank ratio of best + difficulty factor) ------------------ X --------------------------------------- competitor's score 2This means that if you race against tough competition, you are rewarded; if you race in hard races, you are rewarded; if you do both, you are rewarded for both; and if you do well under these circumstances, you are rewarded a lot.
Your ranking is determined by your best three (highest) ranking scores of the year.
Q. How did you come up with this?
A. I did four years of research. I studied the current slalom rankings system here, internationally (i.e. the ICF's) and in other countries. I evaluated ranking systems from other sports (skiing, chess, tennis, football, basketball, etc.). I performed nearly a hundred experiments using thousands of real results. I read articles by people who rank other sports to find out what works for them and what doesn't. I solicited input from other people, had my mathematical approach reviewed by a math professor, and had the writeup checked by a professional writer to make sure it was accurate and understandable.
Q. Is this a major change?
A. No. It's simply an extension of the system that we have been using. It takes advantage of the best parts of the existing system and improves on it. For reference, here's what the current system uses:
best score of race (field strength + importance factor) ------------------ X ------------------------------------ competitor's score 20This results in a number between 0 and 1, just like the proposed system. However, the proposed system provides more accurate rankings for paddlers across the board, as discussed below.
Q. Is it complicated?
A. No. It uses simple arithmetic. While the research behind it is long and complex, the end product -- the rankings system itself -- is designed to be simple enough to be understood by everyone in the sport.
Q. Is it perfect?
A. No. To make it perfect is probably impossible and certainly would require great complexity. This system is designed to be fair, simple and as accurate as possible.
Q. Is it fair?
A. I think so. If we are to do rankings at all, we are obligated to be careful to see that they are done fairly. I take that very seriously, and have tried really hard to come up with something that treats all athletes at all levels of the sport equally.
Q. What are the benefits of this new rankings system?
A. There are several.
1. It removes or reduces several biases in the current system, so that all paddlers are ranked fairly, regardless of what ability level they have, where they race, how old they are, or what sex they are. The current system is biased with regard to all of these factors. The new system continues to accurately reflect the factors it should, such as strength of competition and difficulty of race.
2. It makes good use of the available data. Generally speaking, using more data means generating more accurate rankings. And it certainly helps to make the system inclusive rather than exclusive.
3. It rewards paddlers for doing well on hard courses against strong competition, regardless of when or where those races happen.
4. It works equally well for paddlers at the World, National, advanced, intermediate and beginner levels.
5. It is more accurate than the current system, i.e. the rankings it produces correspond more closely to real race results.
6. It can be used to compute rankings multiple times each year -- for example, once a month -- in under a minute. (Once results are recorded, the entire process is automated.) This can be used by athletes and coaches to track athlete progress.
Q. If I want to be ranked higher in this system, what must I do?
A. Paddle harder. ;-) More seriously:
1. Do better at the races you go to.
2. Go to harder races.
3. Race against better competition.
4. All of the above.
This aligns closely with athlete expectations, and corresponds to the normal development of athletes.
Q. Will I be able to walk away from a race and know what the impact of my performance today will be on my ranking?
A. No -- because somebody else might have had a better day than you did, perhaps at another race that you don't know about. (Example: on 6/4/2000, there were races in New York, Colorado and Nevada.) But you will be able to at least make an estimate. And if rankings are updated once a month, surely that's often enough to track even the most quickly-improving athlete.
Q. Well, what about this once-a-month thing?
A. The single biggest problem I have is getting race results and converting them so that they can be displayed on the web and used in this system. In 2001, I will be deploying a web-based system that hopefully will make it easier for race organizers to submit results, and easier for me to get them. *If* we can quickly get results, check them, and correct any problems, then we can run rankings as often as anyone likes.
Q. What about other proposed systems?
A. I would be happy to test, with real data from hundreds of races, any alternate system that someone proposes. Since I've done a lot of testing by now, I've gotten pretty good at it. And testing with real data is really the only way to evaluate the merits of any proposed ranking system, because it is then fairly easy to tell if the proposed system does a good job of approximating reality. No system will be perfect, but I think that this new system will be the closest we've come so far.
Q. Is there more to say about this?
A. Y'all know me: brevity is probably not my strong suit. ;-) I've written an entire paper on this that goes into exhaustive detail. It's long -- because there's a lot of detail -- but it's worth taking time to at least look at, even if you don't read every word. It's probably the only way to get a really precise idea of what the new system will do for you and your fellow paddlers.
That paper is located here.