|
January 2003Then, a profile of Scott & Glenda Meyers and an overview of scoring methods. At the end, a little about the Awards Banquet and Practice Day. Our next event is on Sunday, February 9. Duplication allowed in whole or in part, provided full acknowledgment is given. Mike Driscoll & Scott Meyers , Co-Editors |
Scott and Glenda both have a special taste for cars.
His parents joked that car was his first word.
Her dad owned a used car business.
His first sports car was a '59 Austin Healey bugeye Sprite, followed by
MGB's, 914's, 911's, a Capri V-6, a Pinto, Dodge Shelby's, a Mustang, an MR-2,
RX-7's, and Neon ACR's.
She, too, has owned Porsches and Sprites.
Just recently, they've migrated to a Mazda Miata.
Scott accidentally drove by an autocross one day around 1970, and was hooked as soon as he stopped to take a look. His lengthy record of volunteerism began with our club itself, as he was one of the crew who formed the Arizona Region SCCA Solo 2 program out of the several local groups that existed at the time. So he was active in early events that 30 to 60 entrants held on shopping center parking lots. He has seen local autocrossing grow ever since, as regular entrant, equipment manager, course designer, newsletter editor, club director, ... pretty much all the jobs. Some more than once: Scott is returning to editorial duties at the Pylon Press this month, and will continue as primary event reporter.
Scott has autocrossed (or will autocross, just make an offer) almost anything that has two or more driving wheels, a steering mechanism, and an engine somewhere. The Dodge Rampage is his most unusual ride yet, if not the most successful. Still, he does hold five Divisional autocross titles, earned in different types of cars. Glenda also competes in our local events. She'd often driven an automobile de jour from her dad's lot, and not always below the speed limit. When she met Scott eleven years ago, he introduced her to autocrossing, that is, to driving fast around corners.
Scott taught eighth-grade mathematics for fourteen years and was an elementary school principal for fourteen more. Since retiring from that office in 1997, he's gone back to Dysart School District, working as a supervisor. Glenda is a professional tennis instructor.
Combining ratings across multiple events involving multiple participants is not simply done. Everyone knows of someone who's been treated unfairly by our club's scoring method, and almost everyone has the intuitive feeling that a different method would just treat someone else unfairly. Those who have a habit of thoroughly investigating things in the abstract (that is, mathematicians) have proved that this intuitive feeling is, in fact, correct. Every multi-event scoring method is a flawed scoring method.
That includes your favorite method, whatever it is. Mine too. We find this unsatisfactory, of course, because we'd like to have a scoring system that is fair in all instances and is consistent with all reasonable alternatives. That can be achieved one event at a time, but not across multiple events. (A mathematician would express this by saying that it's not possible to impose a total order on more than one dimension.)
Scores from a single event admit complete ordering, so that every participant's
standing can be unambiguously compared to every other participant's standing.
The situation is so stable that it makes no difference how 1000 PAX is set,
since changing that determination simply rescales all the participants' scores
without changing their relative standings.
That's why the message board discussion about 1000 PAX determination was
directed at the effect, not on a single event, but across multiple events.
And there's the friction. If scores across events are combined by adding (or, equivalently, averaging) them - the margin of victory approach - then the manner of setting 1000 PAX within the single events becomes important. But if single event scores are converted to ranks before the addition or averaging - the event position approach - then it's not important how 1000 PAX is determined within the individual events.
This last point is perhaps the most salient argument in favor of event-position scoring. But the method still has flaws. As it must! In particular, it drops nearly all of the information about the relative quality of different drivers' performances in the individual events. (Statisticians, who investigate appropriate ways to deal with multi-dimensional measurements and then draw inferences from them, would say that the conversion to ranks in the event-position approach involves the reduction of quantitative interval data to qualitative ordinal data.)
Perhaps the most appealing argument in favor of margin-of-victory scoring is that it allows the come-from-behind-to-win situations beloved of most sports fans. Flaws here, too, though. Of necessity! One flaw is that (unless your class is PAX'd separately from all others, which can be pretty boring) competitors outside your class may influence series-standings within your class. A big flaw is that about the only way of learning the differing effects on series results of various techniques for setting 1000 PAX in the component events is to compare those techniques on sets of actual data. Such analyses (a statistician would call them simulation studies) are time consuming and the results can be exceedingly difficult to summarize and hence to understand.
So, which is the best scoring method?
There is no best scoring method.
Okay, which of the event-position and margin-of-victory methods is the
better method?
Sorry, it depends on what your definition of "better" is.
Okay, okay!
Which is the preferred method?
Now there's a question that can be answered.
If this retrospective has been as unbiased as it's meant to be, then you won't be able to tell from it which method this mathematical statistician prefers. But, if you feel that one of our multi-participant multi-event series is analogous to a multi-inning baseball game, you'd prefer the margin-of-victory method, and you'd be right. If you see it as more analogous to a multi-game championship series in baseball, then you'd prefer the event-position method, and you'd be right again.
What about the club as a collective? The recent online voting showed a preference for the margin-of-victory method over the event-position method that, in any political setting, would be called a landslide.
It began at 6:00 pm with a no-host social hour.
Sharon Roberts conducted a scavenger hunt in which the goal was to identify
individuals with such attributes as having never owned a foreign car or not
having an e-mail address.
Each person approached could be asked only one of the questions on the list,
so it was a challenging ice breaker.
(This writer gave up in disgust when Tage Evanson told him he'd not received a
speeding ticket within the last year.)
A full field of seventy-two entrants enjoyed great January weather and 15 or 16
runs on Chuck Voboril's short but technical course.
Most took all those runs themselves, while some split them between two drivers.
The course: a short straight to a late-apex right turn into a four-cone slalom (left entry better than right), then gently left into a long backend sweeper, followed by a rounded left corner and a quick right corner, exiting into a second four-cone slalom (again, left entry better) to the finish.
The first right hand corner was ever so slightly greater than 90 degrees,
causing many to experience the oversteer condition described as Oh Sh**
by some.
The most vulnerable cones throughout the day were in the slaloms, especially
the third cone in each.
The common mistake was trying to carry more and more speed through those four
cone wiggles, and many runs failed at doing so cleanly.
The sweeper following the first slalom quickly pointed out who was confident
of their braking limits and who wasn't.
Many got caught coasting to the actual end of the sweeper as they chickened
out early.
And some went too deep.
It was challenging!
The final two corners (after the sweeper) comprised an 'S' that encouraged the driver to set up a rhythm. Some of us obviously can't dance very well: there were more jerky hand movements through these corners than bidders at a Kruse auction! Except here the motions caused one to lose time, not win something.
Scores of tires contributed to the day-long scent of rubber, and almost every driver learned something new. Valuable experience it was, having so many runs at a short course. Top prize for consistency goes to Don Sattler, driving a stripped black Miata. He posted 24.657 seconds as his best time of day. Twice! On consecutive runs. On his last two runs, in fact.
The most common comment afterward was "When's the next Practice Day?"
Well, not for a while. But the next performance runs are scheduled for Sunday, February 9.
|
(Posted 20 January 2003) |