2010 SCM Zone Championships

Are the predictions based on the same people as they move longitudinally through the different age groups (i.e., you base the data on Chris Stevenson's times at 20, 21, 22,...107, 108, 109....) or are you looking at different cohorts (i.e., average top times in the current 20-24, 25-29, 30-34...65-69, 70-74, etc. age groups.) If the latter, than you aren't factoring out cohort effects that could bias things pretty widely. Most 70 year old swimmers today did not benefit from the early training, technique advances, etc. of most 20 year old swimmers today. I would venture to guess that when the latter hit their 70s, their times will be a LOT faster than the current crop of septuagenarians can go. Thus the "degradation" or "disintegration" or "let's call it what it is, rotting away" of human speed would not seem nearly so severe. Yes, different cohorts, and I'm aware of the potential problem. But it is much less work to do it based on a "snapshot" of records, and this isn't my day job... It is actually more of a (potential) problem on the men's side, because on the women's side we have a fewer number of people -- the Laura Vals, KPNs, Susan von der Lippes -- who have been dominating the records across the events and over multiple age groups for quite a bit of time. Another problem is just that there are fewer and fewer data in the upper age groups, so the variability is greater in the records at the upper (and even the very lowest) age groups. I usually cut off the fitting range at the 75-79 age group, if memory serves, and extrapolate the fit to make predicts. Extrapolation also magnifies uncertainty. Maybe I'll work up some data one summer, though; it would be interesting. I will say that there is definitely some anecdotal evidence of "snowballing." I am in charge of maintaining records in my LMSC, and since there are fewer people in the 75+ age groups and the same ones tend to hold all the records, I tend to notice their time increases.

Are the predictions based on the same people as they move longitudinally through the different age groups (i.e., you base the data on Chris Stevenson's times at 20, 21, 22,...107, 108, 109....) or are you looking at different cohorts (i.e., average top times in the current 20-24, 25-29, 30-34...65-69, 70-74, etc. age groups.) If the latter, than you aren't factoring out cohort effects that could bias things pretty widely. Most 70 year old swimmers today did not benefit from the early training, technique advances, etc. of most 20 year old swimmers today. I would venture to guess that when the latter hit their 70s, their times will be a LOT faster than the current crop of septuagenarians can go. Thus the "degradation" or "disintegration" or "let's call it what it is, rotting away" of human speed would not seem nearly so severe. Yes, different cohorts, and I'm aware of the potential problem. But it is much less work to do it based on a "snapshot" of records, and this isn't my day job... It is actually more of a (potential) problem on the men's side, because on the women's side we have a fewer number of people -- the Laura Vals, KPNs, Susan von der Lippes -- who have been dominating the records across the events and over multiple age groups for quite a bit of time. Another problem is just that there are fewer and fewer data in the upper age groups, so the variability is greater in the records at the upper (and even the very lowest) age groups. I usually cut off the fitting range at the 75-79 age group, if memory serves, and extrapolate the fit to make predicts. Extrapolation also magnifies uncertainty. Maybe I'll work up some data one summer, though; it would be interesting. I will say that there is definitely some anecdotal evidence of "snowballing." I am in charge of maintaining records in my LMSC, and since there are fewer people in the 75+ age groups and the same ones tend to hold all the records, I tend to notice their time increases.