Friday, May 7, 2010

celebrity thursday

Yesterday was kind of a big day, celebrity-wise for me.

We went to a lecture from Naomi Wolfe

she was on tour for the 20th anniversary of her feminist book "The Beauty Myth". We got front row center seats. We were so close that if we were to get up and walk about 30 feet, we could practically reach out and touch her. Her premise is basically that standards for female beauty are unrealistic and unattainable, but they keep women on a treadmill of buying cosmetics and going on fad diets and so on.

Earlier in the day I met two famous people from the history of forecast evaluation (how good forecasts are). Robert Winkler was visiting from overseas and we got to go to lunch
He was co-author on many papers with the most famous person in forecast evaluation, Alan Murphy (now deceased). When I was a wee young grad student, I had a grainy 4th generation xerox copy of a collection of Murphy's works and they had a big impact on my forecasting and forecast evaluation philosophy. Winkler also wrote papers with a guy named Clemen, who was one of the most famous researchers in how you combine different forecasts together, e.g. the average of 100 peoples guesses of how many jelly beans are in the jar is going to be a better guess than the best single guesser.

If I had to, say, pick one non-obvious piece of philosophy out of the air from Murphy and Winkler, it might be that one measure of "goodness" of a forecast is how well it aligns with the forecasters personal belief. A forecaster should never be in a situation where he has incentive to say the forecast should be, for example, 60% chance of flood, when in his heart of hearts he believes there's only a 10% chance. Think of it another way, a doctor should never say you have 6 months to live when he believes you really have 3... or 12.

Lastly I also finally met and had lunch with Frank Woodcock

He works for the Bureau of Meteorology. In 1976 (when I was 2 years old) he wrote a forecast evaluation paper that lead to what's called "The Woodcock Score", which, again, I read about in my early days at grad school. He was a real friendly guy and quick with a laugh. I imagine that over the years, he's seen it all.

No comments:

Post a Comment