Why not subscribe?

Tuesday, November 20, 2012

Have I made more errors than anyone, ever?

I was comparing notes with a younger, more junior colleague. They said “you’re probably right” and sounded like they meant that in deference to my more senior status and corner office.

“Well, let’s check it out. I’ve been wrong before. In fact, since I’ve been making a living in research since 1969 or so, I’ve probably made a lot more errors than you have.”

This led to the following thought.

1. I’ve been doing marketing research a long time, and most of that time in R&D.

2. There is not a single piece of data at the company I work for – which has some of the largest databases in the world – that doesn’t go through several of the algorithms I developed.

3. These algorithms are fallible, and sometime errors crop up – or the characteristics of the data change, so the algorithm no longer cleanly handles all cases.

4. There are many ways to count errors. For example, if a company messes up and releases personal information about 1,000,000 people, is that 1 error or 1 million errors? Either way is defensible.

I am reminded of the article by Kurt Wolters on the 2000 Bush-Gore election results in Florida – depending on which of a large number of arbitrary rules you applied, either candidate could be determined to have won.

5. So, it is entirely possible that there exists some metric by which I have been responsible for more errors than anyone, ever.

This is, of course, a theoretical exercise. But the fact that it might be true reminds me of some of the (in)famous records in sports. Let’s just pick one: who has struck out more than anyone else in baseball?

Answer: Reggie Jackson, with 2,597. Reggie Jackson is also in the baseball Hall of Fame, for being one of the great players of the game.  “Mr. October” remembered not for those 2,597 small defeats, but for all the big home runs and big victories. I should be so lucky.

Even if my dubious, unproveable claim would happen to be true, it won’t be true for long. The growth of data coming in via the internet clearly provides ample room for me to lose whatever crown I might erroneously claim