The Electric Commentary

Wednesday, March 23, 2005

Gaming the System

UW Law Prof. "Oscar Madison" is firmly against adding an asterisk to records set by players using steroids. He cites 3 purposes of baseball statistics:

Baseball records are a factual record of player performance that function as (1) an aid to memory, (2) a basis to compare the abilities of current players, and (3) a basis to compare the abilities of players across time.

The major thrust of his argument seems to stem from the fact that raw, unadjusted baseball stats are not a good comparison of players over time. For instance:

Frank Baker, third baseman for the Philadelphia Athletics in the 1900s and 1910s, was one of the foremost home run hitters of the dead ball era but never hit more than 14 homers in a season. He certainly had reason to gripe about the bright, new tightly-wound and frequently changed baseballs of the 1920s that Babe Ruth hit out by the dozens.

Prior to 1920, it was legal for pitchers to doctor the baseball with spit, vaseline, tobacco juice or sandpaper. Since 1920, that has constituted "cheating," though numerous pitchers prided themselves on getting away with such cheating, such as notorious spitballer, and Baseball Hall of Famer, Gaylord Perry.

And he drops the obligatory Roger Maris comparison:

Bunning now makes a shibboleth out of Roger Maris's 61 home runs, but he pitched in the AL that season and should remember that many people both among baseball fandom and officialdom did not want Maris to break Babe Ruth's record. And when he did, Major League Baseball, in its wisdom, decreed that Maris's home run "record" would have an asterisk in the "official record books" to signify that his 61 home runs surpassed Ruth's 60 homes only in a qualified sense: Maris hit his 61 in the modern 162 game season, whereas Ruth played in an era of 154 game seasons.

A few years ago, that asterisk was properly given up as idiocy. I think it would be further idiocy to put asterisks next to Mark McGuire's 70 home runs, or Barry Bonds' 73, or any accomplishments by players found to have used illegal steroids, let alone to "wipe out" those records.

It seems to me that Madison undervalues his second function of statistics: Comparing the abilities of current players. Because steroids are illegal, and not used by every player, certain players had an unfair advantage against their peers. It is as if Mark McGwire and Jose Canseco were using aluminum bats on the sly. Obviously then, the statistics that they have accumulated will be inflated not just over the course of history due to rule changes, advances in technology, etc., but also over current players. Perhaps Mike Greenwell has a point, no?

I’m not necessarily in favor of adding asterisks either, as I believe that proper skepticism towards current players will always exist in baseball fans, and that’s good enough. However, this is an interesting position for Oscar to take in light of recent events.

You see, recently Oscar Madison and Tom Bozzo conspired together to move up the rankings of the Truth Laid Bear Ecosystem through trickery. For those of you who don’t know, NZ Bear at The Truth Laid Bear developed a system for ranking blogs based on incoming links and traffic. Madison and Bozzo simply linked back and forth over and over again until their blogs started moving up in the rankings. They have now been banned from the ecosystem for the stunt, although Madison has since apologized (Note: For the record, I don’t think that Madison was really attempting anything nefarious, and basically just wanted to see if he could do it).

Of course the Ecosystem is merely a method of measuring the performance of a blog. Its chief function is to let people know how their blogs stack up against the blogs of others. Madison seems to think that a high ranking is some kind of honorary award, but such a function would be, at best, a distant second in terms of its importance and its fit with the logic of the Ecosystem.

If we follow Oscar Madison’s logic, shouldn’t Ecosystem users give full faith and credit to the ill-gotten links and rankings that Madison and Bozzo achieved with their experiment? After all, a Madison-to-Bozzo link is still a link as much as a Jason Giambi home run (Circa 2000) did in fact fly over the fence.

The fact of the matter is that Barry Bonds’ ill-gotten home run record contaminates baseball records in the same way that the Madison-Bozzo stunt contaminated the Ecosystem. It put bad information into the system, and because of it everyone else’s position was affected. The primary function of statistics in my view, is to convey accurate information, and while it may not be necessary to add an asterisk due to the stigma that is sure to attach itself to many current players, it would certainly be more accurate to include the asterisk than to omit it.


  • My motivation was pretty clearly spelled out here at the get-go, or see Jeremy Freese's definitive post-mortem.

    As it turned out, the ecosystem rank is very easily manipulated, particularly in the mid-to-upper ranks. In contrast (in degree if not kind), steroids may well make a good hitter into a better slugger, but hitting major-league pitching remains hard.

    The accuracy of the asterisk question hinges on the magnitude of the marginal effect. If Bonds, well-conditioned but drug-free, would have hit at least 62 home runs in his annus mirabilis, then an asterisk in the record book wouldn't be so obviously informative.

    By Blogger Tom Bozzo, at 3:56 PM  

  • Hi Tom,

    Thanks for the comment.

    I still think that my comparison is apt, and I hope that this response clarifies exactly why. The explanation given by you at Marginal Utility doesn't seem that far removed from my throwaway line directed at Oscar, that he probably did it to see if it could be done, as an experiment. I'm all for finding a more accurate measurement of blog success, but the whole ecosystem bit was really more of a set up to deal with the baseball point, so let's deal with that.

    Asterisking the stats of abusers, only if their improvements were "significant" will lead to further pollution of information by signalling certain users as cheaters and letting others go, and anyway, there is no way to tell what a given player would have done if he were clean. Also, I like to use the Caminiti curve:

    Ken Caminiti, HRs, 1993-1998 – 13, 18, 26, 40, 26, 29

    Barry Bonds, HRs, 1998-2003 – 37, 34 (in 102 games, 49 over 150, projected), 49, 73, 46, 45.

    I find it very unlikely that Bonds, McGwire, or Sosa would have reached their record setting levels without some assistance, and if they were using, and it was not ubiquitous (it gave them an unfair advantage against current players) it should be noted. Perhaps the stigma currently afflicting the lot of them is enough, I merely want it on record that they cheated, and some of their peers did not.

    Also, an increase in power clearly aids in your consistency as well as your power, as the two, at some point, become related, especially if you can increase power without having to swing harder (when you intentionally swing harder you sacrifice control. If you can make your basic swing stronger, you do not). If you get stronger and you put the bat on the ball with the same frequency that you normally do, you will gain hits due to harder ground balls, would-be fly outs that become both home runs and foul balls with a little added power, and more walks.

    Just because hitting a ball is difficult does not mean that steroids do not make it easier. They do. Bonds never sniffed a batting title before the big home run years (typically around .300 until "The Big Jump" at which point he's in the .350s to .370s. Note that his high average before "the jump" was in a year in which he also hit his at-the-time career high in HR, 46.), McGwire (.312 when he hit 52, .299 when he hit 70, before that he averages about .250, with several seasons of around .230) hit significantly better in his big HR years, and Sosa (a .250 hitter who suddenly jumps into the .300 range) as well.

    Anyway, while it takes skill to be a ballplayer, for those that possess basic talents, steroids are a relatively easy way to improve. The Ecosystem is obviously easy to manipulate, much easier than MLB, but baseball can be manipulated, and it's not all that hard. And for stat-heads, we should want to know if someone is abusing it, and adjust accordingly (especially if you're a SABRmetrician and adjust for era and ballpark and the like already).

    By Blogger PaulNoonan, at 4:40 PM  

  • Paul,

    More evidence that I'm a bad blogger: I respond to your post a week late. Plus it takes me thousands of words to get all my arguments out. (Maybe my subsequent steroid posts answer some of your contentions?) I agree with you that steroid use distorts records, but I think you overestimate the accuracy of baseball stats "untainted" by steroids even for purposes of comparing contemporary players with one another.


    By Blogger Oscar Madison, at 11:36 PM  

Post a Comment

<< Home

Amazon Logo