What is it about sports? Part 4 - Which team is best?

Having just seen the St. Louis Cardinals squeak into the postseason and then win it all, it reminds me of some thoughts I've had regarding postseason playoffs, especially with regards to college football.  There's probably a majority of fans who want to ditch the BCS one-game "playoff" and go to some sort of a playoff for FBS teams to determine a "true" national champion.

Given that virtually all other NCAA championships are determined by some sort of postseason tournament, there seems to be no reason not to make big-time college football the same as other collegiate sports competitions.  But I'm of two minds about this.  Yes, it would be nice to wind up with what would become an undisputable national champion, supposedly settled by performance on the playing field rather than something so heavily influenced by opinion polls and computerized ranking schemes.  Then FBS college football would become indistinguishable from pro football (except for the salaries!).  And that's the rub for me.

As we witness the degrading spectacle of FBS teams engaging in an orgy of "realignment," it becomes quite obvious, even for anyone who has been operating in a vacuum for the past 25 years, that FBS football has become all about the money that football generates.  Traditional rivalries?  Who cares, when big money is at stake?  After all, universities are businesses now - they long ago ceased to be primarily about education, anyway.  NCAA football is simply the low-paid minor leagues for pro football now, and all of the fun and tradition of amateur athletics is disposable in pursuit of the almighty dollar!  For the best athletes that universities can buy (and offer them a showcase for their talent in pursuit of a roster spot on a pro team), they rake in millions in revenue - which allows them to attract even more top athletes.  No one cares much about the majority of college athletes who don't go on to a career in the pros, anyway.  They're only along for fun and to support the play of the real playmakers - big-time collegiate sports is about revenue, not fun!

And just what does a playoff produce?  It inevitably produces a champion at the end, of course.  But the arguments go on about just which team is the "best"!  [My definition of "best" is to be offered below.] In the minds of some fans - the fans of the championship team - their team was best because they won the playoff.  Period.  End of story.

But as we have seen in the World Series of 2011, the outcome of games can be highly dependent on things that may have relatively little to do with onfield performance:  injuries, random bounces of the ball (or whatever), just missing or just making a play, the weather, coaching decisions, bad calls by the referees, the playing field conditions etc.  Athletic contests are a fine example of nonlinear dynamics and it becomes quite possible for the hypothetical best team to lose a game during the playoffs owing to one or more of these non-performance factors.  In fact, the larger the field of teams in the playoff, the more likely it becomes that the hypothetical best team will be beaten.  Although playoffs always yield a final, surviving champion, the determination of which team was best that year remains undone.  Arguments can continue indefinitely.  And that seems rather like the way things are now, using the BCS system:  a champion is determined but the arguments go on.

So here's my modest proposal for determining which FBS football team is truly the best.  Each team will meet every other team multiple times, under a variety of playing conditions, weather, roster status for all the team's top players, home/away/neutral field, and so on.  Just for the sake of a number, let's say each team will play every other team 100 times, and each game will involve different conditions.  Since there currently are 120 FBS teams, this will amount to a rather large number of games for each team [(119 x 100) = 11, 900 games, so the entire regular season of FBS football will involve 11, 900 x 120 = 1, 428, 000 FBS games].  I guess this means the end of Saturday afternoon college football - they'll have to play multiple games every day of the week to fit in the required games (see below).  But of course, college football has already spread to other days of the week!  After all, prime time TV revenue is important.

At the end of this expanded "season" (from 12 games to nearly 12,000 for each team), the 16 teams with the highest winning percentage (ties will be broken by some sort of tiebreaker rules) would then enter the championship playoff.  Each team would again play every other team in the playoff 100 times under a variety of conditions [for a total of 16 x 15 x 100 = 24, 000 games], and the team with the highest winning percentage would be declared the national champion (again with a set of tiebreaker rules).  Think of the revenue this scheme would generate!!  And the old argument that "If X played Y 100 times, they'd win PP percent of the time!" would actually be played out!

This "ensemble" approach to determining a national champion is a logical way to determine which team is actually the best, because the random, non-performance factors that seem to be so important to the outcome of a single game against an opponent would tend to even out over 11, 900 games.  I admit, there are some practical issues that would need to be worked out, such as the degradation of the playing fields from playing multiple games every day, or the fact that it would require each team to play 10 games, 7 days per week, for 170 weeks (3.3 years) to just to complete the regular season, or wear and tear on the players.  But these are just minor details.  What counts is the enormous revenues this would generate and the chance finally to know which team is best!  Once every few years ...

ENSO and tornadoes?

Today, I had a nice discussion with a colleague who showed me some fascinating preliminary results of some analysis he'd done regarding the El Niño/Southern Oscillation (ENSO) and tornadoes.  I don't want to steal his thunder, but what he showed me was that ENSO could explain about 10% of the variability in tornado occurrences.  This number struck me as being a reasonable one - my position is not that ENSO is irrelevant to tornado occurrences, but rather that it's only a part of a much more complex set of processes that ultimately control tornado events.  Much of that complexity is beyond the pale of existing atmospheric scientific understanding, unfortunately.  The notion that only a relatively small portion of the variability can be attributed to ENSO seems both plausible and consistent with what understanding atmospheric science can offer at present.

Imagine my dismay when, later in the day, I read in the prestigious science magazine Nature for 22 September (p. 373) that "El Niño, a quasiperiodic cycling of tropical Pacific water temperatures, causes extreme weather around the globe."  This tidbit of journalistic excess really set my teeth on edge after my earlier discussion today!  ENSO is just one of many quasiperiodic "cycles" (e.g., the North Atlantic Oscillation, the Pacific Decadal Oscillation, etc.) that affect the global weather patterns and no doubt more will be discovered with time.  ENSO is a player but far from the dominant one, and the resulting global weather pattern is the result of the interplay between all these processes (and those as yet undiscovered!).  Attributing events exclusively to El Niño is outright nonsense!

Although ENSO apparently does modulate global weather patterns, there are many causal linkages between that global pattern and tornadoes, many of which remain unknown to science, at least in detail.  To say that El Niño causes extreme weather is a ludicrous exaggeration of its influence.  This sort of easy "explanation" often appears in media coverage of extreme weather events as a sort of non-explanatory explanation, similar to the "clash of air masses" nonsense often heard in media coverage after tornado outbreaks.

Once again, I'm reminded of the vast gulf between what's offered to the public and what I know as a severe weather scientist.  I expected better of Nature, clearly, and was disappointed to see such an ignorant "explanation" for extreme weather events within this magazine.  I've seen many scientists tempted by the siren song of ENSO as a way to explain interannual variability of tornadoes.  It appears to be a seductive hypothesis, perhaps because there is just enough of an actual influence to convince many to pursue the topic - I wound up publishing a paper addressing some of the myriad problems in trying to establish a causal connection between El Niño and tornadoes.  But the beat goes on, nevertheless - like the "clash of air masses," it's a simple "explanation" that just won't go away.

Is something wrong with tornado warnings?

In the wake of this year's massive April tornado outbreaks and the 22 May Joplin tornado, there's been a lot of talk about changing the National Weather Service (NWS) tornado warning process.  I don't know the inside scoop on what is being considered by the NWS, but there certainly has been a lot of media attention on the topic and it's my understanding that NWS is indeed considering changes.  I see all of this as an inappropriate reaction to the large fatality toll this year.  In some ways it's good to review the system and at least discuss whether change is necessary after a tragedy like this year, when so many lives were lost.  But hastily-constructed change is not an appropriate answer to what happened in the spring of 2011.

The public should understand that the NWS system evolved over decades as a series of more or less ad hoc decisions made in a cloud of ignorance.  If one were to ask today if the existing system is the best of all possible systems, I believe the answer would be a resounding "NO!"  The NWS warning system was not constructed after careful study of what is best to do and how best to do those things. It was built on the fly, as it were.  Nevertheless, there can be no doubt that it's worked effectively over the years to produce a much-reduced fatality toll from tornadoes compared to those in the 1920s and before.  If we change the way NWS warnings work, we should be careful not to to do so in haste under political or media pressure.  Otherwise we run the risk of screwing up a system that has worked to save thousands of lives over the years.

This is not the place for a discussion of what's wrong with the NWS warnings - that's a topic far too complex to be dealt with properly in a blog.  What I want to do here is to point out something I've been saying for a long time.  If people in the tornado-prone areas of the country choose not to take warnings seriously, is this because of failures in the warning system?  I find it to be extremely difficult to understand how people could not take a tornado warning seriously!!  Nevertheless, it seems that this is not uncommon.

Consider the tale of three tornadoes - two in 1990 and one in 2011 ... the F5 tornado that hit Hesston, Kansas,  the F5 tornado that hit Plainfield, Illinois, and the EF5 tornado that hit Joplin, Missouri.  In the Hesston event, tornado warnings were issued and people apparently took them seriously - only one life was lost.  In the Plainfield event, tornado warnings were not issued before the tornado, with 29 lives lost as a consequence.  For the Joplin event, warnings were issued about 20 min ahead of the tornado and the current figure I have for fatalities is 159 - the most from a single tornado since 1947.

Not all NWS offices do an equally good job with tornado warnings.  After the Plainfield event, for instance, a disaster survey team found many problems with the performance of the Chicago office.   Having an F5 tornado strike with no warning remains possible, but is much less likely today than it was in 1990.   The Springfield office that issued the warnings for the Joplin event did a decent job with a challenging situation, although they may have had a history of many false alarm warnings. Some of us have been saying for quite some time that large fatality counts are not necessarily a thing only of the past, and 2011 has shown those dire predictions to be true.  We have been lucky for decades and in 2011, our luck simply ran out.  It will do so again, in the future.

The sad, and seemingly inescapable conclusion I draw from this trio of major tornado events is that people in the tornado-prone parts of the USA need to understand the true risks from tornadoes.  They don't need to live their lives in constant fear - far from it.  The NWS offers state-of-the-art information about the threat of tornadoes and all people need to do is take it seriously enough to take actions to protect themselves.  Why is this so difficult?  Is it a matter of public education?  I think that might well be an important factor, but not the only one. 

There's been a lot of talk about too many false alarms from the NWS.  While I agree the NWS warning performance is not perfect and not even as good as it could be, there inevitably will be uncertainty implied (if not stated explicitly!) in their forecast and warning products.  People need to understand and accept that principle.  The public must accept their share of responsibility for what happens when tornadoes strike, rather than placing the whole burden onto the NWS to provide some product for which the science of meteorology offers little hope of being an improvement over the existing system.  Learning about the simple and inexpensive things anyone can do to protect themselves from natural hazards seems like common sense to me, not something too challenging for "the public" to do.