THE BOOK--Playing The Percentages In Baseball

This is very cool - I can’t wait to see what the consensus “most unethical” play ends up being!  On a side note, it appears that you can randomly get the same situation as #1 and #2.  I clicked “Pass”, but just an fyi.

Yup, just noticed it, and I’ve given studes a fix.  He’s going to test it to see if it works, and we’ll get that done in a jiffy.

It’s been an hour, and I’ve already got over 200 votes.

Hardball Times must be in the thousands already…

Yup.  Over 4,500.  I’ll try to get to that fix later today.

I just sent you an email with code so that you can get the rankings in real-time.  Check it out…

A similar voting process was used to determine which ballplayers had the cutest butts.  It was a huge hit on an A’s blog, got tens of thousands of votes, and was eventually featured in a magazine.

http://www.owlmelon.com/cgi-bin/odd.pl?MoneyButt

Very cool.  I love how he documents how the program developed from some tiny little script (like we have) to someone more engrossing.

Real-time results:
http://tangotiger.net/ethics/ethics_results.php

Note that until the “opponent winning percentage” stabilizes to close to .500, you should look at both winning percentage and the strength of schedule.

The “pass rate” is how often that particular scenario came up, but the reader chose not to vote on it (for whatever reason, not the least of which could be that they didn’t like the opposing scenario to consider).

This is an admittedly nit-picky complaint, but there are a few cases in which I think there could be some ambiguity about what exactly is unethical about the situation.  I think I played about a dozen “games”, and the most unethical thing (IMO, of course) that popped up was the Bonds indictment.

But I did not rank it as such, because I assume the survey is approaching it from the standpoint of Bonds being unethical, whereas I consider the government’s actions to be unethical.

Another one is the Eddie Gaedel case, which personally I would never pick as the “winner” in any of these cases.  But I could see some people viewing Veeck’s actions as unethical, and others thinking the decision to ban midgets is unethical.

Or is the point to rate the total lack of ethics in the scenario, in which case the Bonds haters and the Bonds supporters might find common ground?

I think it’s a more than fair complaint, and I asked the professor who created the text of the scenarios to respond.

His response, unedited (I added the bolding):

=================================================
Willy Stern here, the guy who taught the Carleton College course. 

Patriot makes a very valid point, and one that my students and I wrestled with a fair amount last semester.

Quick background, which may be a bit illuminating, though it may do little to erase Patriot’s reasonable frustration: The scenarios were developed as a starting point for class discussions at a top liberal arts college. They were designed to be provocative and to force smart young students to think hard and well about ethics and mores in American society over time.  Baseball was simply the vehicle, or lens, for those discussions. We teachers revel in the pedagogical possibilities of such ambiguity. But now these self-same scenarios are being used for an ethics game (which, to be honest, delights me no end.)

The obvious structural problem is that incidents written for one purpose are being used for another. They adapt reasonably well, but not perfectly. And Patriot raises an intelligent point, and one also raised by my students during the homework part of the course which required rankings: When more than one (potential) ethical lapse shows up in a single scenario, what ought one focus on when doing the ranking? There is no right answer, of course, which is the beauty of teaching ethics at a liberal arts college. What I asked my students to do was to find the “worst” ethical transgression in the incident, and to base their ranking on that transgression alone.

This is an imperfect solution, as Patriot well knows, as the determinant of that single transgression is open to much ambiguity. Reasonable and well-informed people can and ought disagree. I was kind of hoping that the “pass” option would dissipate some of the frustration. But Patriot’s criticism is accurate and fair in every way.

Willy

This brings up an interesting problem—what is the most efficient way to rank a subjective list. Let’s say you have 800 CDs and want to put your top 100 on a special shelf—how do you rank them?

If you have a list of 10 items, the best approach is to compare every item against every other item. That makes for 45 comparisons, which is easy enough. Off the top of my head I think the number of comparisons is n(n-1)/2. But ... if you have 100 items, then you’ve got 4,950 comparisons. For 800 items, that’s 319,600 comparisons.

One way around this is a merge sorting system (http://en.wikipedia.org/wiki/Merge_sort). If you have to sort eight items—A, B, C, D, E, F, G and H, then sort them in pairs :

A
B

C
D

E
F

G
H

Then merge the pairs into groups of four. Because the pairs are already sorted, you only need to compare the top item in each group:

A
B
C
D

E
F
G
H

Then merge those groups together (again, because the groups of four are already sorted, you only need to compare the top item in each list. So you compare A vs E, and if E wins, then the next comparison is B vs E etc):

A
B
C
D
E
F
G
H

Any other suggestions?

Meanwhile, I imagine you would have to apply a “schedule adjustment” to the ethics list, unless the opponent winning percentages converge at .500 (would probably require a lot more votes). Seems like a problem right up Andrew Dolphin’s alley.

Dackle, that is EXACTLY the plan here!

However, THT has received so many responses (each scenario has played 450 games), that I suspect that their “opponent winning percentage” will have converged to .500 by now.

In mine, I’m nowhere close to getting a .500 opponent winning percentage, and therefore, Andy will be providing a “strength of schedule” result.

Project linked to from here:
http://sportsillustrated.cnn.com/2008/sioncampus/08/08/the-10-hottest-cheerleading-squads/

Look for “best class ever"… at some point.

I included an opponent-adjusted Win% for the time being:
http://tangotiger.net/ethics/ethics_results.php

It’s adjusted using the Odds Ratio method.  The long multi-step process boils down to:

1/(1+(1-win_percent)*(1-opp_win_percent)/(win_percent*opp_win_percent))

THT’s results:
http://www.hardballtimes.com/main/article/how-ethical-was-it/

which you can compare to link at #15

For those who missed it the first time, feel free to respond.

Andy has also written a “strength of schedule” program for this project, and so, I’ll publish the results of that later this week.