The Nominees for Best Result are …
A new system of awards for excellence in theory
Oscar is, of course, the name of the most prestigious award in film making. The award ceremony is this coming Sunday, and it will be watched by millions. This made me think about awards for our community.
Today I will talk about how we might create a new type of award for excellence in theory, based on a model that is quite different from all the current awards in computer science.
The Award Name
I need a name for the new awards, so for now let’s call them a Euclid. Obviously, the awards are named after the famous Eucild of Euclidean Geometry. It seems fitting to have an award named for the “reporter” of one of the great algorithms of all time: the Euclidean algorithm for computing the greatest common divisor (gcd) of two integers. This algorithm is beautiful, fast, and in use over 2,000 years later—without the ability to compute gcd’s fast, modern cryptography would just not work.
There is general agreement on who invented the algorithm: someone else besides Euclid. Beyond this I cannot find any definite claim of who invented it. Some believe it was Pythagoras’ school, others credit Eudoxus of Cnidus, others believe Whoever invented it, the algorithm is associated with Euclid, and so his name will serve as an initial name for the award.
The Award Process
I think the place where we can be most inventive is in the process of selecting who gets the awards. I suggest a multiple phase process:
- First, a select committee will propose a list of 3-5 candidates in each category for the awards. More on categories in a moment, but for now think “best result.”
- Then, members of the theory community would vote for their choices. The members could be restricted to the obvious societies or other groups, and the voting would be done on-line.
- Finally, after the vote the results would be announced on a known day to maximize the press.
Rich DeMillo suggested the whole process could be an online social system. He suggests some social filter or clustering method could eliminate any “select committee.” I am not sure, but this does seem like a possible approach.
The Award Categories
I think that we could easily have many categories following the Oscar model. I do not suggest an award for “best editing,” but we could have a number of awards. For example, here are a few suggested categories:
- Best paper.
- Best survey article.
- Best exposition.
- Best book: textbook, monograph, or other.
- Best conference presentation.
- Best new result.
- Best new method.
- Best Journal Editor.
DeMillo had some very interesting ideas on categories—I would usually have him add these as comments, but they are too neat to not include now:
Why not best actor awards? You can have best contributor in a starring role (full co-authorship). That way results that span many individuals get recognized. Same thing with best contributor in a supporting role (acknowledgements or citations of the form “this result is based on ” get recognized). Also best director to recognize heads of groups, labs, funding offices, etc
In our area, theory, the significance of anything takes time to be realized. This has been the rationale for making existing awards “best paper of the last ten years,” rather than “best paper for 2010.” I wonder whether this should be a requirement for the Euclid’s. To be different, perhaps, we could actually have awards that are based only on work done in the last year. Or we could have a specific category of “best paper of the last twenty years,” and so on.
An interesting study of the Oscar awards has shown they have not been very accurate in identifying the “best” films in the long run. Best pictures have not always stood the test of time, and correspondingly pictures skipped for Oscars have often become classics. A quick look at the IMDb Top 250 tells one that 6 of the top 10 movies, including the first, haven’t won the best picture award. I think the Euclid’s might have exactly the same problem, but who cares. Why not increase excitement and honor more—if a paper or result is honored and later we discover it should not have been what is the lasting damage?
The Rationale for the Euclid
There are many reasons for the Euclid. One key reason is even those nominated would get some recognition. Today, most—all?—of our awards are completely secret. Those not in the inner circle have no idea who was nominated, we only find out who won. Being nominated for a Euclid and not getting one would be still a great honor. This is a simple amplifier used in many other awards from Oscars to the Booker Prize: why can not we also use the same amplification method? These other awards use essentially a short list notion: a list of those who are being considered for the award.
We currently announce our awards in a way set to minimize the impact. I would suggest that we have a pre-set day, a day known to all, and on the day announce all the winners in all the categories. We can do the same with the nominations, all would take place on the same day. For example, for next year we could:
- On February 5, 2011 we announce all the nominations—the short list. There would be a formal press release and also the list would be announced on the web in all the obvious places.
- On April 5, 2011 we would announce all the winners. Again it would be released via the web and other means.
I understand from Atish Das Sarma some conferences already do something like this with the “Best Paper Award.” At WSDM they announced the nominees months before the business meeting, and there they announce the single winner.
Winners would get some medal—perhaps a statue of Euclid—we can figure it out later. They would not be asked to give a talk, nor to give a speech, or to do anything else. They would just enjoy the honor of being the recipient of the 2011 “Best New Result Euclid.” The point here is what they do not get: they will not get a large check. My hypothesis is an award’s prestige is not linear in the amount of money given. The Oscars are a perfect example: the amount is . The same with many of the literary prizes.
Is this a good idea? What about the name? What about the categories? What do you think?