Tony Blair loses Munk Debates with Christopher Hitchens! Yeeeeaaaaaahhhh!!!

That's the conclusion one reaches if you're a fan of sloppy statistics. The debate between Blair and Hitchens just ended, and since it wasn't streamed live without paying I haven't seen it yet. Nor do I care who wins - I enjoy a good debate and prefer to avoid commentary on it after watching one as I prefer to let my first impressions sink in on my own. After about a day then I'll become curious whether others had the same impressions as mine, and what is often the case is that they didn't.

More interesting than who won the debate is who came up with the most interesting point. In a recent Hitchens - Dembski debate I thought the most interesting part was when Hitchens talked about whether a non-specialist can contribute to fields of knowledge as well. This was in response to a point made in Stephen Hawking's latest book where apparently he says something to the effect of human knowledge has progressed to the point where you have to be a specialist in order to make a meaningful contribution to any field now, in comparison with before where a layman with a good deal of thought could contribute without even necessarily having an academic background in that area...the example he gave was one where Immanuel Kant contributed greatly to understanding of seismology after the 1755 Lisbon earthquake:

The concept of the sublime, though it existed before 1755, was developed in philosophy and elevated to greater importance by Immanuel Kant, in part as a result of his attempts to comprehend the enormity of the Lisbon quake and tsunami. Kant published three separate texts on the Lisbon earthquake. The young Kant, fascinated with the earthquake, collected all the information available to him in news pamphlets, and used it to formulate a theory of the causes of earthquakes. Kant's theory, which involved the shifting of huge subterranean caverns filled with hot gases, was (though ultimately shown to be false) one of the first systematic modern attempts to explain earthquakes by positing natural, rather than supernatural, causes. According to Walter Benjamin, Kant's slim early book on the earthquake "probably represents the beginnings of scientific geography in Germany. And certainly the beginnings of seismology."

So that's the kind of thing I most look for in a debate.

Obsession with who won the debate though has produced the following sloppy statistic (from here and here):

Preliminary results on the Munk website said 68 per cent of the votes backed Hitchens and 32 per cent Blair.

All right, 68% for Hitchens and 32% for Blair. Blair wins, as the second article says (Hitchens 1 - 0 Blair).

But wait a second...

Both men gained about 10 percentage points from the pre-debate standings, when 21 per cent were undecided.

It turns out the pre-debate standings (before a single word had been uttered) were 57% Hitchens, 22% Blair. After the debate they both went up by 10 points each, winning over the undecided.

Bad conclusion: Hitchens won the debate!
Good conclusion: Debate watchers overwhelmingly preferred Hitchens before it began, both men won over an equal number of undecided.

Expecting Blair to get 51% in order to be the 'winner' would be like expecting Barack Obama to win over a group of 57% firm Republican backers in 90 minutes, or Dick Cheney to win over a group composed of 57% Daily Kos users, in order to be declared the winner. Not going to happen. So it seems like the only loser in the debate was statistical comprehension.