Thursday, August 9, 2012

Errol Morris falls on his font face

We all love Errol Morris, but he needs to read more In Progress—specifically 'Why Smart People Are Stupid' by Myk—or, more modestly, the work by Nobel Laureate Daniel Kahneman.

You may recall Morris' intriguing quiz in the New York Times Opinionator which was referenced in a comment to the 'Why Smart People Are Stupid' article. It appeared that he took his quiz question right from Kahneman. (In truth, however, he took it from David Deutsch’s second book, The Beginning of Infinity.) I think it was an unfortunate choice.

Here is Errol Morris' follow-up where he explains the true purpose of his first article.
The test consisted of comparing the responses and determining whether font choice influenced our perception of the truth of the passage.
In traditional psychological study fashion, the whole thing was a ruse. It was a great idea, and he includes some interesting information, especially an anecdote about how a student's grades went up when he switched to Georgia font. The article is also a wonderful exposition on fonts and their inventors. His subject is interesting, the articles are fascinating, but his conclusion, for all the statistical analysis, is suspect.

I suppose Mr. Morris feels that the content of the question or passage doesn't matter since the only thing that varied between the six groups was the font. The problem is that the passage, which people read in different fonts, has nothing to do with deciding how to answer the question. As pointed out in the passage, the chance of an earth-hitting-astroid is one in 250,000 years, so, in essence, having modern astroid destroying capability adds nothing to our safety, especially since the invention of the automobile. Whether we believe the passage is true or false, in whatever font, makes no difference when we answer the question—at least for those who see its irrelevance. This is a Kahneman moment. 

Perhaps we do change our perception of truth depending on fonts, but the best Morris can say is that after reading a story in a certain font, we may be more incline to answer questions differently. I don't see where truth comes into play.

Of course, we know that truth never comes in sans comic.

UPDATE:
The more I think about this, the more I feel that I shouldn't be quite so quick in dismissing Errol Morris' conclusions. The relative truth of the passage doesn't come into play if we accept the one-in-250,000 years part of the passage. But the font may also keep us from believing that too! So, if we don't believe the stated chance, do we think the chance is greater or lesser? With all this talk of astroids I'm going to say greater, say, one in a hundred or one in a thousand years. So now, because of the font's power, the passage becomes relevant in answering the question. Now we consider how much we believe the passage—font and all. It's the all inclusive nature of the font that lets us discount or accept each and every part of the passage.

However, if it is a truth inducing font, then we believe the one-in-250,000 years, and the passage again remains irrelevant. Hmm…this is a lot more complicated than I thought. Maybe I'm just having a hard time believing Morris' follow-up article(s) because of the font he is using.

1 comment:

James R said...

Further questionable practices by Errol Morris have been revealed by readers of Daring Fireball. "The graphs that accompany the article (for example, this one and this one are highly misleading insofar as they omit the first 10,000 responses from the bottom of the chart, so as to exaggerate the results."