Monday, September 13, 2004

technical issues in the news

boy howdy, the major media just acts ignorant when it comes to technical issues. Two recent issues have made this clear to me. First, the allegations, started at Little Green footballs and other right wing blogs, that the Killian memos must have been faked. As is typical with this kind of allegation, the basis of the allegation is a combination of lies, innuendo, and flawed reasoning.

All of the arguments are based on the supposition that the memos could have been forged by MS Word. This is a patently false supposition. It took me about five memos of typing to reach this conclusion.

An example:

"The probability that any technology in existence in 1972 would be capable of producing a document that is nearly pixel-compatible with Microsoft’s Times New Roman font and the formatting of Microsoft Word, and that such technology was in casual use at the Texas Air National Guard, is so vanishingly small as to be indistinguishable from zero."

Problem: the CBS memos are not pixel-compatible. They are close, but they do not match.

At this point I would ask anybody who really cares to consider the difference between the
following two letters:

l and l.

The letter on the left is what is produced by MS Word using the Times New Roman font. The CBS memos were printed in some font that is very similar to Times New Roman. Unfortunately for the conspiracy theorists, the 'l' in the CBS memos looks like the letter on the right, which is from Courier.

The thing to notice here is the serif in each letter. "Serif" is the technical word for the top bar of a letter like 'l'.
In Times New Roman, the serif is diagonal. In the memo, the serif is horizontal.

That's the end of the argument. I'm tired of hearing crap about how the two memos are virtually identical. They are not. What I've shown is one example of the difference - others exist also. The CBS memos, for example, do not have character homogeneity. A number of different characters look differently within the document itself!! It is virtually impossible to achieve this kind of behavior with a word processor, but would happen naturally with a typewriter.

Now I run into the usual problem dealing with truculent wingers - they refuse to let go of a hypothesis.

Consider the main source of my irritation, the Washington Post, which, incredibly, decided to make this a front-page story based on an Internet rumor.

After leading with the exciting blurb "Documents unearthed by CBS News that raise doubts about whether President Bush fulfilled his obligations to the Texas Air National Guard include several features suggesting that they were generated by a computer or word processor rather than a Vietnam War-era typewriter, experts said yesterday. "

Then they have about eight paragraphs of filler that doesn't say much interesting about why anybody would think it would be a forgery. Then we have
"An examination of the documents by The Post shows that they are formatted differently from other Texas Air National Guard documents whose authenticity is not questioned. "

That's like saying Yankees are baseball players. Yankees wear pinstripes. Strangely, Manny Ramirez does not wear pinstripes. *suspicious expression*

Well, duh, so what.

  • "William Flynn, a forensic document specialist with 35 years of experience in police crime labs and private practice, said the CBS documents raise suspicions because of their use of proportional spacing techniques. Documents generated by the kind of typewriters that were widely used in 1972 space letters evenly across the page, so that an "i" uses as much space as an "m." In the CBS documents, by contrast, each letter uses a different amount of space. "

This is a poorly worded sentences. "Documents generated by the kind of typewriters that were widely used in 1972..." at this point the phrase "kind of typewriters" is undefined, but it is suggested that proportional fonts are somehow uncommon. Well, so what? The Gutenberg Bible has an unusual font; that fact alone does not mean that it's a forgery.

  • "While IBM had introduced an electric typewriter that used proportional spacing by the early 1970s, it was not widely used in government. "
It doesn't matter if it is widely used or not. We have a document. We know it exists. Unless you can prove that it is impossible for it to have been created in the 1970s, a likelihood argument is useless here. Unlikely things happen all the time. Grrr...innumeracy rears its ugly head again.

  • "In addition, Flynn said, the CBS documents appear to use proportional spacing both across and down the page, a relatively recent innovation."
I would disagree about whether proportional spacing is used down the page.

  • "Other anomalies in the documents include the use of the superscripted letters "th" in phrases such as 111th Fighter Interceptor Squadron, Bush's unit. "
As has been amptly demonstrated by now, the "th" superscript is seen in other documents. No, the other "th" doesn't look the same. So what?

When I read an argument like this, I have to wonder if I'm the only person who remembers the 1970s. Back in the 1970s, they had these magical devices known as electric typewriters, and electric typewriters used a ball to type, and often on such a ball, there would be special characters for commonly used symbols, like "th".

And this is on the front page of the Washington Post!

  • "It would be nearly impossible for all this technology to have existed at that time," said Flynn, who runs a document-authentication company in Phoenix.
What the hell does this mean? Either the technology existed at the time, or it didn't! This isn't a likelihood question. If Mr. Flynn were really providing expert analysis, he would not use a phrase like "it would be nearly impossible for all this technology to have existed at that time." The phrasing here means "I have no fucking idea what I'm talking about." If he knows that the technology didn't exist, then he would simply say "The technology didn't exist at the time." This "nearly impossible" caveat is chickenshit. Had the Post cared about the veracity of this story, they could have simply done 5 minutes of research checking the website of IBM. They would have found out that the technology did exist at the time - IBM Selectric Composers and IBM Executives could both have printed this document.

  • "Other experts largely concurred. Phil Bouffard, a forensic document examiner from Cleveland, said the font used in the CBS documents appeared to be Times Roman, which is widely used by word-processing programs but was not common on typewriters. "

Who cares if it was common or not? You would think one media source insinuating that another major source is presenting a forgery would give a little credit to the second media source. But no..somebody at the Post (*cough* Howard Kurtz) has a hunch that the documents "look forged" and then decides that this must be front-page news.

The second issue is the latest Time poll, that has 52% Bush voters and 40% Gore voters. Why did nobody at Time check the poll for bias?