As groklaw and other groups show, the internet blog user population has been producing more rapid and insightful criticism than mainstream media. It’s a beatiful thing to see. Powerline’s initial analysis of the suspicious CBS memos began a torrent that threatens to bring down the honour of a news network, even as opposing groups try to present contrary findings. Amongst the noise of “me toos”, there is an aggregation of deep thought that I considered the sole domain of Usenet in its heyday.

As for whether the memos are fake or not … one idea I’ve not seen elsewhere is analyzing the text layout more quantitatively. If indeed there was a 1970s era typewriter that can make a document look like that, and if in fact such a machine was used by the deceased, it may be possible to measure more carefully the horizontal layout granularity of individual characters. Somewhat like the Millikan oil drop experiment, there should be a quantum or minimum amount of distance, whose multiples all characters line up with. Someone might take a high-resolution scan of the memos, and measure carefully the horizontal offset of each character relative to a margin or tab line. All these distances would have to be an integer multiple of some greatest common denominator.

For a modern typesetter or laser printer, this quantum would be too small to be seen, since the imaging pixel resolution is so high. For the rather impressive-sounding old IBM Selectric Composer, this quantity seems like it should be one-third of the width of the “i” character, based on this IBM technical paper. Other manuals for the machine are available at this web site . It’s a marvel of mechanical engineering, but can it clear CBS of the forgery charges? Probably not .

UPDATE: Someone else had a similar idea.