Sunday, September 12, 2004

Modern Times

With the New York Times reporting that a key 60 Minutes source has turned on CBS, their earlier decision to "stand by their story" has doubled a bet on a losing hand. Retired General Bobby Hodges of the Texas Air National Guard repudiated the documents which CBS said he would corroborate.

Sept. 11 - A former National Guard commander who CBS News said had helped convince it of the authenticity of documents raising new questions about President Bush's military service said on Saturday that he did not believe they were genuine. The commander, Bobby Hodges, said in a telephone interview that network producers had never showed him the documents but had only read them to him over the phone days before they were featured Wednesday in a "60 Minutes" broadcast. After seeing the documents on Friday, Mr. Hodges said, he concluded that they were falsified.

Worse, Hodges virtually accused the network of deceptive journalism. Commenting on the process through which he was interviewed, "Mr. Hodges, 74, who was group commander of Mr. Bush's squadron in the 147th Fighter Group at Ellington Field in Houston in the early 1970's, said that when someone from CBS called him on Monday night and read him documents, 'I thought they were handwritten notes.'" They were not; they were supposedly typewritten notes which may now turn out to be forgeries prepared on Word for Windows.

CBS's last hope had been to show that Colonel Killian -- whose wife maintains did not type -- prepared the documents on an IBM Selectric or Composer. Those probabilities took a dive now that experimental attempts to reproduce the document on such equipment have failed. Worse, Computer Science Professor Robert Cartwright of Rice University (hat tip: Hugh Hewitt) shows that the variable letter spacing based on the adjacency of letters found in CBS's documents was computationally impossible on any mechanical device available in 1973. Modern word processing processing programs, like Microsoft Word, contain information in the font definition which, for example, tuck a small "i" under the overhang of a capital "T". No mechanical typewriter then available could do this.

... in 1971, even the most powerful available computer systems were not equipped to produce documents like the Killian documents. In Fall 1971, I entered graduate school in Computer Science at Stanford. I soon gravitated to the Artificial Intelligence Laboratory, which had the most powerful time-sharing system (a PDP-10) on campus. In either 1972 or 1973, Xerox gave the Stanford Artificial Intelligence Laboratory a prototype xerographic printer called a "Xerox Graphics Printer (XGP)". Two similar prototypes were given to the MIT Computer Science Department and the Carnegie-Mellon Computer Science Department. The programming staff at the Stanford AI Laboratory was thrilled with the gift because it was the first opportunity that computer science research community had to develop software to support printer quality type-setting. The three Computer Science Departments cooperated in developing the word processing programs to support the XGP. I wrote my first published research paper and my doctoral disseration using the XGP in Spring 1976. It would take another decade before comparable word processing systems were available to most computer science researchers on minicomputers running Unix. It would take nearly another decade before they were widely used on personal computers.

The typed text in the "Killian memos" is kerned (check out letter combinations like "fo" and "fe"), but the (IBM) Composer text is clearly not. Kerning is a computationally complex task beyond the capacity of any mechanical typewriter--even one as expensive and elaborate as the IBM Selectric Composer.

The CBS attempt to escape the kill zone and regain the offensive on the Bush National Guard story appears to have failed. By clutching the faked documents closer to the center of their story they may have effectively destroyed their own expose. But the true magnitude of the catastrophe is hinted at by the Los Angeles Times. In an article entitled No Disputing It: Blogs Are Major Players, Peter Wallsten says:

These days, CBS News anchor Dan Rather and his colleagues at the network's magazine program "60 Minutes II" are enduring an unusual wave of second-guessing by some of the public and fellow journalists. For that, they can thank "Buckhead." It was a late-night blog posting by this mystery Net-izen that first questioned the validity of documents Rather cited Wednesday as proof that George W. Bush did not fulfill his National Guard duty more than 30 years ago.

Although the article half-humorously suggests "Buckhead" is actually Karl Rove, "Buckhead" maintains he acted alone. "But once I posted the comment to Free Republic I was no longer working alone, and that is the real point of the story about the story about the story." The real catastrophe for CBS is that Killian incident is probably not an isolated setback so much as proof that maneuvers which worked in the past can no longer be attempted with impunity. The equivalent of the longbow had arrived on the media scene. When the longbow was first deployed on the European battlefield, it was obviously a formidable weapon.

Such was the power of the Longbow, that contemporary accounts claim that at short range, an arrow fired from it could penetrate 4 inches of seasoned oak. The armored knight, considered at one time to be the leviathan of the battlefield, could now be felled at ranges up to 200 yards by a single arrow. One account recalls a knight being pinned to his horse by an arrow that passed through both armored thighs, with the horse and saddle between!

But it was long years before it was taken seriously. After all, mounted cavalry was the aristocratic weapon and the longbow that of the despised yeomen, the medieval equivalent of bloggers in their pajamas. It took Crecy, Poitiers and finally Agincourt to bring home the fact that the longbow threat was real. As the Christian Science Monitor remarked:

The English longbow plied by yeomen ended the military power and social reign of knights. "Shining" armor fell to a taut string, a cured piece of wood, and a tipped arrow. The military dynamic of the Middle Ages - knight, squire, and armorer - ceased.

It did not bring an end to history: a new dynamic replaced the old, but an era had passed.

Update via Roger Simon. Joseph Newcomer, one of the pioneers of desktop publishing, analyzes the Killian memos and pronounces them fake. He does not outright deny the computational possibility of producing the Killian memos, given then-prototype equipment and esoteric skills, but merely invokes Occam's Razor to pronounce them utterly improbable. Newcomer's conclusion is:

So we have the following two hypotheses contending for describing the memos

  • Attempts to recreate the memos using Microsoft Word and Times New Roman produce images so close that even taking into account the fact that the image we were able to download from the CBS site has been copied, scanned, downloaded, and reprinted, the errors between the "authentic" document and a file created by anyone using Microsoft word are virtually indistinguishable.
  • The font existed in 1972; there were technologies in 1972 that could, with elaborate effort, reproduce these memos, and these technologies and the skills to use them were used by someone who, by testimony of his own family, never typed anything, in an office that for all its other documents appears to have used ordinary monospaced typewriters, and therefore this unlikely juxtaposition of technologies and location coincided just long enough to produce these four memos on 04-May-1972, 18-May-1972, 01-August-1972, and 18-August-1973.

Which one do you think is true? Which one would a 13th-century philosopher think made sense? How many totally unlikely other juxtapositions are expected to be true? How could anyone believe these memos are other than incompetent forgeries?

BTW, I think the phrase "computational possibility" in this context means being able to produce the Killian document from an initial state (typewriter, fonts, hammers, tweezers, etc) given a set of instructions, however long but finite, using the standard operations available on the equipment. The idea is that the the computation "stops" at some point and spits out the Killian document with1973-available operations. Newcomer is saying that a suitable set of such instructions exists but is surpassingly unlikely to have occured to Colonel Killian or anyone in his circumstances.