David Kriesel's sheet demonstrating how his Xerox copier was replacing numbers.[/caption] There were always plenty of ways to mess up a Xerox copy: smudges on the glass, broken scanners, paper jams, toner issues, and sticky notes obscuring a key portion of a document could all affect the final image rolling into the tray. But now a handful of researchers have discovered the tour de force of Xerox errors: under the right conditions, Xerox WorkCentre and ColorQube machines can change individual numerals on pages filled with figures. A German Ph.D. student named David Kriesel discovered the problem while scanning building plans using a Xerox WorkCentre. After checking the plans, he found that the scanned versions included three room measurements that were different from the originals. In each case, the number "6" had been converted into an "8." Since software is always the most obvious culprit in any odd problem involving electronics, Kriesel turned off the optical character recognition (OCR) functions to be sure the machine wasn't simply reading the numbers wrong. But the problem kept happening. "This is not an OCR problem," Kriesel wrote. "It is a lot worse—patches of the pixel data are randomly replaced in a very subtle and dangerous way: the Scanned images look correct at first glance even though the numbers may actually be incorrect." Since the machines didn't call attention to the problem, he theorized, it could easily result in incorrect invoices, blueprints, schematics, and other vital documents. Picture medical prescriptions with mistakes in dosages, or construction plans with the wrong load-bearing figures. Kriesel blogged about the problem August 1, a few days after reporting the problem to Xerox, and touched off an unexpectedly loud response from others who had also noticed the same problem. On August 6, one reader reported a notice in the administration panel of his copier warning about the potential for "character substitution" and recommending the copier's resolution be reset from normal to "higher." Xerox eventually responded with an explanation (PDF) stating that the problem is due to a bug in the scanning function that occurs only on "stress documents" that "have small fonts, are difficult to read, contain stray pixels and/or have been scanned multiple times." It also issued a 32-page guide (PDF) to resetting the default quality/file size settings for WorkCentre and ColorQube models. But an actual patch won’t ship for about three weeks. Image: David Kriesel Xerox Copier Bug Will Take Weeks to Fix
[caption id="attachment_11580" align="aligncenter" width="567"] David Kriesel's sheet demonstrating how his Xerox copier was replacing numbers.[/caption] There were always plenty of ways to mess up a Xerox copy: smudges on the glass, broken scanners, paper jams, toner issues, and sticky notes obscuring a key portion of a document could all affect the final image rolling into the tray. But now a handful of researchers have discovered the tour de force of Xerox errors: under the right conditions, Xerox WorkCentre and ColorQube machines can change individual numerals on pages filled with figures. A German Ph.D. student named David Kriesel discovered the problem while scanning building plans using a Xerox WorkCentre. After checking the plans, he found that the scanned versions included three room measurements that were different from the originals. In each case, the number "6" had been converted into an "8." Since software is always the most obvious culprit in any odd problem involving electronics, Kriesel turned off the optical character recognition (OCR) functions to be sure the machine wasn't simply reading the numbers wrong. But the problem kept happening. "This is not an OCR problem," Kriesel wrote. "It is a lot worse—patches of the pixel data are randomly replaced in a very subtle and dangerous way: the Scanned images look correct at first glance even though the numbers may actually be incorrect." Since the machines didn't call attention to the problem, he theorized, it could easily result in incorrect invoices, blueprints, schematics, and other vital documents. Picture medical prescriptions with mistakes in dosages, or construction plans with the wrong load-bearing figures. Kriesel blogged about the problem August 1, a few days after reporting the problem to Xerox, and touched off an unexpectedly loud response from others who had also noticed the same problem. On August 6, one reader reported a notice in the administration panel of his copier warning about the potential for "character substitution" and recommending the copier's resolution be reset from normal to "higher." Xerox eventually responded with an explanation (PDF) stating that the problem is due to a bug in the scanning function that occurs only on "stress documents" that "have small fonts, are difficult to read, contain stray pixels and/or have been scanned multiple times." It also issued a 32-page guide (PDF) to resetting the default quality/file size settings for WorkCentre and ColorQube models. But an actual patch won’t ship for about three weeks. Image: David Kriesel
David Kriesel's sheet demonstrating how his Xerox copier was replacing numbers.[/caption] There were always plenty of ways to mess up a Xerox copy: smudges on the glass, broken scanners, paper jams, toner issues, and sticky notes obscuring a key portion of a document could all affect the final image rolling into the tray. But now a handful of researchers have discovered the tour de force of Xerox errors: under the right conditions, Xerox WorkCentre and ColorQube machines can change individual numerals on pages filled with figures. A German Ph.D. student named David Kriesel discovered the problem while scanning building plans using a Xerox WorkCentre. After checking the plans, he found that the scanned versions included three room measurements that were different from the originals. In each case, the number "6" had been converted into an "8." Since software is always the most obvious culprit in any odd problem involving electronics, Kriesel turned off the optical character recognition (OCR) functions to be sure the machine wasn't simply reading the numbers wrong. But the problem kept happening. "This is not an OCR problem," Kriesel wrote. "It is a lot worse—patches of the pixel data are randomly replaced in a very subtle and dangerous way: the Scanned images look correct at first glance even though the numbers may actually be incorrect." Since the machines didn't call attention to the problem, he theorized, it could easily result in incorrect invoices, blueprints, schematics, and other vital documents. Picture medical prescriptions with mistakes in dosages, or construction plans with the wrong load-bearing figures. Kriesel blogged about the problem August 1, a few days after reporting the problem to Xerox, and touched off an unexpectedly loud response from others who had also noticed the same problem. On August 6, one reader reported a notice in the administration panel of his copier warning about the potential for "character substitution" and recommending the copier's resolution be reset from normal to "higher." Xerox eventually responded with an explanation (PDF) stating that the problem is due to a bug in the scanning function that occurs only on "stress documents" that "have small fonts, are difficult to read, contain stray pixels and/or have been scanned multiple times." It also issued a 32-page guide (PDF) to resetting the default quality/file size settings for WorkCentre and ColorQube models. But an actual patch won’t ship for about three weeks. Image: David Kriesel 
