These images compare the original PDF file to a Word document with various names replacing the redactions. The original file is in red, the comparison Word document is in blue.
Armitage has long been suspected of being Novak's original source. Fleischer and Rumsfeld were chosen because in Word 'Fleischer' is slightly shorter (about 1.6%) and the 'Rumsfeld' is slightly longer (about 3.3%) than 'Armitage'. Grossman was added because he was also deemed to be a likely Novak source; though, as can be seen by the images, his name is almost certainly too long.
The PDF excerpts were produced by using the "select region" feature of Adobe Reader to copy the region of the document, scaled to 150%, to the clipboard. The Word excerpts were produced by displaying the region on the screen in Word, also scaled to 150%, then using "print screen" to copy the screen image to the clipboard.
It may seem more sensible to actually print the Word document, then scan the printed page, since this is what was done to produce the original PDF. However, this adds more mechanical variation into the system. I did try printing and scanning a Word document, then comparing the results with a screen-capture, and the two images seemed to align quite closely.
The excerpts from the original affidavit PDF file were scaled by approximately 97% to match the Word versions. Because even the non-redacted regions didn't always align, both the scaling and the exact placement over the comparison versions were somewhat subjective processes. To minimize unconscious bias, I determined the scaling and placement in comparison to the unredacted areas of the 'Grossman' sample (which I was confident wouldn't match in the redacted region), then copied it over the other samples in the same relative position.
----------Double Click Images to See at Full Scale-----------
Page 14
--------------------------
This section names Novak's source.
There are obviously several non-redacted sections where the characters are not aligned, making it more difficult to reach a conclusion about which name fits best. Clearly, judged by the remainder of the total line, 'Armitage' is the best fit. However, the right parenthesis and period following 'Fleischer' appear to be in exact alignment, and the section of the line with the most shift is below a section that has a similar shift; yet, the following line is in close alignment in the same region. Both 'Fleischer' and 'Armitage' are reasonable possibilities.
Page 15:
--------------------------
This section names a source without specifing whose, followed by Woodward's source, then Novak's source. (The partial line "Moreover, ..." is not included in the comparison Word document, since its exact starting position is unknown.)
For the unattributed source, both Armitage and Rumsfeld fit well. For Woodward's source, 'Armitage' seems to be the best match, with 'Rumsfeld' as second best. 'Fleischer' seems too short, but is possible. For Novak's source, 'Armitage' is an excellent fit, while both 'Fleischer' and 'Rumsfeld' are close.
The big question is whether both sources are the same. On the one hand, if they are, the phrasing is rather unusual, repeating the same name over and over, like a car salesman talking to a customer. On the other hand, the use of the singular "piece of information" suggests only a single name has been withheld. Also, the fact that following "source for" the entire phrase is redacted seems to indicate the source can't simply be assigned to either Novak or Woodward, so perhaps it's both.
Page 16:
---------------------------
This section names Woodwards source twice.
This is the most useful redaction, since by repeating the same name twice the offset is doubled. Only 'Armitage' seem to fit; and the fact that the letter offset increases from the first to the second redaction suggests that the shifts are not due to some other factor. This is very good evidence that the last name of Woodward's source is almost exactly the length of 'Armitage'.
An additional bit of evidence is a speck following "with" in the
original document that aligns well with the left-lower serif in the 'A' in 'Armitage'.
Page 18 comment:
---------------------------
This section names Woodward's source.
Clearly, 'Armitage' is the best fit, though 'Fleischer' is reasonably close. There are shifts in the non-redacted sections that make any conclusion somewhat less certain. If only the line-break had not split the first and last name! It may be noticed in the 'Grossman' example that the line beginning "In addition" following the redaction aligns almost exactly. This was achieved by including one fewer spaces after the sentence then in the other versions. Notice the final 'e' in the Word version extends a full
character-width beyond the 'e' in the original PDF version. It's a
curious fact, however, that producing the spacing between the period and "In" seen in the original document requires three spaces, so the 'Grossman' version has the normal two.
Armitage has long been suspected of being Novak's original source. Fleischer and Rumsfeld were chosen because in Word 'Fleischer' is slightly shorter (about 1.6%) and the 'Rumsfeld' is slightly longer (about 3.3%) than 'Armitage'. Grossman was added because he was also deemed to be a likely Novak source; though, as can be seen by the images, his name is almost certainly too long.
The PDF excerpts were produced by using the "select region" feature of Adobe Reader to copy the region of the document, scaled to 150%, to the clipboard. The Word excerpts were produced by displaying the region on the screen in Word, also scaled to 150%, then using "print screen" to copy the screen image to the clipboard.
It may seem more sensible to actually print the Word document, then scan the printed page, since this is what was done to produce the original PDF. However, this adds more mechanical variation into the system. I did try printing and scanning a Word document, then comparing the results with a screen-capture, and the two images seemed to align quite closely.
The excerpts from the original affidavit PDF file were scaled by approximately 97% to match the Word versions. Because even the non-redacted regions didn't always align, both the scaling and the exact placement over the comparison versions were somewhat subjective processes. To minimize unconscious bias, I determined the scaling and placement in comparison to the unredacted areas of the 'Grossman' sample (which I was confident wouldn't match in the redacted region), then copied it over the other samples in the same relative position.
----------Double Click Images to See at Full Scale-----------
Page 14
--------------------------
This section names Novak's source.
There are obviously several non-redacted sections where the characters are not aligned, making it more difficult to reach a conclusion about which name fits best. Clearly, judged by the remainder of the total line, 'Armitage' is the best fit. However, the right parenthesis and period following 'Fleischer' appear to be in exact alignment, and the section of the line with the most shift is below a section that has a similar shift; yet, the following line is in close alignment in the same region. Both 'Fleischer' and 'Armitage' are reasonable possibilities.
Page 15:
--------------------------
This section names a source without specifing whose, followed by Woodward's source, then Novak's source. (The partial line "Moreover, ..." is not included in the comparison Word document, since its exact starting position is unknown.)
For the unattributed source, both Armitage and Rumsfeld fit well. For Woodward's source, 'Armitage' seems to be the best match, with 'Rumsfeld' as second best. 'Fleischer' seems too short, but is possible. For Novak's source, 'Armitage' is an excellent fit, while both 'Fleischer' and 'Rumsfeld' are close.
The big question is whether both sources are the same. On the one hand, if they are, the phrasing is rather unusual, repeating the same name over and over, like a car salesman talking to a customer. On the other hand, the use of the singular "piece of information" suggests only a single name has been withheld. Also, the fact that following "source for" the entire phrase is redacted seems to indicate the source can't simply be assigned to either Novak or Woodward, so perhaps it's both.
Page 16:
---------------------------
This section names Woodwards source twice.
This is the most useful redaction, since by repeating the same name twice the offset is doubled. Only 'Armitage' seem to fit; and the fact that the letter offset increases from the first to the second redaction suggests that the shifts are not due to some other factor. This is very good evidence that the last name of Woodward's source is almost exactly the length of 'Armitage'.
An additional bit of evidence is a speck following "with" in the
original document that aligns well with the left-lower serif in the 'A' in 'Armitage'.
Page 18 comment:
---------------------------
This section names Woodward's source.
Clearly, 'Armitage' is the best fit, though 'Fleischer' is reasonably close. There are shifts in the non-redacted sections that make any conclusion somewhat less certain. If only the line-break had not split the first and last name! It may be noticed in the 'Grossman' example that the line beginning "In addition" following the redaction aligns almost exactly. This was achieved by including one fewer spaces after the sentence then in the other versions. Notice the final 'e' in the Word version extends a full
character-width beyond the 'e' in the original PDF version. It's a
curious fact, however, that producing the spacing between the period and "In" seen in the original document requires three spaces, so the 'Grossman' version has the normal two.
posted by Rick Ballard at 1:34 PM
<< Home