— Or the end of a ‘god of the gaps’ argument —
The Cold Case Posse released the following document on March 1, 2012 as a preliminary release of their findings. In it, its experts outline how they reached a conclusion of “forgery”. Note that they never really identified a ‘forger’ other than by claiming that regular scanning processes would not create specific artifacts.
Zullo appears to have understood this when he issued a challenge to show how a simple workflow would indeed create artifacts similar to the ones found in the WH Long Form Birth Certificate.
And here my challenge started. While it was almost self evident that the artifacts were the result of a Mixed Raster Compression process, the Cold Case Posse as well as its critics, had remained unable to locate a specific workflow/scanner that would result in a PDF that matched ‘all’ of the artifacts. Until serendipity and some creative digging unearthed not only a wealth of confirming evidence, but also led to the identification of a likely ‘suspect’: A Xerox WorkCentre…
Thus my task is relatively simple:
Show how a repeatable workflow creates the artifacts identified (and more). My hypothesis is not even competing with a positive hypothesis but rather one based on what is commonly known as an ‘argument from ignorance‘ aka a ‘god of the gaps’ argument. The ‘argument’ goes as follows: We do not understand how ‘xxx’ could have resulted in ‘yyy’ and therefor ‘yyy’ must have been ‘designed’/’forged’. Of course, when one can show how a simple process can result in ‘yyy’, the conclusion of forgery is significantly weakened. When one can show how a simple workflow can explain close to a dozen ‘artifacts’, any argument from ignorance is totally destroyed.
Of course, simple does not mean trivial and it took some careful investigation to locate some predictors (the quantization tables and the YCbCr jpeg comment which appear to have been overlooked by the CCP experts) and then, through serendipity I ran across a scan created on a 7655 Xerox WorkCentre, followed by one found on one owned by the White House… Then the pieces started to slowly fall into place while more and more ‘features’ found in the WH LFBC pdf were matched to the workflow. And note that several of these ‘artifacts’ had been identified before the ‘forger’ had been located…Quite powerful…
Mara Zebest, raises various issues that disappear when the workflow involving a Xerox WorkCentre is understood. She came close to understand it, when she observed that
Examine how the layers divide the image into pieces. It is analogous to taking a scissor and cutting the image into random rectangle.
This is a somewhat crude description of the Mixed Raster Compression technique used. So let’s look at some of the failed conclusions.
The document was indeed divided into pieces, in a process that is called Mixed Raster Compression. The algorithm separates the background from foreground text in approximately the same color. The background is encoded in highly lossy JPEG and sub-sampled to 150 ppi, while the foreground is represented as a colored bitmask, sub-sampled to 300 ppi. This process causes exactly the appearance of ‘anti-aliasing’ versus bitmapped text, and which letters become part of the a foreground bitmap or the background, depends on the details of the algorithm. It is indeed, the ‘random’ distribution of letters over the foreground and background that defies a forgery conclusion as one would have to explain why a forger is so sloppy.
In fact, a simple experiment with a Xerox WorkCentre shows exactly this effect. And the preview step causes an extra clipping mask. This is a repeatable process and totally runs counter to this claim about a ‘normal scan’ being a flat file. This is perhaps true for lower end desktop scanners but high end devices offer a large variety of ways to capture a document.
Well, yes, by the Xerox MRC algorithm. Again, this is a repeatable process where the images, which are captures in landscape form have to be rotated for proper display. In addition, the 300 ppi images have to be scaled 24% to fit the 72 PPI settings for the PDF, and the 150 ppi image has to be scaled 48%. This was already explained before the Xerox had been identified. Anyone familiar with how MRC works, could have understood the scaling aspects.
The safety paper is ‘painted’ first and the white halos are a side effect of the MRC process, which at least logically can be understood and has been observed in real Xerox created documents. It also does no make sense from a forgery perspective that the background layer would be done so poorly. The CCP has never explained why the forger appears to have been so ‘incompetent’… A simple work flow however explains it all.
Note that John Woodman contacted an expert who also observed
If it was a forgery it was a very sloppy job. Any photoshop-knowledgeable person, of the garden variety, can do a much better job than that. If it is automated, it is a lousy job too, but bear in mind that algorithms for these jobs are not trained on specific documents. They were more likely developed, trained and tested on magazine pages and books. A US birth certificate is unlikely to give good results because it may be an outlier in the big picture of all documents they had in mind when developed their MRC tool.
In summary I can only say I see much stronger signs of common MRC algorithmic processing of the image rather than some intentional manipulation.
Ricardo L. de Queiroz
Now Ms Zebest did manage to put to rest an early explanation of OCR, which however was quickly rejected by those who had proposed it.
As to the ‘smiley face’ and the ‘misspelling’ there are quite reasonable explanations for this that do not require a jump to an accusation of ‘forgery’. These will remain minor mysteries of little relevance.
Ms Zebest also raised the issue of x-ray, where the ‘white’ behind the text that was separated from the background showed color texture. Again, this is a predictable side effect of JPEG encoding and has been successfully repeated in experiments.
Now we get to the stamp layer, which nicely separates as one layer. Which is of course easy to understand, realizing that the stamp ink has a distinctive color that allows the MRC algorithm to detect it. In some of the experiments, full lifting of the stamp layer was achieved.
As to the attempt to recreate the Halo, sufficient to say that the example fails to even get close to the observed effects. But that is far less damaging than the simple observation that Xerox scanned PDF’s show the same ‘x-ray’ and halo effects.
Others who have fallen victim to the Gap Argument…
I’ve read your book. I give you credit for trying, but your conclusions are wrong. Neither adaptive nor MRC compression can result in the type of layers seen in Obama’s birth certificate. I’ve been testing this for over 6 months and can prove it conclusively.
The fact is, until you provide even one example of optimization creating multiple 1-bit image mask layers, your hypothesis is untenable.
And it isn’t just an issue of using the right combination of settings. The fact is that both Adaptive compression and MRC compression only create a single 1-bit image mask layer. Therefore it is not possible that these processes created the layers on the WH PDF. The WH PDF defies the characteristics of an optimized file.