If you want to see an example of another document created in this fashion,
Nathan Golding at the National Review Online addressed the "layers" issue at the time the long form certificate PDF was released, and if you scroll down to his second update, he has another PDF that he generated using the same software, Quartz PDFContext for Mac.
The article by Dr. Krawetz glossed on the reason for the separate color and individual greyscale layers, but there's a better description of the process
here.
Roughly, in order to create a PDF with a reasonable file size, the software does its best to separate the text and foreground images from the background image. The background is then highly compressed into a jpg. Foreground images are only compressed a little, to maintain image quality, and are saved in a separate layer. A "mask" (the "blotch" layer you describe) is used to further save space, since those holes in the background layer will be filled by the text and foreground images. Finally, the extracted text is, as the above link describes, "analyzes the outlines of similar characters in the document, creates an average character template and uses it instead of a character itself." That's why some letters, such as the i in "inside" and the i in "judicial" you point out, appear identical: you're not seeing the two i's as they appear on the original document, but the optimized, averaged i's created by the PDF software.
Because the original birth certificate is so irregular, the process wasn't exactly perfect when it was applied. The software recognized as much as it could of the text and images, which is why you see things like numbers and letters appear in the background jpg, and why some objects (like checkboxes and signature fragments) that should have been recognized as foreground images appear in the text layers.
For a further example of the flaws in Ms. Zebest's analysis of the PDF, regardless of her expertise with Photoshop, let's take the mysterious "Derek's PDF" tag encoded into the long form birth certificate. Zebest (and you) think this may point to White House staffer Derek R. B. Douglas as the possible forger. But it doesn't.
Take a look at
this PDF, for example. It's a PDF hosted on the University of Michigan's Center for Health Communications Research website, last modified back in April of 2008. It has the
exact same "Derek's PDF" tag.
And it's far from the only PDF out there with that tag. The ASCII appears as "Derek's PDF", but in Windows encoding, it appears as ƒÂÚÂÎßÛ. A quick Google search of "ƒÂÚÂÎßÛ pdf" reveals
hundreds of PDFs with the same "Derek" tag.
Why is that? Because these PDFs, like the PDF of the President's long form birth certificate, were created on an Apple computer. In 2007,
Apple, Inc. acquired the CUPS printing software to incorporate into their computer operating systems. CUPS includes, as its PDF filter, a sub-program called
Xpdf. Xpdf was created by one
Derek B. Noonburg (scroll down to the "Legal" section). That's why most PDFs created on an Apple computer contain the tag "Derek's PDF", because they were made using Derek's PDF code.