[DaxtonBrown]

[If you scan a document, the whole document will be made up of identical sized pixels... ]

This actually seals the deal. One could argue that optical character recognition software had somehow separated layers. But that wouldn’t have caused pixel changes.

[Gondring]

Please see below. Note that the pixel sizes vary from place to place on the PDF.

The image I'm showing is merely to illustrate what I see on my screen. But, to confirm that I didn't do anything funny with it, here's what I did:

1. Scanned document to a non-searchable PDF (i.e., without OCR); I had similar results for various presets on the Ricoh machine. It seems to be the compression algo that the machine uses, not anything to do with the quality or resolution.*
2. Zoomed into the area of the PDF and grabbed a screen snapshot that I pasted to MS Paint.
3. Added some quick labels with MS Paint.
4. Saved image as 24-bit BMP.
5. Uploaded file and confirmed that the site's conversion to JPG didn't change the image upon casual view.

Sure, there might be funny business...but the pixel size doesn't provide "proof" of anything.

*Also, IIANM, it's been noted for years that Quartz PDFContext (what seems to be the PDF writer used for the PDF provided by the White House) has problems and shouldn't be used for important documents such as those going to press. A Google search (or some of the experts here) should reveal details.