- Messages
- 35
The recent piece on the de-facto federal gun registry at LegalInsurrection
https://legalinsurrection.com/2026/02/congress-demands-answers-from-atf-over-backdoor-gun-registry/ referenced other work at Reason
https://reason.com/2026/02/13/the-atf-created-a-backdoor-gun-registry-lawmakers-want-an-explanation/ which mentions "PDF and JPEG". Sure, PDF and JPEG are very commonly used by the ignorant for scanning and photographing high-contrast printed forms. Anybody see the problem there? JPEG is closely related to evil, and has no place in any chain of investigation or evidence. The details on howJPEG is used in PDF (and PS) are at https://www.adobe.com/jp/print/postscript/pdfs/PLRM.pdf, and you can go get the JPEG specification at their website.
JPEG is valid only for image that have smooth changes in color value, like photographs. Indeed the "P" in JPEG stands for "photograph". When the JPGE transform is used for high-contrast image data, like say scanning or photographing a form, the result is usually bad. (Note that lossless JPEG does exist and is about as common as winning lottery tickets. You may never see it or hear it mentioned.) It is bad in a crucial way: it inserts pixels into the digital representation of the original that do not correspond to the original. That is to say that, in the case of scanned forms, JPEG encoding makes up data.
Then OCR is done on the resulting image data, and the OCR engine guesses at what the text was in the original form. The guesses are then inserted as invisible character objects atop the JPEG-compressed image data. But mistake most certainly are made.
It is already illegal to have a federal firearms registry. It is beyond idiotic to have a registry made using JPEG-OCR with DCT-injected noise. There is no guarantee that *any* valid information can be recovered from the originals. Of course this makes for wiggle room to have open-ended fishing expeditions. Note the asymmetry: past Democrat administrations were imposing harsh penalties on FFLs for trivial paperwork errors, but the Feds have no limit on errors from scanned forms (in their illegal registry).
There is already a bit of judicial notice on this topic... brought to you by the evils of Hillary Clinton. When required by a federal judge to provide evidence in "digital form", her crack team of obstructionists
1) printed the original vector PDF documents -- thousands of pages
2) sent them by fax to a second fax machine in the same room
3) scanned the faxes that came in
4) saved each one as a low-resolution JPEG-compressed PDF file with a random name
5) handed the resulting mess to opposing council
All the text information was removed... and the evidence was "delivered in digital form". IIRC, sanctions were involved shortly thereafter.
https://legalinsurrection.com/2026/02/congress-demands-answers-from-atf-over-backdoor-gun-registry/ referenced other work at Reason
https://reason.com/2026/02/13/the-atf-created-a-backdoor-gun-registry-lawmakers-want-an-explanation/ which mentions "PDF and JPEG". Sure, PDF and JPEG are very commonly used by the ignorant for scanning and photographing high-contrast printed forms. Anybody see the problem there? JPEG is closely related to evil, and has no place in any chain of investigation or evidence. The details on howJPEG is used in PDF (and PS) are at https://www.adobe.com/jp/print/postscript/pdfs/PLRM.pdf, and you can go get the JPEG specification at their website.
JPEG is valid only for image that have smooth changes in color value, like photographs. Indeed the "P" in JPEG stands for "photograph". When the JPGE transform is used for high-contrast image data, like say scanning or photographing a form, the result is usually bad. (Note that lossless JPEG does exist and is about as common as winning lottery tickets. You may never see it or hear it mentioned.) It is bad in a crucial way: it inserts pixels into the digital representation of the original that do not correspond to the original. That is to say that, in the case of scanned forms, JPEG encoding makes up data.
Then OCR is done on the resulting image data, and the OCR engine guesses at what the text was in the original form. The guesses are then inserted as invisible character objects atop the JPEG-compressed image data. But mistake most certainly are made.
It is already illegal to have a federal firearms registry. It is beyond idiotic to have a registry made using JPEG-OCR with DCT-injected noise. There is no guarantee that *any* valid information can be recovered from the originals. Of course this makes for wiggle room to have open-ended fishing expeditions. Note the asymmetry: past Democrat administrations were imposing harsh penalties on FFLs for trivial paperwork errors, but the Feds have no limit on errors from scanned forms (in their illegal registry).
There is already a bit of judicial notice on this topic... brought to you by the evils of Hillary Clinton. When required by a federal judge to provide evidence in "digital form", her crack team of obstructionists
1) printed the original vector PDF documents -- thousands of pages
2) sent them by fax to a second fax machine in the same room
3) scanned the faxes that came in
4) saved each one as a low-resolution JPEG-compressed PDF file with a random name
5) handed the resulting mess to opposing council
All the text information was removed... and the evidence was "delivered in digital form". IIRC, sanctions were involved shortly thereafter.