My working theory is in the middle between yours and what they said.

This is based on my viewing of the content in question yesterday, and my decades of experience as a software developer, including web page development.

You didn’t have to compare it to the original, it just looked wrong as it was. It looked to me like all of the content starting from Article I Section 8 until the end of Article I was all wrong. Like rendered weird and/or removed. All of a sudden, it was just a bunch of paragraphs that seemed out of place. No headings, for example.

I suspect that some MAGA was trying to change something, rather than delete all of the stuff that was deleted. Because A LOT was deleted. It can’t be a coincidence that these sections are relevant to ongoing criticisms of Trump, but not all of the deleted parts were relevant. Somebody either tried to add/edit an annotation, or they tried to edit the wording of the Constitution directly.

And I suspect that whatever method they use to store the constitution and annotations is extremely error prone. And so, let’s say that the person edited in a quotation mark ("). If the software doesn’t handle characters like that properly, something like that alone can cause problems like what I saw.

So, in that case, they could call it a “coding error” and pretend like they weren’t lying. Both the software for the website and markup for the data can be called “code”.

On top of that, government software is usually done by the lowest permitted bidder, so it’s not surprising if it is basically done by an amateur who doesn’t know how to escape characters. Finally, the incompetence of trying to edit something without reading the instructions that surely exist, and without checking the result for unintended consequences, is exactly the sort of incompetence that I expect from MAGA.