Evidently, there are some fundamental errors made by the original reporter that change the timbre of this story altogether. This report has Joseph Lorenzo Hall of the Centre for Democracy and Technology in DC, asserting very strongly that the tabulation machines are “air-gapped” — the tabulation results from the original voting system are in actuality walked over manually (via a data export to, say, a thumb drive or flash card) to the tabulation machines. Apparently, no code run on those machines can access the primary system because they’re isolated. So what the code has write access to, then, is apparently the export of the database, not the originals in any way.
It still means that processes should be followed to ensure the integrity of the data, to ensure that the exported data matches the CSV conversion. But I suspect these folks are more “with it” than I’d originally thought.
See below the fold for my original story.