Since we can’t observe how an automated vote count works, one safeguard is the manual audit. By law it is supposed to be random. I posted some views on this previously.
Here are the views of a gentleman (Dr. Mahar Mangahas) of what I might call the Old School.
It appears that the random manual audit (RMA) is mandated in 230 out of 75,471 precincts, or a sample size of the order of 0.3%. This seems rather small, but that seems to be what the law says. However, there is mandate to undertake a wider manual count, if the RMA results in “any difference” between the automated and manual count that suggests that other precincts were “affected by the computer or procedural error.” Presumably, “procedural” covers “human” errors, such as if, for example, the wrong computer program was “burned in” into the machine, or if the machine was tampered through what might be called an “internal rigging.” “Computer error’ could cover such events as when the machine fails to scan certain ballots properly or if it turns out there is a program “bug” that was not previously discovered at the so-called source code review.
Note that 0.3% accounts for about 150,000 votes out of about a (potential) 50 million ballots. Is this sample size enough for what Dr. Mangahas calls “representational value”? I believe Comelec has sufficient mandate of discretion to use a larger sample size if this were to be more sensible.
Dr. Mangahas also emphasizes that the RMA process should be “completely trustworthy,” and he suggests transparency in how the 230 precincts would be chosen, including when, and how the manual count should be made. My best guess is that the RMA can result in what election lawyers call “manifest error” that would provide a window for a pre-proclamation controversy. This is untested legal territory.
To Comelec officials, please consider.