Most of the competitions I scrutineered started with recalls of 24 or 48 numbers, and 3-7 judges. Each judge writes down the numbers of the couples being recalled (so 24 or 48 of them in a typical first round), and then those are combined with the couples being marked by the greatest number of judges being advanced to the next round. So I'd enter the marks for each judge on my summary sheet, add up the total received by each couple, figure out what the "threshold" should be to advance, and then write down the couples to be recalled for the MC to announce (and also on my own next sheet for dancing the next round). Typically there would be about 2-3 competitions going at a time with the rounds interleaved. (So we'd do the newcomer latin first round, then the bronze latin first round, which might or might not have some of the same couples in it, then the silver latin first round, then back to the next round of the newcomer, etc.) During the bronze latin first round, I'd be entering the scores from the newcomer, then during the silver, entering the scores from the bronze, etc.
I never bet the farm on noticing MC errors - but I did catch them from time to time. I don't think there were a lot of them that I missed, because I didn't have to field many complaints from couples who looked at a sheet and saw that they weren't announced when they should have been.
If your DS is doing it like I do, one telltale is that if you ask him a comprehension question immediately after you finish the book while he is still talking at you, there should be a short delay while he "reads back" the auditory stream before he gives you the answer. If he stares into space for a minute then comes up with it, that's a clue. If he gives it to you immediately, then I think he is comprehending it as he is listening, so he's doing something slightly different.
