I agree that 200 is an obscenely high number and can't think of anything kind to say about anyone who'd feel the need to have so many, but I did do a quick test a while back where I backed up the ignores table, emptied it, observed performance, and restored the data.
No detectable performance difference between 200 per person and 0. Lots of difference between 1000 and 200, though. And we have db horsepower to spare again anyway. Although it's quite possible that when I did my testing, there weren't any or many 200-ignore people actively using the site.
The performance issue isn't so much the total size of the ignore table. 10,000 people having 200 each would likely have little impact on performance. It becomes costly when someone with a really large ignore list is reading messages, because the database has to do a LOT of work to grab the 50 most recent messages in a board NOT written by (and sometimes to) a list of 1000 people.
