When you say a “nice” pool, how big is that pool? Are there many examples of it being exploited on a scale that affects election outcomes?

I agree there is a risk that bad actors could exploit this vulnerability, but there is also a risk purging voters prevents legitimate voters. That is especially concerning when the purges come from a government controlled by one party in an area heavily populated by voters of another party.

So it is not cut and dried. There are reasonable arguments on both sides, but my personally calculation is it is less likely someone gets away with finding people who have moved and casting votes on their behalf. It is more likely that a bunch of politicians purge voting rolls in districts that lean the other direction. Especially as we have 150 years of evidence of the latter but few examples of the former.

The risk of purging legitimate voters on a scale that could actually affect the election results is substantially overstated, from what I can tell. In particular, there was a claim which went viral after the 2016 election that this had actually happened in the swing states whose (award-winning journalist) author clearly knew it couldn't work as advertised. It compared the number of entries on the purge lists with the victory margin, pointed out it was smaller, and claimed this proved the vote was rigged that way. There were two huge issues - the first is that many people genuinely wouldn't have been eligible to vote in those states, and the second is that in at least one and probably all three states being on the purge list wouldn't actually stop the people from voting if they did turn up at a polling station, and all three were necessary for Trump to win. (People are only actually removed from the register if they don't vote for a while after being added.) The author clearly knew this, because buried in the article was an odd little disclaimer about how what happened when people were added to the purge lists varied from state to state without any further details. The ACLU and the press in general also like to run articles claiming every name removed is a "suppressed vote".

I think that may have been after just a few years of not being able to purge voters due to lawsuits from the ACLU, so there is definitely a big enough pool of registered voters who are likely no longer living in that state to potentially swing elections at least some of the time. If I remember rightly, some recent key swing states have a lot of outward migration compared to the victory margins.