Until I started using the new suggested-edit review system I wasn't aware that you were capped at voting approve/reject on 50 edits a day (having less than the required rep for the existing tool meant running into posts requiring approval were quite rare).
Since using the review system and taking a look at the stats page, I've noticed that there's a reasonable number of users that are reviewing more than the 50 edits that they can vote on. This can be seen in the screen grab I've borrowed from this question, where you'll notice that John has reviewed 108 posts in one day:
This isn't an attack against John, I'm sure that he's doing good work and he certainly has enough rep to be making the edits by himself anyway but is this a desired effect? Is it an unexpected effect possibly motivated by the 'Top Reviews all time' list?
I have a few concerns (and some of these may be covered by tools available to mods/users with greater rep anyway).
As far as I can tell, when we chose to improve, we can select helpful/not-helpful for the edit we're improving and that approval/reject is marked as community. Is this still tracked against the individual making the decision (I can't see it in their suggested edit review page)? For example, it seems like there would be a reasonable amount of work involved in determining if one user was always selecting 'improve + unhelpful' to reject a particular editor.
When you're using the review system, you see a lot of posts from tags that you wouldn't necessarily normally look at. If you're making decisions on your own in those tags, it's less likely that somebody else will catch any mistakes you make (for example). We all make mistakes (I'm sure there are plenty in my review history), but the current double review systems mitigates that to some extent, a rise in unilateral decisions seems like a bad thing.
If people are regularly hitting the max number of votes, would it be better to increase the number of votes per day?
Would it be better to count improve decisions count in your votes per day, or would this just create an unclearable backlog?