Actually I use Hit catcher for panda's now and Panda Crazy for search jobs only. The search jobs seems to be on a different timer than Panda's so it doesn't cause more pre's in my testing. I use overwatch for alerting to special requesters. Hit finder I'll use when I have a bad day and I need more hits. Hit Finder is a great way to find more requesters with batches. Just let it run for a few hours per day and then check the logged hits for any interesting batches that occur every day.
Just to clarify on this a bit wont this still cause issues with certain timings?
Correct me if I'm wrong but my understanding is the way PC runs its requester searches is by pulling up the requester's profile on MTurk (essentially typing the req name into the search box, but obviously using the rid link instead), if it doesn't see anything then no big deal (no problems in HIT Catcher).
However if it sees something, its gonna turn on a panda for it, and then you'll have problems if both timers are low & you have something running in Catcher?
Even if it doesn't see something running PC Search AND Finder at the low intervals should still cause PREs, just in the scraping of new HITs instead of the panda'ing of them.
The request pools are split, but this is still going to use too many in a # of scenarios. More or less trading off panda pool PREs for scraper pool PREs right?
Again, please let me know if I'm wrong I'm absolutely not intimately familiar with either set of tools (which is why I'm mostly asking! Its def a setup that makes a lot of sense if you have the timings done well) but I do know there are still
some limits to running these kinds of things side-by-side and users who aren't aware of that may end up shooting themselves in the foot pretty heavily.