[the block decline of flags at the end yesterday] was indeed to send the message of not to take this sort of "shotgun" approach to flagging.
I'm sorry that this essentially boils down to some sort of miscommunication since not only did I have the above line, minus the bold, but very much there and explicitly saying what not to do, but I also figured that the many lines I had spoken about what is bad and not helpful, versus what is efficient and organized, would be directly implying the underlying intent without me needing to explicitly say "Don't do it". My bad, I should've been more direct and I understand that now, I messed up. I think underlying messages should be inherent and sticking only to what I said is a bit self-deprecating on the reader's part, though. But without further ado...
Don't do it.
Before I went to bed last night, I was working with others on the community team to compose a general announcement about these scripts. At that point I also got a complaint from the moderators that the walls of flags had returned, and so I piped up about that before wanting to sleep on this and come back at it with a rested head. In the mean time, my team was the one that mass-declined your recent flag load. It was only done on Arqade because the rest of the team was only aware that you were doing this on Arqade (and Arqade was the one, again, who poked us). You were in fact explicitly told why there was a decline in the previous incident, and so you persisted, and so we continued to decline for exactly that reason.
Here's pretty much what I got to writing last night with Shog9's help:
Do not flag based purely on naive keyword searches. If I thought a simple keyword search was sufficient to remove these comments, I would remove them en masse and cut out the middlemen.
If you come up with a system for identifying noisy/obsolete comments with a very, very low false-positive rate, post it up on meta & we'll just clear 'em all out. If you want to test your system by using the results to flag small batches, go for it - but don't just flood the queue, especially if you're still in the early "naive keyword search" stages.
If you refuse to follow these guidelines and flood the flag queue with what amounts to noise, your ability to continue flagging at this rate - and possible a large chunk of your past flags - will be removed.
This goes for everyone working on these scripts.
We're actually thinking of blocking this script from functioning any further on our API.
Arqade isn't the only site that has had issue with these flags. They are, however, the most recent ones that came to us about it. As well, different teams take different approaches - in this case the moderators had noted, which I relayed in our exchange, that they do like to check context on these things and don't always agree that all of these needed to be removed. A clarifying comment can be signal for some folks, and acknowledgment of the request for clarification is signal when joined with that kind of signal. That's how it worked over here. I can't speak to the experiences on other sites as I haven't analyzed the latest in flags or interrogated the mods there.
Yes, moderators are the only ones who can handle these. But this approach to flagging is far from an ideal means to get these taken care of. Rather than strongly urge you as before, I'm going to outright say it, if your main interest is clean up then the best way to get things done is the way that is most organized and most useful and helpful to those who actually have to do the job behind it. Your comfort level with regards to how much you're willing to do at once, your ability to do 200 edits in a single day, that's nice but it's not comparable to doing flags because that is no longer on what your comfort level is. And comfort dictates how well the job gets done.
I've not gotten to read the comment exchanges here and I shall be doing so shortly, but I wanted to get this message out first.