If iabot has done some kind of editing, but an user has overridden the edit, most likely with a revert, then do not try to redo the edit.
Check if a previous edit has been done, and later been partial or fully removed, and if so do not redo the same edit.
Cost of implementation is too high.
Closing as technically infeasible. IABot has opt-out methods in place to prevent edit warring. Please refer to the FAQ.
I don't think you will be granted rights on nowiki to run this bot unless this is fixed.
I will post your comment on nowiki, and it should be up to the community to decide.
Some additional info
Note that I am not the one blocking this bot, and that I can not do anything about the blockage.
Another trigger-happy admin apparently. I was not informed of a block, and the bot has a run page. If anyone ever bothers to read the instructions to shut off the bot.
I can hardly understand any of that, and no one is bothering to inform me of any issues, and are instead just blocking the bot. I cannot fix anything if I'm not aware of problems. And rather having the bot pick up when it gets reverted and to not repeat the edit, it's better to know why the edits are being reverted, and to fix the root cause. I need diffs for that, examples, something to investigate.
So Bing is useless.
Yandex seems to work. The first linked thread, is already resolved.
Paged through 250 contributions and checked those that had later changes
That gives about 4% failures of some kind, which is too much in my opinion.
So this goes back to my feature I am working on T172737.
Why aren't these being reported at https://tools.wmflabs.org/iabot/index.php?page=reportfalsepositive in the meantime? It raises my attention and I investigate them quickly. Reporting them brings the failure rate down.