Failed a few tests recently for reasons that fail

Question

Just failed this test: https://superuser.com/review/first-posts/644425

How did I fail? By commenting. I commented

OMG. That made my heart hurt.

Yes, the OMG wasn't necessary. The entire comment wasn't, but it was a comment of approval on a painful picture of horrific carnage.

But Test Bot's automatic interpretation of my comment was incorrect. I'd guess the pass action on this would be an upvote or No Action Needed review.

There have been a couple other reviews I've done recently and also failed usually due to commenting where TB made assumptions about what my comment meant that were not accurate.

I'm not worried about failing too many tests. I've failed enough and passed enough and have enough rep it's just not an issue.

But I think TB perhaps ought to be tweaked a bit.

While it would be more work for mods (you know we love y'all), perhaps there ought to be two buttons on the STOP, LOOK, LISTEN message following failed audits:

1. You got me
2. I protest

Answer 1

I'd guess the pass action on this would be an upvote or No Action Needed review.

Exactly. Your first thought would, preferably, been to upvote the high quality and helpful content.

Comments should primarily be used to indicate problems with the answer. This answer did not really need any more commentary and I suspect, but have not checked, that this answer got more than enough comments along the lines of "Ahhhh! Bad Thing!" which are not entirely helpful.

It is a nice picture, I admit.

_{just checking...}

22 deleted comments. All of them commenting on a pretty picture without adding much to the conversation or giving any useful insight.

Answer 2

I very much dislike failing an audit. I know I'm not to take it personally, but that's so hard for someone who strives to do everything with excellence. I feel your pain.

But I think an I protest button would create more problems that it would solve. It would invite reviewers to disagree with (typically correct) TB. That invitation would flood Meta with complaints and claims of bleeding, overshadowing the instructive purpose of the audits.

TB, is an automated system. We all know such systems have their flaws. This is not an excuse for the flaws, but a reminder that ultimately what we must reach for in automation is close enough to perfection. As Mokubai pointed out, TB would have granted a passing grade had the action taken during the audit been correct (admittedly, in the strictest sense).

Given the tradeoffs I perceive with enabling users to protest their audit, I think TB's performance is "close enough."