Putting the "Twit" Back in TwiXter

​Ah, Sweet Twitter, what hath thou become under MusXk...

Yesterday, in response to a tweet left by new White House press secretary Karoline Leavitt that was a song-and-dance attempt to explain away something that wasn't true, I wrote the following --

"You have started your tenure VERY poorly for someone who said she 'was about the truth.'  This below at best is a gobbledy-gook to explain a jumble. You lied about the military going into California. And lied about $50 million meant for Gaza that supposedly went instead for bombs."

To my surprise, I received an email from the folks at TwiXter that said --

​
Needless-to-say, after ungnashing my teeth and unclenching my typing fingers, I saw that at the bottom of their email, I was allowed to appeal their algorithm's "decision."  I did not, explaining that I merely criticized the press secretary for lying and there was nothing in my tweet that promoted violence or threatened, and nothing was based on age, gender, religious or whatever.  (I'd have said much more, but I was limited to only 280 characters.)

I explained this online, and then rephrased my original tweet, trying to figure out some way to tone down what was only bluntly critical but not threatening in any imaginable way.  I trimmed out a few words and also cut out the reference that Ms. Leavitt herself made to bombs, since I thought that that word might be been what the algorithm saw and thought was a "threat."  And so, my post now read this kinder, gentler way --

Moments later, also to my surprise, but this time a happy surprise, I received a quick follow-up from TwiXter to my appeal that said --

​
Oh!  How nice.  Honestly, I didn't expect that.  And so quickly.  Sanity ruled the day.  So, okay, good!

But then came the funnier part. 

Much to my, once again, surprise -- after I had toned down my tweet (written before getting their ruling that it was perfectly OK), the algorithm limited my toned down tweet that -- which moments before they had just ruled that the original tweet was fine!

I decided not to challenge the "limited visibility" of the rewritten tweet, since the original was more pointed and better.  And I just deleted the rewrite.

Mind you, I still have absolutely no idea what the algorithm saw that it considered even remotely violent or threatening about my original tweet, since (in my rewrite) I deleted what I thought was the only conceivably algorithmically-threatening word, "bomb" -- and it still got tagged.

Some days, it seems you just can't trust a MusXk-based algorithm.  Go figure.