I knew this tidbit from a briefing between the mall management and the retailers’ staff during Jem’s pre-opening and thought it’d be fun to share since even some of the retail staff at the time were surprised to hear this as it looked like an acronym, but it didn’t even get to a vote.
I don’t think the AI is capable of judging the truth of such edits, and having things like this rejected is precisely why some of us don’t bother trying and just stick to simple, boring edits like fixing spelling and capitalisation.
Note: Don’t tell me to appeal this, it’s useless advice. I’m not wasting appeals on edits, because they’re better off getting spend on adding new wayspots.
Well, I don’t think it should be completely unable to reject since there are some sorts of edits like ye olde “please pick the option on the right to reveal hidden pokestop” that are kinda sorta technically abuse (even if I don’t blame the people using edits that way), and obvious nonsense spammy and gibberish edits like “hhasdkjha”.
But I do think Niantic should give submitters more options to appeal AI decisions, like a separate appeal stream for edits so we don’t have to choose between appealing nominations and edits, or perhaps some sort of “request/appeal to community vote”.
I feel like ML might of rejected based on how JEM is written in the title, JEM, versus how you wrote it in ypur description edit, Jem. It’s not even rhat big of an issue to me, as your edit is a lot better, but maybe ML is getting more picky with grammar.
I will say that ML rejected a title edit I submitted for a health sciences building at a local college. I didn’t think the college’s name needed to be a part of the title, as the description has that included, so I submitted just the building name without the college name included to clean it up. ML rejected my title edit, but the description edit I made is in community review.
Sometimes it’s just strange what text edits ML rejects. And then there are the text edits ML accepts that you later realize had an error, and you have to resubmit with the correction.
Niantic uses the term “Machine Learning”. The definition of Machine Learning is that you’re constantly feeding it resolved data, so the machine can learn. If you want more learning, you need to generate more resolved cases.
A challenge would be a great way to get people to generate cases - and review to resolve them - so the machine can continue learning.
or maybe… just a very wild speculation here… Niantic values an accurate and database, and what better than to encourage people to help update that database than to create an edits challenge.
Sorry, my tinfoil hat might be on too tight.