I have a theory about why the AI isn't able to identify contact details when you report them.
When you set up a listing, the system automatically identifies certain violations - eg e-mail addresses, websites, postcodes, phone numbers. If you type in something that it doesn't like, it tells you and makes you change it. Obviously, it is programmed with certain rules to do this. And it doesn't take long for a determined rule-breaker to work out what the rules are, and how to show their contact details in ways the system doesn't recognise but actual people will spot.
I think the AI that "deals with" reports of contact details, has been set up with two major flaws:
- It works on the same rules to identify contact details, as the listing system does. So anything that got past the listing hurdle, will get past the report hurdle. So it can NEVER identify an active listing as breaching these rules.
- Because it's AI, it "learns from its own success" - so it soon decides that ALL reports about contact details are false alarms - or worse, false reports - and turns its attention to punishing the people who made the report in the first place, for "abusing the report system to make false reports."
That's why we still see the old favourites - "disguised" email addresses, photographs of business cards, disguised phone numbers using letters instead of numbers, and all the other ways that some sellers know will never get picked up. Because anything that made it on to the listing, will necessarily pass the report check too.
It's basically like using the same spellchecker twice to "double-check" a document. Any flaws in the spellchecker (like it being set to American spellings instead of English ones) will be replicated.
*****************
Cesario, the Count's gentleman
