I wonder what the accuracy rate of identifying stolen items on ebay would be (not just bikes) when passed in front of a handful of subject experts. I reckon I could get a pretty accurate hit rate with 30 seconds of what I could see on a listing (pics, words, seller history) and I don’t have the account analysis insight that ebay/gumtree etc have on the setup and use of the account. Of course I’d struggle with power tools, electronics and other commonly stolen items but it wouldn’t be difficult for an organisation the size and scale of ebay to turn the problem over to machine learning and then have humans doing some of the finer-detail validation and questioning.
Agree 100% it’s all just business as far as the marketplaces care but that just isn’t good enough.