No. of Recommendations: 14
They self planted images of Hitler in their own fake accounts they created in the posts they created.
Then they counted and over reported the number of times the "ads were shown" next to images of "Hitler images".
example was 3 IBM ads next to Hitler image in 5 Billion posts 2 of which were viewed by Media Matters themselves.
How is that biased? And more to the point, how is that fraud or defamation?
If you want to test whether Twitter's algorithms actually prevent major brand's ads next to pictures of Hitler, you need to set up an account that will have pictures of Hitler into it. They're not testing whether typical users will frequently - or even ever - see pictures of Hitler. They're testing whether the Twitter "speech not reach" algorithms will prevent the ads from being put next to Hitler. It's not really relevant whether there were 5 or 5 billion non-Hitler posts on the site at the time; what they're testing is what the software does with the Hitler images.
It doesn't matter how uncommon the occurrence is. To use a ridiculous example, if someone claims that their truck body is bulletproof, you would test that claim by taking a truck and shooting bullets at it, and reporting the results. It doesn't matter that it's insanely unlikely that the average truck being driven by the average user would ever get shot at. It doesn't matter that 99.99999% of the time that someone's driving the truck, it's not being shot at. You're testing the claim about how the product will respond to a specific situation, so you recreate the specific situation that the company has made the claim about to see what happens.
Old Twitter tried to protect brands from having their ads next to Hateful Conduct by removing Hateful Conduct, and the posters who post it, from their site. Musk's Twitter changed that policy, allowing the Hateful Conduct (and the posters) remain on site - but claimed that it had precautions in place to protect the major brands from being exposed to the Hateful Conduct. MM showed that those precautions don't guarantee that the brands will be protected. Musk showed that toxic content can still get massive exposure, because if power users (like him) that choose to consume toxic content choose to retweet it, it can be seen by millions. It's hardly surprising that major brands decided to reconsider whether Twitter is a safe venue for their brand-building.
