The more that algorithms get used, the harder it is to see when manipulation is occuring. This is especially true if there is confirmation bias. If I'm an exe

The more that algorithms get used, the harder it is to see when manipulation is occuring. This is especially true if there is confirmation bias. If I'm an exec that leans libertarian (or left, or right, or whatever) and those are the articles that I see, I assume it's working. Moreso, if I see what's getting posted elsewehre, I just assume the rest of the world is smart like me. You have to really dig in to see that something is going wrong. This involves setting up unbiased metrics that can't be gamed (hard) and listening to data that disagrees with your preconceptions (even harder).