Re: Has anyone else noticed that these reccomendation algos are basically cancer?
The talk of these confirmation algorithms is interesting, FTA: "Previously the Silicon Valley biz has been criticized for sending users down a conspiracy theory wormhole where watching one video about, say, the Earth being flat, leads to another and another"
Surely this is no different to the "You've just bought a toaster, hey look at these adverts for toasters!" advertising algorithms that are oft criticised in these esteemed forums. We're told these adverts are much more useful* to us as a result of the creepy tracking but in reality, no. This really is all they've got - there is no AI, the Machine Learning touted as "AI" is still really fucking dense and can't do anything other than feed an echo chamber.
So for all that data they slurp and activity they track (emails, messages, chats, searches, friends networks, page views around the web) all they've got is something that can do simple reactive regex matching. I can't see the promised automated moderation being any better so the usual perpetrators will be able to circumvent it with the usual tricks like spaces and substituting letters for numbers as they do on the Daily Heil message boards.
* = for no values of "useful"