> For me or anyone to believe this, is a huge ask. Agreed. > Any attempt to algorithmically shape perspectives is crude with today’s state of the art. I susp

> For me or anyone to believe this, is a huge ask. Agreed. > Any attempt to algorithmically shape perspectives is crude with today’s state of the art. I suspect that the truth of this statement is very difficult to ascertain at the moment. Reasonable people might dispute both which technologies represent the "state of the art" and also which outcomes represent successfully shaped perspectives. Although I don't have the data to back up this assertion (and again, I'm not sure such data is cognizable in the current environment), I think that for some reasonable definitions of the above that indeed perspectives can successfully be algorithmically shaped today, though as I said above, I think that this "white couple" nonsense is not a meaningful example. > How precisely would such an algorithm work, while not creating obvious artifacts and errors throughout the search results for all people using the service. Again, it may not be clear to anyone, including the authors, precisely how such an algorithm works. That's why A/B (and other) testing is needed. And I think people in this thread are saying that indeed obvious artifacts are starting to creep up (although again I think that the "white couple" thing is a total misfire and not at all an example - the "gun rights" thing appears to my eyes to be much closer). > And how would it work for people only in the us and not elsewhere? Isn't this already at a point of triviality? Doesn't Google already do this all the time? I mean, if you travel outside North America, indeed the Google experience is palpably different from the one in the US.