For such a long piece, this is disappointingly light on any details or discussion about how one would go about building a "Contextualization Engine." I mean, w

For such a long piece, this is disappointingly light on any details or discussion about how one would go about building a "Contextualization Engine." I mean, we all saw what happened with Microsoft's Tay Bot [1] -- any system that attempts to factor in user input is ripe for abuse. Who would control what context is valid, and what context is discarded? If you decide to factor in "all" content, how do you rank it? Is it simply a factor of popularity, or are trusted sources given more weight? Who is a trusted source? I don't mean to sound overly pessimistic: I agree with the author that context is the biggest thing lacking from the internet today, and is a valuable tool for fighting misinformation. But any discussion like this needs to pretty immediately grapple with the intricacies of how to implement context -- and what context really is. [1] https://en.wikipedia.org/wiki/Tay_(bot) https://en.wikipedia.org/wiki/Tay_(bot)