-
braedon Basically: Ad tech companies are responsible for representing Topics into the ecosystem for bidders, so it becomes valuable for them to be able to have Topics as consistently as possible. It becomes very likely that they will use Topics to train ML systems to predict Topics...On twitter.com
Mood +2 🙂
-
braedon The vast middle of the ad tech ecosystem will end up not being real Topics on any given request, but ML systems trained on seeing Topics & a variety of other factors attempting to predict Topics w/some or no work to distinguish these predicted Topics from real ones to buyers...Permalink On twitter.com
Mood -1 🙁
-
braedon The result will be a predictive market in Topics as opposed to a market in Topics where what is being purchased and sold is representations of likely Topics not the actual Topics, all tuned with ML so ad tech middlemen can represent themselves as doing something of high value...Permalink On twitter.com
Mood 0
-
braedon I'm specifically referencing Zuboff's "shadow text" the idea of ad tech systems creating a digital doppelganger of the user that is transacted on instead of the truth of the user's signals, as an attempt to extract greater profits. A very similar thing likely to happen here.Permalink On twitter.com
Mood +5 🙂
-
braedon Hopefully that helps!Permalink On twitter.com
❤️ 1 Favorite
Mood +4 🙂