Not that I disagree that they shouldn't be doing a better job with the dickbar thing, but analyzing 12TB of data per day and cross referencing it with petabytes of earlier data does actually seem very challenging. I can't think of too many companies currently doing this, so it's not like you can just google "how to data mine petabytes" to get a step by step tutorial.
You are very right. But, why not start simple and just lump people in broad categories based on follows.
Jessica follows a lot of celebrities. Bryan follows lots of tech icons. Billy follows lots of comedians. Etc. Etc.
Even that isn't easy but it is hundreds of times more likely to work than blanket ads and last I heard they have millions in VC money. Spend it on talent specifically for this.
