The point is that due to network effects (which lock users in to a platform), ads are monopoly-priced. But network effects are not super well understood and platform collapses do happen.
So we know the ads market to be large -- or at least people on HN are testifying to that effect, but we don't know that Facebook or MySpace or Lycos get to appropriate it to a long enough period of time that current valuations make sense on a no-bullshit NPV basis.
A fair point was made elsewhere that multiples (which are more or less like stock market folklore, but still allow for comparison across multiple industries) in these companies are relatively modest.
Before I’d run different affiliate offers through them. Though back then for those, Instagram wasn’t that big. Might start again if the opportunity works out.
The returns are good or it would be crazy to put that much money into the ads.
For one: we still don't really really understand network effects. There's huge value to being Facebook (you're forced to be there because all your friends are there and most nowhere else), but people also switch platforms for reasons that are not excellently understood. And when I say bubble I don't mean this or that company -- Silly Valley big money is basically all bet on network effects.
(I.e. Instagram is worth $100B under the premise that I can't build an Instagram killer (that actually kills it) in a short period of time. And jeez, it's not fundamentally different from Geocities + Frontpage, just has better UX)