ways to think about feeds

Toricelli highlights some difficulties in turning a netowrk evolution as a feed. There are cool principled ways to approach this, as well as nice naive approaches.

• recently updated is surpsiningly useful on its own
• Quieuing theory probably tells us how to maximise liveness, surprisal, or other more abstract properties
• the diff between toricelli and recents probably tells me about how predictive toricelli is, how tasty it is
  • cool strange loop problem