This is probably my (and my model's) biggest disagreement with other election forecasters right now. Per our model, the large number (70!) of lean/likely GOP seats—places where Dems have between a 5% and 40% chance of winning—is a huge reason why they're favored to win the House.
About one-third of Dems' projected gains, in fact, come from the "long tail" of seats rated as Lean R, Likely R or Solid R. Without the Dems' having the possibility to win those seats, the House would be more or less a toss-up right now.
Per our model, it is *not* true that Democrats will only win these seats in a landslide. House races are noisy and weird and not as correlated as presidential races. There's almost a 40% chance that the decisive, tipping-point district will come from a Lean/Likely/Solid R seat. 
Anyway, that's what my column was about this morning. Those lean/likely R seats matter a lot. We don't know which ones Dems will win and (unless there's a mega-tsunami) they'll lose most. But they're a big liability for the GOP. 53eig.ht/2yk3DZA
