Mike the Mad Biologist (my buddy) says that that's where the discussion ends.
The DLCers will claim that they knew all along that Hackett couldn't win, and that supporting him was a waste of time (of course, they never predicted that he would finish so strongly in a Republican stronghold). The grassroots will claim that if the DLC had supported him several months ago, he could have made Schmidt look even worse (and he would have gotten his message out more effectively). The grassroots will also claim that this is a success for the grassroots, mobilization, the internets, etc.
Wrong. We lost.
Well, we did. But we lost 52-48 in a district that elected the previous incumbent (in his first race) 72-28. A district that broke for Bush by 26 points was dead even in a campaign fought largely over the problems with the war in Iraq.
Coingate didn't hurt, as Schmidt and her staff have various ties to the guy who walked away with millions of dollars of the Ohio Worker's Compensation Bureau pensions.
The ethics problems surrounding Tom Delay helped as well.
No matter what, a district that went heavily Republican last November in the state that decided the Presidential election was almost even this time. A quarter of the voters changed their mind about the Democratic party in that district.
Mike is right that there are a lot of close districts, and Democrats need to be out there running and funding good candidates in each of them. But that's not the lesson of Paul Hackett. Mike's lesson is right, Democrats need to target every close race.
The lesson of Paul Hackett is that the right candidate can make any district tight. If you were a party strategist, you'd ignore the Ohio 2nd every time. It has too big a margin, you'd think.
Paul Hackett showed that progressive–even liberal–principles, can sell, if you try it. And that's an important message. Last time, Jim Ryun won his tightest race in a long while, 56-41. That's a race that's winnable by the Hackett standard, but may not be worth looking at from the "target close races" angle.