So I'm now averaging the daily tracking polls to introduce some precision into my brooding. Since Saturday--which is to say, since Friday night's surveys--the average of the eight polls Nate lists in his tracking poll primer has fallen from 7.8 points to 5.5 points. The erosion has been pretty steady from day to day, though it appears to decrease somewhat as the days go by, and it almost disappeared between yesterday and today (yesterday's average was 5.7). For what it's worth, the trend is more or less the same if you exclude Zogby, though the deceleration is less pronounced.
It's the stability between yesterday and today that prevents this from being really lousy news. While it's true that, at this rate (an average loss of .575 points per day), Obama would be down to roughly a two-point national lead by next Tuesday, any temporary stabilizing, even if the downward trend were to resume, should keep him in the three-to-four point range at worst. I guess the question is: Is whatever happened between Friday and Monday night just a blip, or does it reflect the race's underlying dynamic as we head into the homestretch? I'm honestly not sure, though I'd guess Obama's prime-time TV appearance tonight will buy him a little insurance either way.
One other optimistic note (I promise it'll be the last for a while): For the last week or two, I've been treating the possibility that state and national trends might diverage as a cause for concern. That is, I've assumed that even states where Obama had consistently polled above the national numbers (liked Pennsylvania) could begin to tighten more quickly than the national numbers, as McCain focuses on them almost exclusively and ignores the rest of the country.
In fact, I now think the divergence may occur in the other direction. That is, McCain's socialist/redistributionist attack may be getting him a lot of media coverage nationally, which is tightening the national numbers. But Obama's standing in the key battleground states is more robust, because he's got them so well wired with paid staff and volunteers and is flooding them non-stop with ads. If that's the case--and the state numbers seem to support it at this point--then McCain really can't win. Stay tuned.
Update: Some of my office-mates are more pessimistic today than I am because polling indexes like Pollster.com's have abruptly narrowed. (Pollster.com's national number is down to 5.8). My normal impulse in these situations would be to jump from methodology to methodology depending on which is the most pessimistic. But I'm going to be the bigger person and stick with the tracking poll approach I've laid out.
As I've said before, Pollster.com-type averages are somewhat backward-looking by design. My sense from the tracking polls is that the real deterioration for Obama happened between Friday and Monday. I suspect they're just picking that up now.
Second update: Nate makes a good point about how two of the trackers--Battleground and Investor's Business Daily--are pretty backward-looking because they have five-day time horizons (and Battleground doesn't even poll on weekends). I've redone my average without those two polls and the trend is the same, except even more pronounced. Between Saturday and today, Obama's lead has shrunk from 9.1 to 6.3, but the erosion gets smaller each day, to the point that it basically disappeared between yesterday and today. I think the worst may be behind us, though it's hard to know without understanding the cause.
--Noam Scheiber