Each week since the middle of the primary season I've been running Monte Carlo simulations of the general election using polling data from Votemaster Andrew Tanenbaum's www.electoral-vote.com.
This week's polling data shows has been the most favorable for Barack Obama all year. Without some significant change in the electoral landscape, John McCain simply can't win.

I'm still doing two simulations, of 10,000 trials each, but with a different margin of error. First is a 4% margin of error common to most state polls, a run which reflects what would happen if the election were held now. The other uses a margin of error I computed using a linear regression of 2004 polling data errors compared to the final result, with time to the election as the independent variable. That now uses a 10.5% margin of error.

4% Margin of Error:
Obama wins 100.0%, averages 353.5 EV (low 296, median 354, high 384)
McCain wins 0.0%, averages 184.5 EV (low 154, median 184, high 242)
No electoral ties.

10.5% Margin of Error:
Obama wins 99.94%, averages 340.6 EV (low 247, median 343, high 403)
McCain wins 0.06%, averages 197.4 EV (low 135, median 195, high 291)
No electoral ties.

The 10.5% margin of error does overstate Obama's chances of winning (since each state trial is independent; it's more likely that if McCain gets momentum, he'd gain in many states at the same time, as Obama has over the past two weeks), but not necessarily by that much. Nate Silver now estimates Obama has a better than 90% chance of winning. Obama is clearly in a dominant position.

Almost all of the recently toss-up states have moved strongly towards Obama.
The two latest Colorado polls both gave Obama a 6 point lead. Obama led in all four new Florida polls, with margins between 2 and 7 points.
Likewise Minnesota's five new polls showed Obama with leads, ranging from 1 to 18(!) points.  Obama led by 2 and 7 in the two new Nevada polls, and he led in the three new New Hampshire polls by 9, 13, and 8.
Ohio had 7 new polls, with McCain leading in just one (by one point), and Obama ahead in the other 6, by 3-6 points. Obama's leads in the 5 new Pennsylvania polls were all 10 points or more. Even traditionally red Virginia had two new polls giving Obama double-digit leads (the third still had him ahead by 2).

North Caolina is the lone exception, perhaps still slightly leaning to McCain, with three new polls splitting Obama +1, tie, and McCain +3.

Other previously safe Republican states are now perhaps in play. The two new polls in Missouri both showed a 3 point lead, but differed on who was ahead. The first new poll of West Virginia in over two weeks showed Obama ahead by 8. Three new polls in Georgia showed McCain ahead by 7, 9, and just 3. A new poll of Montana saw McCain's lead shrink to 5.

As the financial crisis has dominated the news, Obama has pulled further ahead. The McCain campaign seems disorganized and unfocused, unable to settle on a consistent message. Its handling of the Ayers association is a good case in point - at times they are trying to paint Obama as unacceptable and perhaps a terrorist sympathizer, but McCain backed away from such direct accusations in the town hall debate. One night a campaign spokesman says it's not an issue, the next day another raises it. McCain did the right thing by calling out some supporters who got out of line with outrageous statements, like calling Obama a terrorist, but much of his campaign is essentially pushing that very idea.

It is a campaign that knows it is losing, but has run out of ideas for yet another tactical diversion, and so it may now be resigning itself to defeat.