This post is inspired by a couple of others that I'd suggest you check out. First, I'd recommend TPM's Eric Kleefield's post on Obama's electoral map and his possible paths to the White House. Then I'd check out two posts over at Ben Smith's blog at politico.com:

1. First, this post shows us that Survey USA and other pollsters continue to have a hard time polling African-American voters accurately. The trend in the primary polling was a general underestimating of Obama's support with African-American voters. Expect that trend to continue unless the pollsters change something or until the general election is closer.
 
2. This other post at Smith's blog shows the electoral map predicted in May 2004 based on the polling back then. You can look at this glass as half full or half empty for Obama, but really the only thing you should take away is that polls in May aren't very accurate predictions for the November election.  

How accurate it the polling happening right now? I think there's some compelling evidence to say, not very accurate. Let's focus in on that first post from Ben Smith's blog. Survey USA was one of the more accurate pollsters this year, although they all had hits and misses.  SUSA says that McCain is beating Obama by 4 points in Michigan (and of course Hillary supporters are going to pounce on that as evidence that he is a weak candidate). 

But let's look at SUSA's numbers:
- Obama is winning African American vote 62% - 26%.
- Obama is winning Wayne County 44%- 42%. (Wayne County is home to Detroit and about 1 million of Michigan's voters.

Now here are two things that are going to happen in November that are related and important: Barack Obama will win over 90% of the votes from African Americans (given that white Democrats have been getting in the 85 - 90% range of those votes) and ANY Democrat will win Wayne County, Michigan by more than 2:1. In 2004, Kerry won the county with nearly 70% of the vote.

So my word of caution is to avoid reading too much into any poll released for any election to be held in months rather than weeks. Even from respected pollsters like SUSA. They are just relaying the data as it comes to them - but when we dig a little deeper I think we can see how flawed opinion polls are this far out. Generally, I think the polls are pretty close a week or two from an election (minus New Hampshire) but months before an election they are going to be way off. Hell, remember how just a couple of months before Iowa and New Hampshire the polls stated that Rudy Guiliani and Hillary Clinton were sure to be the nominees?

But please do make plenty of guesses and predictions about what will happen in November - that's the fun part.