Should you believe the polls?

Washington Monument Source: E. McBride

The polls can be misleading at times and dead-on accurate at others. This week there are conflicting polls in several major publications.

A Bloomberg News poll shows Cain at 20 percent, Paul at 19 percent, Romney at 18 percent and Gingrich at 17 percent among the likely attendees with the caucuses that start the nominating contests seven weeks away. This information was cited in an article entitled "Romney Two-Way Race Now Republican Four-Way Race in Iowa." According to the poll, participants were 2,677 registered Republican and independent voters in
Iowa ages 18 or older, of which 503 said they would definitely or probably participate in the January 2012
Republican caucus. It does not cite the intentions of the other 81% of respondents.
Meanwhile, Salon.com published a story on the same day saying that Romney is the "inevitable nominee they all can't stand."  And cites a Wall Street Journal/NBC poll where Romney has 32% of the votes.Salon.com focuses on comparing Romney to Bob Dole in 1996, GW Bush in 2000, and McCain in 2008, who all maintained significant leads throughout the campaign trail and primaries. It also mentions the significant differences in this race as compared to years previous. In the past the contenders had pointedly different campaign platforms, and were not appealing to the same base. (For instance, Pat Buchanan's extremist points of view as compared to Bob Dole's moderate appeals. Or Ron Paul's base versus the McCain base in 2008.) This race is different as the three leading candidates- Romney, Cain, and Gingrich, are pulling in close to 70% of the base combined, and are not significantly different in policies (as compared to the considerably more conservative Santorum and Bachmann who only have 3% of the base combined). In other words, if just one of the big 3 were to surge forward in one of the first primaries it could completely change the rest of the race. But that's just Salon.com's point of view. Don't forget Bloomberg has Ron Paul at 19% in Iowa!
At the same time Real Clear Politics has a story citing a Public Policy Poll  putting Gingrich into first place with 28%, Cain in second place with 25%, and Romney "sliding" into third with 18% .PPP surveyed 576 Republican primary voters from November 10th to 13th. The margin of error for the survey is +/-4.1%.
In an almost humorous twist, Real Clear Politics is running the following poll in a widget next to the article-

(click on image to enlarge)

What does it say? In a Bloomberg poll (the same Bloomberg who above has Romney at 18% in Iowa), Romney is leading at 40% in New Hampshire.

So can you believe polls? Or accept them with a grain of salt?