The wisdom of the crowd (not the 2004 book The Wisdom of Crowds) is a perceived phenomenon whereby groups of people can make better informed decisions than individuals alone.
Most of the evidence for such phenomena are entirely anecdotal. The original observation was supposedly seen at a church or village fair (or fête or wherever) in the 19th century, where the crowd were asked to guess the weight of a pig (or the number of beans in a jar, you know the sort of game). Of course, estimates vary wildly for this sort of thing, from sensible answers to outright embarrassing answers. The majority of these answers would, of course, be completely wrong, and it is possible that not a single one could be right. However, one observer noted that if you took all the answers, and calculated the arithmetic mean of them, not only was the answer very close, but it was exactly right. Hence the crowd, collectively, could calculate better than an individual.
 Is it real?
Some evidence suggests that this sort of works, but only in certain situations. Namely, situations where the knowledge required is very well defined. This can be seen in the game show Who Wants to be a Millionaire where a player can ask the audience for an answer out of 4 possible ones. Here, the audience can be split into two groups; those that know the answer, and those who don't. Those who don't know the answer will, in theory, select randomly and on average, these answers will be uniformly distributed amongst the possible answers. This leaves the people who know the answer, who will choose the correct one and bring that answer above average. With the exception of circumstances where the answer may be part of a widely believed myth or the question is ambiguous, the situation works out well on average and the wisdom of the crowd holds.
The whole idea that groups of people are better at decision making than the individual is the cornerstone of voting and governmental systems such as democracy. It is also the key component of Wikipedia, where the contributions of many editors over time will produce a reliable and good quality article and of news aggregation sites (RationalWiki's WIGO system being one example) where articles are voted up/down based on group consensus.
The idea of "the wisdom of the crowd" does not hold where the the knowledge is poorly defined. An exact outcome, such as the "number of beans in a jar" discussed above, is very well defined, but the results of the lottery (which Derren Brown cheekily tried to claim he predicted using the "wisdom of the crowd" method) isn't, and thus isn't subject to the method. Similarly, the Who Wants to be a Millionaire scenario works because the questions are on general knowledge, there are limited options and enough of the population will be expected to know the right answer. If the answer was more open-ended, subjective or far fewer individuals would be expected to know the right answer - so few that their contribution would be statistically significant - then the system would fail. The whole wisdom of crowds, therefore, falls short when the "crowd" in question is subject to groupthink or other biases, in including a lack of specialist knowledge.
The lack of specialist knowledge in a crowd is one of the main factors working against "wisdom of crowds" having a universal application to open ended or specialist questions. The majority population at large would not have the knowledge of atmospheric chemistry, climate models, hydrosphere dynamics or many other areas of knowledge to effectively evaluate the effects of climate change on the planet. Indeed, when chess master Gary Kasparov played against "the world", he won. Given that the only way to effectively implement the system in most circumstances is by vote majority consensus, what the majority may believe may not necessarily be right. The high prevalence of some urban legends and conspiracy theories is certainly evidence of this.