It seems to me that the previous reviewer is either a wooly-head theoretician or that the previous reviewer hasn't actually done any research in a laboratory. Because in this book, there are many sparkling insights into the way that science is practised.
It takes a while for Latour to get going as he is quite verbose in the early section, where he discusses his "anthropological" approach to science studies. However, after that, he makes a couple of points that as far as I know, he was the first philosopher of science to make.
First, Latour demonstrates the intimate relationship between the publication of scientific papers, scientific prestige, laboratory finances and actual experiments. He makes the seemingly obvious, though not so when the book came out, that the possibility of experiments in a lab requires the influx of an amazing out of money. The acquisition of this research money takes up a large proportion of the time of the head honcho scientist in a laboratory .
Second, Latour shows that entities in science are always defined by a network of properties that are experimentally determined. Scientific entities are hardly ever seen as objects with a few simple analytical properties. In fact, the more properties the better. And it doesn't matter if the mesh of properties is convoluted and seemingly contradictory. For each property concerned, there must be a vast array of material techniques to measure, control and manipulate that property. A new entity in science is accepted as real only when there are enough inter-locking properties to guarantee its existence. No method, by itself, is ever convincing.
Latour points out that once an object is deemed to be real, scientists often invert the logic and argue that the reason why the combined set of experiments worked in the first place was that the object was in fact real. Whether this inversion of logic stands up to philosophical scrutiny - I do not know - but I have seen many practising scientists make this jump in logic. I've even used it myself. It is here that the "realist" and "anti-realist" debate rages. However, I think Latour reports it just as he sees it.
Third, Latour carries out an analysis of scientific texts, which I have yet to see anywhere else. Scientific statments take on 5 modalities - from speculative hypothesis to proven statements to unspoken assumption. Latour gives a account of how the modalities of each statement are modified by how every other scientist in the field cites the statement in future scientific papers. They can ignore it, attack it as a useless hypothesis, bolster it by citing it as a supporting statement, adulate it by assuming that is a proven statement, and finally they just assume it's true. This scrutiny occurs continuously both inside the lab and in conferences.
However, the difference between this process in the sciences as opposed to the humanities, is that these statements are often associated with machines that act in the material world. Proving a statement means that a material effect is generated.
Using this method, Latour can analyse the fortunes of the scientists in a lab. And analysing the citations of scientific papers results in a reasonably good definition of scientific credibility. As a grad student in a biophysics lab, I've seen this happen - albeit on an intuitive level.
Although Latour has since gone onto to more and more abstract studies, the beauty of Laboratory Life is that it is firmly grounded in the actual practises of an existing laboratory, the Guillemen Lab at the Salks Institute.