The atlas is likely to appeal primarily as a coffee table book - and in that sense it probably works. It looks sufficiently pretty, and from an aesthetic point of view it is certainly deserving of more than just two stars. The basic idea - a metropolitan atlas certainly appeals as well, and the basic information that the author aims at presenting covers some of the more interesting aspects.
Where the atlas fails completely, in my opinion, is in the care and attention lavished on the actual data contained, as well as in the tools used for the graphical presentation of the data.
The author admits in the short introduction that only a select few sources were consulted and that little effort has been made to ensure the data on the different metropolitan areas (partially coming from different sources) is comparable. This will result in the book claiming that the city with the highest amount of crime is Stockholm (followed by other criminal hotspots such as Frankfurt, Berlin and Copenhagen (rememeber, this is supposed to be violent crime here)). You will have cases, where even percentages are not calculated correctly (very often apparent at a glance already), where the growth rates implied are completely unrealistic (a city growing from 400k to say 3m inhabitants over 5 years), etc. To make matters worse, no attempt has been made to present the complete picture for all metropolitan areas - if the handful of basic sources did not give the info, the author certainly did not seem to bother digging further. This results in some metropolitan areas being represented with little more than population numbers (the empty page does look funny in such cases).
Adding insult to injury, most of the data is presented in the form of pie and bubble charts. From a design focused book I would certainly hope for a better graphical way of representing data - as it is, the format makes glancing comparisons pretty difficult - at the end one needs to actually look at the number to really get a feel (negating much of the advantage of representing the data graphically in the first place).
Overall, the effort looks rushed, with little editing (the 2007 edition is no better), and not enough work put into actually making the book useable as a source of information. If you are a collector of atlases, or if you just want to leave a good looking book lying around, by all means go for it - if you care about the quality and presentation of the content, it is best to avoid this one.