[Lightly edited from Bill Gasarch's original.]

Reviews can be 3-6 pages and usually have:

a) An introduction to the area so that non-specialists
can understand what the book is about.  This is less important if
it's a textbook or a book on stuff everyone knows. 

b) A summary of the book that tells what's in the book.  Hypothetically
anyone who understood the intro should be able to follow this.

c) An opinion. This could say clearly WHO would benefit from the book.
These are usually not negative, but they can be. If they are, they should
be explained in measured terms, thoughtfully, in fairness and with
all due respect to the authors.

There are a few issues:

i) What level is it supposed to be on?
(e.g., you can't blame an advanced monograph for not
defining some basic terms)

ii) Is it well written?

iii) Is it technically correct?

Especially for (ii) and (iii), explain your answers. Why is
it well-written (or not)? If there are technical problems, what
are they? Examples in both cases are fine, but also if there
are systematic problems (or virtues for that matter), what 
exactly are they?

EXAMPLES:

	There can be a well-written book on Davenport-Schinzel
	sequences (a very specialized topic) and the reviewer
	can say it's a well-written book but still caution that if
	you don't care about Davenport-Schinzel sequences, you
	won't like the book, or won't find it helpful.

	There can be a well-written book on Crypto that is just
	plain WRONG on some important points (e.g., RSA is equivalent
	to factoring) whose audience is undergrads. Mention all these
	points: well-written BUT flawed. You can interject your own
	opinion about whether the flaws outweigh the value.

	There can be a badly written book on a topic near and dear to
	your heart.  Even so, you must tell the reader it's badly written.

d) Reviews must be in latex. I will email you a template for
format. 

e) Bill's website has (almost all!) past columns, so
you can see what
kinds of reviews are appropriate:
   https://www.cs.umd.edu/users/gasarch/bookrev/bookrev.html
Columns edited by me are here:
   http://cs.clarku.edu/~fgreen/SIGACTReviews/bookrev/bookrev.html
You may note that NOT all reviews fit the
proposed format above (sections a, b, c). That's fine; if some other format
fits your review better, you are welcome to add and/or delete sections or
change section names, but you should convey
the INFORMATION mentioned above. 

f) You get to KEEP the book. WOW!

g) I want to stress again that the non-specialist should be able
to read your introduction and understand something about the area.
You may use the intro as a way to sell the area if you want
(see next point)

h) You may want to use your review to make some POINT of more
general interest. (E.g., "I will be reviewing a book that is
mostly pre-RSA crypto and will make the point that pre-RSA crypto
is worth studying").

i) You really need to read/skim the entire book. To just
paraphrase things from the introduction is not good.
Books of this nature are hard to actually read and finish
but you really should KNOW what's in them cover to cover,
perhaps glossing over (most) details.
					Fred, and "bill g."