So I started off thinking about companions with this question: What would I want the ideal one of these to be like? I reckon, it ought to have a number of virtues, certainly including these four:
- It should offer a reliable account of the subject area. For a historical subject, this would mean saying what the evidence is, what is generally made of it, and so on.
- It should give a good idea of why the subject area is interesting.
- It should give a sense of what at present are the major scholarly debates or schools of thought on the given subject.
- It should guide the reader to more specialised discussions, related areas and the like.
But there is something else I think they ought to do. For companions (or whatever) to philosophy, I would think they ought to give an impression of the practice of philosophy. So a companion to metaphysics ought not to be a list of positions or a survey of what conclusions might be reached; instead it would also have to show what it is like to be engaged in philosophical inquiry. It would therefore have to exemplify what it is about.
For my line of work, a companion to some historical period or school of philosophy, a good companion would introduce the reader to the practice of thinking philosophically and to the practice of interpreting and contextualising the particular subject matter. So it would have, for example, to show how to read a bit of Aristotle both by engaging with the argument and also, perhaps even initially, by showing how to get from a bit of Aristotle's writings to a relatively clear view of what it means. Of course, these two practices are not easy (or desirable) to keep apart; that, it seems to me, is what working on historical philosophy is all about: it is both history and philosophy.
I think I am coming to the view that a good companion will have two, perhaps very different, functions. It will (i) lay out the state of play, explain where to go to find various texts, say what they are generally thought to be about and so on; but also (ii) it will function as a protreptic to further deeper work and offer a set of examples of what is involved in working 'unaccompanied' with this material. This second might take the form of more specific or specialised bits of research. The authors can feel liberated from the need to 'cover' an area because their job here is to offer up an example of what the next stage of work would look like, exposing the difficulties and wondering explicitly about methodological questions.
 There are some interesting thoughts in G. R. F. Ferrari's introduction to the new Cambridge Companion to Plato's Republic (2007, Cambridge), xv. Ferrari also stresses the idea of 'accompaniment': 'This Companion, by contrast [sc. with a scout striking a new path], seeks to walk with those who are already on the road...'