This is how I interpreted Wenger:

Each person belongs to one or more communities. Identity is the interplay between the individual and the community. Identity is dynamic, will vary according to the community, and can also be seen as a buffer or membrane between the individual and the community.

This short introduction gives an overview of the second part of the book that focuses on identity. The takeaway is that we all belong to a community but with unique identities. Beware! As usual, Wenger uses fuzzy terms and discusses what they are not rather than what they are.

Key points:
-    The focus is narrowed onto the individual, but from a social perspective. Identity is the social, the cultural, and the historical with a human face. The focus is also expanded beyond CoP, calling attention to the broader social and identification processes.
-    The unit of analysis of identity is at somewhere between the individual and the community. Wait, the focus is now on their mutual constitution! Identity is not narrowly individual, nor it is abstractly collective.

Assumptions to avoid:
-    There is an inherent conflict between the individual and the collective.
-    (Referring to the individual and the collective) One is good and the other bad; one a source of problems and the other is a source of solutions.

Structure of Part II:

Chapter 6: Identity in practice: Establishes a parallel between practice and identity.
Ch. 7: Identities of participation and non-participation: Introduces non-participation as central to identity formation.
Ch. 8: Modes of belonging: Extends the notion of belonging beyond local communities of practice.
Ch. 9: Identification and negotiability: Discusses issues of belonging in terms of social identification and negotiability - variably owning meaning that defines communities.
Coda II: Learning communities: Summarizes Part II describing the basics of a learning community

Wegner points out that individuals and the community are not dichotomous, nor do I think that they mutually constitute one another as Wegner claaims. Instead, I think that they represent two different languages used to describe observations about the world. I think that the presumed struggle between these ultimately two "levels" ultimately comes from disagreements about the appropriate level of analysis to answer a question being reduced to disagreements about the value of each level of analysis (the types of question posed are mistaken for the types of answer given).

For some questions, it seems more important to us to ask questions about the communities role in constructing a certain situation (e.g. education). For other questions it seems more important to ask questions about the individual's role in constructing a situation (e.g. criminal culpability). In both cases the other level of analysis is typically only appealed to as a mitigating factor.

The two different levels of analysis are not really in conflict with one another, but occasionally they describe forces that are in conflict with one another. I am willing to bet that these types of conflict are easier to remember than situations in which the two kinds of description seem to operate in harmony. This is probably because there is not need to appeal to both levels of analysis when they are in harmony with one another (because either level is "good enough" on its own).

It would be conceptually difficult to manage thinking about a problem on more than one level at the same time (imagine trying to consciously model the physics and sociology of handshakes at the same time) but it is a reasonable idea to approach problems through by alternating between both lenses and trying to gain insight into how one might affect the other.

I agree with Jesse's point about the existance of social-individual dichotomy in the 'analysis' as a necessary evil. I once talked with Mimi Ito, an anthropologist (e.g., study camera phone users in Japan), about Wenger's book, and she commented that in the education systems, they learn (practice) collectively, but gets evaluated individually. There's a mismatch between the ideal educational practice and the outcome of the evaluation, still going back to the individual basis of analysis at the end of the day.

However, I guess Jesse's point about supposing physics together with sociology is fundamentally against Wenger's supposed readers of his book.

There are words for overestimating the power of little things (e.g., greedy reductionists), but not for those who are overestimating the power of collectiveness. If physicists analyze society, they may become greedy reductionists. But if sociologists study physics (which they would never do), they just become holists. Wenger just says that let's be holistic, but nothing more.

Mutually constitutable argument
You have to be able to reference the idea of community if you wish to talk about identity issues for individuals.

Likewise, if you're going to talk about community identity, you need to talk about the individual identities that compose those communities.

Identity does not exist just through individual declaration -- there must be a community that recognizes that identity (see Yong-Mi's diagram). In interactions with communities, only parts of your identity is revealed to each community