It is a well-established dictum that "trust is required for knowledge sharing and re-use". In other words, people will share what they know, and act of the knowledge of others, if they know them, if they have a social relationship with them, and if they trust them.
There is another rule of thumb - that it is difficult to form trusting social relationships in groups of more than 150 (Dunbar's number). This seems to be the average size of tribes, military fighting units, and (according to some sources) web 2.0 circles. Within a group of 150, strong social ties of relationship and trust can form.
However that seems to break down when it comes to large online communities of practice within an organisation. Shell CoPs can exceed 2000 members. The US Army "PlatoonLeader.mil" community covers 12000 Platoon leaders. Other companies have communities which are just as large. Dunbar's number would suggest that these communities far exceed the limits of social cohesion, and therefore should not work as knowledge sharing mechanism. Who would trust 12000 people that they have never met?
Yet these communities DO work.
People DO trust them.
People open up to a surprising extent, and share all sorts of things in these large communities, expecially the military communities. Somehow it has proven possible to generate trust in a social institution, rather than in social ties with individuals.
No I don't know how this works, and it would be really interesting to conduct some research in this area. I speculate that there are a few factors at work here.
Firstly the community leaders and facilitators must engender trust. They must be trustworthy themselves, and must be trusted to ensure that the community of practice remains "a safe place to be vulnerable".
Secondly the community of practice must build a brand and a reputation as being "trustworthy". This is built over years, and built through "delivery of value" to members. The community leader will be instrumental in coordinating brand-building.
Thirdly the users must get a rapid, quality response. If they are brave enough to open up to the community of practice - to expose their ignorance and ask for advice and help, for example - then the advice must be quick, and the help must be helpful.
There may be other factors at work, but these must be within the top 5.