A user boss is any person with command and control authority over end users. They dispense infrastructure owners trust and hire and fire but do not necessarily exercise direct control or supervision of intra-user exchanges.
 often pointy
In some countries, most user bosses are incompetent. In Canada, for instance, most talented people move away to work internationally, leaving no-talents in charge at home. Those that remain in Canada typically do so for political reasons and are therefore not as likely to work in the private sector. User bosses often know this, and adopt tactics to discourage innovation, initiative, independent thinking and creativity among employees, so as not to be shown up, or fall out of control. They may maintain control via PowerPoint presentations that, by definition, the employees cannot edit or refute but must sit through. Or, they may simply rub in the fact that they are the boss, and their employees are not, and that this situation will continue regardless of any incompetence, because an organization composed of pointy persons will rarely fire anyone other than for ("rudely") pointing out that everyone in the organization is in fact incompetent. See trolls.
Parkinson's Law applies to all pointy types.
 less pointy
In less pointy organizations, prediction markets are used to limit the reliance on a centralized boss perception. Google shops do this. The duty of any boss is to recognize reactive behaviour and make it more reflective, including their own.
 control of intranets
The major decision a user boss makes with respect to a corporate wiki is whether to buy or build its core content and core ontology. In other database applications, notably in banking, it was common to purchase a strong ontology for use in mainframe applications. In the early 1990s one was marketed by an Australian bank to banks worldwide, for a price about US$1.5M per customer.
Today the expense of building a robust core ontology has come down about tenfold - but it's still unreasonable to expect to make all the basic distinctions required to run a large privacy-sensitive and scenario-sensitive business for less than about US$100,000 for a large organization. Small ones might pay half of that for a single qualified chief editor who knows the application area very well - plus the technology required to support them.
 least pointy
An inhouse team can probably not produce a better information architecture than this wiki. In particular, they are likely to abuse categories to represent non-operational distinctions, or popular ontological metaphors that do not match the problem domain. The costs of this can be quite extreme compared to doing it right in the first place.