There is a common organization structure found in many large professional firms, including global consultancies and the big industry analyst firms.
For any important topic, we may suppose that the firm employs some of the most knowledgeable people in the world, who are available to give presentations to major clients as well as keynote speeches at international conferences. These high-profile experts attract lots of business to the firm, but of course they don't have time to get heavily involved in the day-to-day work, which is mostly executed by their less famous (and less expensive) colleagues.
It is interesting to consider how this structure affects the overall capabilities of such a firm for handling different kinds of client problem. Where a client problem fits neatly within a single knowledge domain, then we might expect it to be handled reasonably efficiently and effectively, according to the "best practice" principles defined by the centre of excellence for that domain, and with some benefits from the economies of scale and scope. But for large and complex problems spanning multiple knowledge domains, the lack of coordination between the knowledge silos seriously impairs the ability to execute, and the economies of scale are outweighed by the diseconomies of scale.
Ironically, it is precisely these large complex problems where the large global firms claim superiority over their smaller more agile competitors. But this claim is not compatible with an organizational structure based on "best practice". The challenge here is not just building more effective knowledge management platforms (with fancy Enterprise 2.0 software) but architecting the organization to achieve real organizational intelligence.