Too Big Not to Fail

We all know about "economy of scale," with Wal-Mart as the canonical example.  What about diseconomies of scale? 

The linked article counts off a number of factors that might contribute to decreased efficiency as companies get larger.  To that list I would add:

• Beyond hierarchy and its associated communications challenges, size also brings scaling problems.  Organizations have a natural limit on growth rate beyond which the existing employees will be overwhelmed in their attempt to instill the existing culture into the new recruits.  Many large companies grow faster than this rate and "lose their way."
• Small companies have little to lose and much to gain.  For large companies, this situation is reversed.  As a result, larger companies will adopt more conservative approaches, counting on network effects and bullying in order to maintain their position rather than taking big chances or moving in new directions.  Microsoft is an obvious example here, but so is GM.
• In some industries, a strong temptation might appear to move from a strategy of providing value to one of rent-seeking.  Interaction with the government becomes important here, as size can have enormous advantages when it comes to lobbying, regulatory capture, no-bid contracts, bailout money and the like.  Defense contractors are terribly inefficient at producing the goods and services that they nominally sell, but probably much more efficient when dealing in the market where they actually compete with one another: the pursuit of influence in Congress and the military.

The answer is probably that more and more medium-sized companies are better.  The mom and pop shop is inefficient, but so is the sclerotic GM.  How to get there?