WillJ: For the increasing returns relationship to exist there is no need for an invisible hand theorem (or any similar theorem) to operate, nor can I see how its existence would make any difference. In this sense I consider the increasing returns conundrum not to be important in an economy. The division of labour does not lead to increasing returns of the kind that offers the spectre of the entire market for pins, say, finally being captured in one factory, any more than increasing returns from planting on a single acre would mean the entire worlds food could be grown there (except in a last person on Earth scenario). Regular predictions of large firms replacing small firms and the demise of small businesses appear in the media, but for over a hundred years, when large firms have got so much larger (with more to come), the demise of the small firm has been put off each decade. In fact, in the UK, for example, small and medium-sized enterprises are still the largest employers of labour, and for one expect them to continue to do so. In state organisations too, specialisation ensures that no one department replaces every other.
Knowledge is a non-rival good, but we also have the phenomenon that no one person can possibly know more than a small percentage of the available knowledge, even if that person knew more about everything that anybody else, and that discrepancy is getting worse and will continue to do so. In Smiths day, professors taught several subjects and could produce valuable work across several disciplines; today that situation is rare even getting disciplines to talk to each other is extremely difficult.
A private economic good is defined as having excess demand at zero price; knowledge is not absolutely free; there are costs in acquiring and accessing it and its delivery can be rival (only so many seats in a library, or hits on an Internet site). Public non-rival goods have costs of provision (those like the air we breathe is an obvious exception), and that immediately runs into scarcity problems and rivalry, whether funded by taxation or prices. Knowledge about how to design and manufacture is non-rival; the provision of such goods from that knowledge is rival. If the civilisation was to collapse catastrophically, even though the accumulated knowledge of the millennia that have passed continues to exist, it may take the small numbers of survivors millennia to benefit from that knowledge and return to anything remotely civilised.
On David Warshs book, I too was impressed when I read it in the summer, though I did not agree with his presentation in several places of the Chicago myths about Smiths works, and I wrote and told him so (and why). He published a version of a paragraph I wrote on his website (
http://www.economicprincipals.com/issues/06.12.24.html) and we have corresponded since, and reached a degree of friendly agreement. As so often happens in economics, people associate their ideas with Smith, even when they would stand on their own without claiming an endorsement from him. I think David Warshs book is well worth reading and discussing but lets leave Smith out of it, at least as he is currently quoted.