Offered, with great examples from American history, by John Steele Gordon (HT - Don Boudreaux):
1. Governments are terrible investors
2. Politicians have self interest too
3. Immigration is a good thing
4. Good ideas spread, bad ones don't
5. Markets hate uncertainty
I don't like all the discussion in all of these - particularly the last one which has a good headline but botches the 1920-21/Great Depression comparison in the usual way. But I do like all the headline points.
I think the first one is more complicated than Gordon implies too. And it's kind of funny in light of "good ideas spread, bad ones don't" too, since there are lots of public investments that lots governments make... so doesn't that mean it's a good idea! I would put it this way: governments are bad investors because "politicians have self interest too" and because there is no price signal or profit and loss discipline. The reason why it is nevertheless still good for governments to make certain investments is because of the externality problem. Profit and loss is only as good as the signal provided by the profit and provided by the loss. So when externalities are a big, relying on a government that's generally a bad investor can be worthwhile if you discipline that government and perhaps stick to plans that can leverage the power of profit and loss (i.e. - the government sets out a plan for infrastructure or defense or research and then funds or subsidizes competing private actors to do it). That won't be perfect either, of course - but who was expecting perfection? You certainly wouldn't get perfection from a pure market solution. That's a good idea, and lo and behold - it has spread widely.