I'm not sure yet.
I quite like Lomborg's suggestion of $2/ton on CO2 to be spent entirely on R&D of new energy technologies, with patents available to all. $25bn/yr would be a ten-fold increase in investment and it _should_ lead to breakthroughs that will bring renewable techs to market quickly.
Right now I think we need to stabilise somewhere between 500-550ppm CO2 (compared with politicians who're talking about 450ppm), and my favoured solution is either a carbon tax, with money raised split between countries by population and pollution levels (eg Americans start with much more money per person, but slowly it contracts and converges). This is simpler than cap & trade and countries can do what they want with the money.
Alternatively, a simpler 'cap and trade' system would sell all permits at auction and companies would have to buy permits to dig up fossil fuels or chop down trees. Permits would be generated by planting trees or burying CO2. Only real carbon fluxes are considered, and there are fewer companies (so less bureaucracy) dealing with this. Easier than regulating every smokestack and car exhaust... Again, money would be given to countries to spend using 'contraction and convergence'.
In the case of my country, I'd hope a portion would go to renewable R&D, and the rest would be split evenly between the population. This way it's 'progressive'.
The costs would be the 'deadweight loss':
Which would be significantly lower than the headline figure.