1. There are (at least) two Laffer curves - one that relates top tax rates to GDP, and on that relates them to tax revenues. The two probably do not coincide. For example, if a top earner responds to higher taxes by continuing to work but by dodging tax, GDP is unaffected but revenues fall. For this reason, the revenue-maximizing top rate is probably lower than the GDP-maximizing rate. To this extent, Richard is right to stress the importance of tax dodging. As Alan Manning says, a rise in the top rate must be accompanied by efforts to reduce avoidance and evasion. (Film tax relief has always struck me as an invitation to conmen.) Given that the deficit doesn't much matter, I would rather we thought more about the effect on GDP than on revenues.
2. Even if high top tax rate do reduce incentives, this needn't be a bad thing. They disincentivize rent-seeking, office politics, exploitation and negative externalities such as risk pollution as well as productive activities - maybe more so, given that the intrinsic motivations to engage in the latter are greater.
3. We should distinguish between incentive effects upon employees and the self-employed. The former might well be trivial: if someone stops working because of high taxes, his employer will replace him. The latter might not be.
4. There's very little robust UK evidence here, simply because the top tax rate didn't change for years. And it's hard to infer anything from the rise in the rate to 50% in 2010 because of forestalling and because the rate didn't last long enough to have longer-term behavioural effects.
5. The shape of the Laffer curve depends upon tax morale. if you think taxes are theft, you'll try and dodge higher ones or stop working. If you think they're the price to pay for living in a civilized society, you'll accept them. This implies that the revenue-maximizing tax rate will vary from time to time, place to place and people to people. For this reason, even good international (pdf) evidence might not be a good guide to tax-setting in the UK. The first two of Kling's laws apply: sometimes it’s this way, and sometimes it’s that way; and the data are insufficient.
6. Relevant evidence here also comes from laboratory experiments. Some of these suggest that a 50% tax rate maximizes revenue.
8. There are two contrary but tenable positions here. One is "the revenue-maximizing tax rate might be high, but high taxes are undesirable because they infringe freedom." The other is "The revenue-maximizing tax rate could be low, but high taxes are justified to reduce the adverse effects of inequality." Both of these positions are rare - which makes me suspect that there's quite a lot of motivated reasoning on both sides.