The problem is with first moving funds to and from mtgox - this takes time and there is much risk that the prices will change, second with exchanging USD and GBP - for small amounts this is rather costly. Statistically this should work - but there is not enough trade in local markets to justify the hassle. But yeah - I do that sometimes, does not pay for the work - but it's like a game.
What about when one adds in this assumption: GBP/USD is stable.
(I accept it is not a guaranteed assumption, but it's pretty good, it varies only slowly, especially compared to BTC at present).
With that in mind. All one would need to do is supply some "float" capital, and push money from GBP to USD via BTC; and then take as long as you want pulling it back from USD to GBP.
As others have said, the market depth is not sufficient that one could just run round and round the loop constantly gaining 16% each time - the second time around, the same prices are not available.
I don't think there is a hugely profitable business there, but there is certainly scope for balancing the currencies correctly, which is beneficial for Bitcoin.
Anyway, I'm pleased to find that no one seems to think my maths is wrong. :-)