Urban legend has it that when a plane is about to crash, passengers are asked to adopt the brace position not because this will save their lives, but because doing so will protect their teeth and so make it easier to identify the corpses from dental records.
I was reminded of this by the Vickers' report recommendation that retail banking be ring-fenced from investment banking. The purpose of this, as Jonathan says, is not to reduce the risk of failure. It is instead to assist the clean-up operation after failure - to make it 'easier and less costly to resolve banks that get into trouble' because such relatively simple operations are easier to either unwind or sell on than investment banks with their countless counterparties and opaque assets.
Instead, Vickers' plan to reduce - not, note, eliminate - the risk of banking failure lies in the call for higher capital ratios. In themselves, these aren't the problem; the 10% he recommends is actually slightly lower than banks have now (though any losses from the euro area debt crisis and associated economic slowdown will reduce these).
There are, though, two other problems:
1. Vickers calls for a further 7-10% of banks' assets to consist of 'loss absorbing' liabilities such as bail-in bonds, cocos or unsecured debt.
However, given the riskiness of banks' activities, investors will only want to hold these on Buffett-style usurious terms. This will raise banks' costs.
2. Vickers threatens the basic model of investment banking. A lot of its activities are low-margin near-arbitrage trading - hoovering up pennies. To make this profitable, you need lots of hoovers - high leverage. Which Vickers opposes.
For example, last year Barclays Capital made ''4.8bn profits on ''191.3bn of risk-weighted assets (p46 of this pdf), a return of 2.5%. But this was in a very favourable time - near-zero borrowing costs. In 2006-07, return on RWA was just half this (p59 of this pdf). This means that if Barcap must hold 10% of its assets in equity, return on equity would be around 13% in normal times. This is in line with Bob Diamond's target - but it shows how hard it will be to much exceed that.
In these two ways, the Vickers report is a threat to the banks.
But this is not a bug. It's a feature. A major purpose of the report is to remove the implicit subsidy which governments give to banks both by offering to bail them out if they get into trouble, and by giving tax relief on debt. Vickers says:

The risks inevitably associated with banking have to sit somewhere, and it should not be with taxpayers

Although Vickers estimates these subsidies to be more than ''10bn a year, he estimates that his proposals will cost banks less than this: ''4-7bn. In this sense, Philip Aldrick is right. Banks got off lightly.
Which brings us to questions raised, but not answered, by the report:
1. How much would banks be worth if the implicit subsidy did not exist? Last year, the combined profits of Barclays, Lloyds and RBS were just ''10.2bn, suggesting that banks might be almost worthless without that implicit state aid. And this raises the question; if banks were of nugatory value, how on earth could they raise sufficient capital from the private sector to provide the loss absorption which Vickers wants? There is a trade-off between reducing the tax-payers' exposure and increasing banks' loss-absorbers.
2. How much do the Tories really believe in the Thatcherite virtues of self-reliance and standing on your own two feet? If they did, their only quibble with Vickers would be that he doesn't quite go far enough. Which doesn't seem to be the case.