PKC

[Ross Anderson]

> However, it seems that the current human check
> may not be as effective as people think.

There are fatalities. For example, there's been more than one case
where a drug supposed to be given intravenously was injected into
the spine instead. In one of them there was a lawsuit about who was
responsible - the doctor who wrote the prescription, the pharmacist
who filled it or the nurse who administered it? The answer from the
court was `all of them'.

Without several people being jointly responsible, safety is bound
to be eroded; the question is by how much.

It's tempting to think that with e-prescriptions you could do away
with pharmacists altogether. It would certainly be cheaper to have
mail order fulfilment of non-urgent prescriptions than to subsidise
all these little country village pharmacies. There are people who
talk in such terms. But for safety reasons I'd not like to be the
first to implement this. Let the Dutch or the Americans do it first
and wait to see how many more people die. Then decide

Ross

PS: this is only one of many examples of cases where the `high
security' that can in theory be provided by modern cryptography may
end up undermining security because of the Titanic Effect: people
believe too much in the latest technology and assume that just
because multuple watertight compartments can save a ship, they will
actually do so in practice