MTAS and other NHS websites

[Ian G Batten]

On 5 May 2007, at 13:41, PeteM wrote:

>
> In last night's Ch4 item, they said a doctor only had to change two  
> digits of the URL that pointed to his own data, to get a valid URL  
> for data referring to another doctor. (I think it was two.)


You can see how the conversation went, can't you?  A developer  
pointed out that checking credentials on every transaction will slow  
things down, both in development and runtime terms.  So they opted to  
just check credentials at the outset, and then bury a tracking number  
in the URL.  That's bad: anyone who gets hold of the URL from a proxy  
log or the footer of a printout is in free.  But the conversation  
never got on to the problem of the search space.  If every number is  
a valid number, they're screwed.  Had they instead hashed them into  
128 bit quantities, so that only a tiny fraction of possible numbers  
were actually valid, they'd have got away with it.

[[ Ever stayed in a Formula 1 hotel?  They use six-digit pins for  
access to rooms.  Why?  Because it's also a pin for the front door,  
and with 100 rooms it's only equivalent to four digits. ]]

ian