Last month, Laura Bush turned up to a party at the White House in a $8,500 Oscar de la Renta dress, which three other women also happened to be wearing. Laura Bush quickly changed her dress, but the damage was done and the story of this amusing accident quickly spread across the Internet.
I was thinking about this while dawdling in a shopping mall last week, and wondered how this sort of thing might be prevented in future. Perhaps some sort of centralised dress list might work? If all attendees to a formal function submitted their dresses to the organisers, any embarrassing collisions could be quickly detected and avoided. Of course, the obvious problem with this (from a security standpoint) is that the list is rather valuable and if it was leaked, there would be hell to pay. Even if it wasn’t leaked, there’d still be the opportunity for the organisers to get up to all sorts of mischief by manipulating the list, for example, to fake collisions in order to prevent people from wearing particular dresses.
So, how about hashing the dresses? By this, I mean taking someone’s dress and applying a hash function to the name (e.g. ‘oscardelarenta2006’). This function would then spit out a unique hash sum (e.g. ‘4FA043BD’) which could then be stored in a database. If anyone else submitted an identical dress name, it would produce the same hash sum and the collision could be detected. The interesting thing about hashing functions (good ones, at least) is that they are one-way; in other words, if you are an unscrupulous White House party organiser with access to the database who wants to sell the dress list to the paparazzi, you only have the hash sums – and you can’t get the dress names from the hash sums.
This is not really a novel idea – ‘friend-of-a-friend’ applications (FOAF) do the same job, wherein people take their email contact list, hash it, and then compare that hashed list against other people’s hashed lists. If there’s a collision, then you have a friend in common, leading to all sorts of possibilities such as friendship/business partnerships/hooking up – but importantly, you can’t figure out someone’s friends from their hashed contact list.
Given that fashion designers already maintain their own private lists of who are wearing their dresses to which parties to avoid collisions and drum up publicity (and presumably someone was fired by Oscar de la Renta), I doubt that a hashed dress list repository will ever come about. But it’s an interesting exercise.