Saturday 13 June 2009

Crime, punishment and reinventing the wheel

Crime and Punishment, the classic 19th century novel by Fyodor Dostoevsky, describes the execution and aftermath of a brutal double murder committed by the poor ex-student Raskolnikov. I am not spoiling the ending by telling you this - the murder itself takes place early on and the bulk of the novel deals with Raskolnikov's guilt and mental anguish.

Initially, Raskolnikov justifies his crime by imagining himself to be one of the elite few who transcend ordinary morality. Like Napoleon Bonaparte, these extraordinary men are destined to seize society bend it to their will. Their higher purpose excuses them from the constraints of morality that ordinary members of society must abide by.

The reader soon realises that Raskolnikov is not a member of this elite cadre. True Napoleons are too busy invading Spain to construct self-serving psuedo-philosophical justifications. As the novel progresses, Raskolnikov's crippling doubts reveal to him the fallacy of his delusions of grandeur. He realises that men who are preordained to shake civilisation to its very foundations do not agonise over their calling.

In the world of software, it is not at all uncommon to encounter a developer who is convinced that they are a Napoleon. Perhaps it's ignorance. Perhaps it's arrogance. Whatever the reason, they are motivated to create their own inadequate solutions to problems that have already been well and truly solved. Often they take it on themselves to improve upon things that ordinary programmers take as given (like the nature of truth itself).

Google Wave may just be an example of a revolution that we actually need. Email is a tried-and-true technology, but it has its limits and could benefit from a ground-up redesign. The success of Google Maps certainly suggests that the Rasmussen brothers are candidates for web Napoleons.

On the other hand, Google's non-standard implementation of OpenID looks more like it was designed by Rodion Romanovich Raskolnikov. The whole point of OpenID is that it is a universal protocol, yet they have extended it for their own specific needs (they want to be able to use gmail addresses rather than URLs). What's worse, every developer who wishes to accomodate Google OpenIDs on their site will have to contaminate their code with a special case to handle gmail addresses.

If you are contemplating producing your own version of a well-established technology, it is just possible that you possess a unique insight and that by reinventing the wheel you will drag software in a bright new direction. But if you are not sure, then your code is more likely to resemble an opportunistic act of violence than the Code Napoléon.

And even if you are certain that your way is better, you're probably still wrong.

No comments:

Post a Comment