?

Log in

No account? Create an account

May 5th, 2009 - Balinares — LiveJournal

May. 5th, 2009

10:01 pm - The Science of Cataclysms.

In non-baby-related news, Slashdot relayed this article, which at first glance you might be tempted to overlook as yet another boring viewpoint on Finance and The Crisis and Stuff.

Don't.

This is a personal account by the man who found himself writing the software which the financial industry has been using to automatically consolidate debts into bonds. First, prime mortgages; then subprimes; then subprime-backed bonds and then the bonds backed on those bonds.

It's like watching a frame-by-frame replay of the world's most interesting train wreck.

I am having a hard time describing my feelings after reading this. There is a pure, cold and perhaps a little morbid mathematical beauty to this, that mortiferous and perfectly logical chain of causes to consequences -- all happening beyond the understanding of anyone involved, because the entire purpose of the system was to encapsulate entire worlds of arithmetical complexity to the point where the sole human-facing interface could be narrowed into a commoditized dashboard sporting but a few variables.

A few months ago, I was sent to attend a conference organized by Oracle on the topic of Business Intelligence -- the core business of the group that owns the company I work for -- which roughly amounts to spectacularly expensive software that purports to model your entire business, and lets different people supervise and control what aspects thereof fall under their responsibility on the basis of convenient, bite-sized atoms of data. It left me with a similar feeling, something uncomfortable, precariously balanced between the fascination for the mathematical beauty of the engineering involved, and sheer terror at the stark disconnect between those people's now fulfilled craving for artificial understandability and the organic complexities of the real world their actual businesses -- and we're talking Fortune 500 here -- exist in.

I asked someone a question to that effect. "How does this system account for mismatches between the models and reality?"

There was this awkward shifting -- the embarrassment of the high-born as a dirt-soled peasant asks the council of princes about beet and pigs -- and a reply which, as I remember it, went something like: "Well, that's not really the purpose of BI, is it?"

Probably not, I guess.

Perhaps all understanding is an illusion, in the end, leaving us, ironically, with the cold beauty of mathematical logic as our lone solace. Bitchin'.

Previous day (Calendar) Next day