You’ve probably never heard of Knight Capital before the past week, when it suddenly showed up in the news as a financial firm crippled by a systems error.

Knight acts as an intermediary. Your broker — or the broker that handles your pension fund manager’s orders, or your mutual fund manager’s orders — puts those orders into the market. Knight acts to ensure an orderly market, matching buyers and sellers and allowing minor gaps in price to close to keep the market moving, and earning its living from a fraction of a cent per trade.

They put in a systems change for the market open, and suddenly discovered they were buying to the bid for a house account they don’t maintain. Within an hour or two the firm’s capital was exhausted.

Put the blame for this where it belongs.

Most people would jump and say “they should never have put that into production without testing it thoroughly”. That’s quite true, of course, but it’s not the real problem.

The real problem is having designed a process that’s filled with smart technology but dumb people.

The system supposedly handled all use cases automatically. It therefore didn’t signal what was going on (why have a dashboard when there’s no one to intervene?). It was no person’s job to monitor the health of the company.

That’s the way far too many IT solutions work today. No human thought required, just follow the screen and do it the one and only way it can be done. Since the system is fully specified, it doesn’t need the ability to intervene in a thoughtful way.

Wrong, wrong, wrong. Smart systems need smart people making decisions. Not just to save the organization’s bacon when something goes wrong — but because the ability to flex the process, innovate on the spot, create a new use case on the fly (it’s called “customer service”) is essential to success.

IT’s process myopia has created great, sometimes even elegant, systems. But we’ve left the ability to deal with the unexpected out of the picture.

And it’s the unexpected that makes fortunes — or loses them.