Kan Kotecha spent 18 years at
What could be better?
Google is all about innovation, in technology. Pure and simple. All the time. Anywhere.
It’s now July 2010 and Kotecha is back. At Morgan Stanley. Where it’s more important to build systems that stand the test of time and every possible manipulation and misconstruction of users – or hacking by the most macabre of miscreants. All the time and anywhere. “Tried and true” systems are what counts, as one financial services programmer puts it.
“Tried and true.” Rings rather dull, compared to “innovation.’’ Google’s brought us amazingly fast search through incredible amounts of unstructured data. And a search window that also does currency conversion, movie lookups and more. It’s brought us satellite views of what seems to be every inch of the world. Straight-ahead and 360-degree street views, too. Etc.
But, like Kotecha, I’ve learned to love, respect and wish for “tried and true,” in the space of one week.
MONDAY, JULY 19: I close the clamshell on my MacBook Air. The right hinge buckles. Has to go to the shop.
WEDNESDAY, JULY 21: Get home. My 24-inch desktop iMac won’t read its hard drive. Has to go to the shop.
SATURDAY, JULY 24: Travel to Maine for a bike ride. Do the ride. Try to call home afterward. Keypad on the cell phone is dead. I can only take calls.
SUNDAY, JULY 25: I communicate with my wife on the drive home by pulling over, sending out text messages on my iPod Touch.
MONDAY, JULY 26, in Connecticut: Load the Wall Street Journal onto my Touch. Drive to train station. Take out the Touch. It’s dead.
MONDAY, JULY 26, at the office: Ask for a laptop computer to replace my desktop computer. Have to keep working on train. Get one.
WEDNESDAY, JULY 28: Video board on my “new” Dell Latitude D630 fails.
By then I am sure I’ve diagnosed the problem: Telekinetic user error. I am a walking hot spot. Whatever karmic frequency I’m sending out will destroy any device that stays within 100 feet of my body for more than four hours over a two-day period. I start warning my seat mates on the 85-minute ride between Grand Central Terminal and my home town, on Metro North.
But what this illustrates is the unpredictability of unpredictability. You can’t project when one device will fail, precisely. Much less, many.
This is why you can’t build enough redundancy into a stock exchange (see
This explains also why there’s constant fear that a faulty algorithm will run amok and take down trading within and across all-electronic networks. Or that interactions between algorithms will spawn unexpected results. You just can’t predict every outcome.
In 1991, I spent several weeks chasing the cause of a series of outages of telephone service on the East coast, while I was reporting for the Dallas Morning News. Why? The switches involved were made by
The cause, which we reported first, was the most simple possible: A single mistyped character of code. The outage was due to a typo.
More recent events on Wall Street make you worry about innovation, quite severely. Subprime mortgage lending seemed like a great idea at the time. Hey, why let documenting income stand in the way of a good deal? And, hey, if we can create new types of securities that are layered on top of other types of securities, we’ll just grease growth in the economy (and our bottom lines). Right? Doesn’t matter if our information systems can’t tell us whether the people at the bottom of the pyramid are actually able to pay off their notes. This is innovation. It’s unstoppable.
Not.
Making systems rock-solid is job number one. "Tried and true" comes first.
Then, innovate.
And watch what you’re doing, every step of the way.
Karma will catch up to you, every time.
Or a typo will.
