September 29, 2005

Another Reason Why IT May Not Matter

While the question had been bandied about for years in the roadhouses of the chat-forum-blog culture, Nicholas Carr's lightning rod article in the May 2003 Harvard Business Review was the first time it had been posed in respectable circles.

Does IT Matter?

Elaborating, defending, and expanding in response to the firestorm of responses from mainstream IT pundits (those who think "change" means a new version of Cisco's router OS, not a sudden replacement of whole professions), Carr invoked the well-worn metaphor of industrial production capacity. You know, companies devoted capital, space, and staff to things like generating their own electricity until economics and enlightenment caused private power plants to be supplanted with electric utilities; now information infrastructures are following the same path.

Yes of course.

The public information utility was predicted by Ed Lias of Univac in the 1970's as I recall. By 1981 we at Delphi (along with H&R Block's CompuServe, Reader's Digest's The Source, plus Telenet and Tymnet) were offering access the first public information utilities. Ten years later when the brilliantly devious Commercial Internet Exchange scheme forced the feds to permit commercial traffic over the NET, we at Delphi quickly adapted our information wallplugs to the new Internet information utility.

Here we are fifteen years later and we're talking about a coming great leap forward from steam driven corporate information plants to the plug-in information utility for companies. Of course there's hugely more complexity to this particular version of the great leap forward: how do you make the information utility fit the immense assortment of established business processes without really messing up those processes?

I'm glad I don't have to manage a big public company through a transition like that, because I think the answer is: you don't. You might attempt what the big airlines tried to do, you know, Delta knew that its uncompetitive nature was so entrenched it had to start over with a new airline called Song.

But Song isn't just new paint on old planes, it's new talk on an old walk. Face it, they just can't think and act like Ryan Air, not just because of old ways of thinking but because of old commitments and old relationships. Wise management can only comfort itself with the thought that the outcome of these Greek tragedies is knowable.

Picture a Fortune 1000 IT department meeting on a timetable for moving toward SOA: how shall we respond to the COO's suggestion that we accept a joint $20 million proposal from Accenture and BMC. The idea is that will hopefully, after five years, tens of millions of dollars, have the company a third of the way toward full SOA implementation. Easy enough to picture, right?

Now assume that the company is Google. Or Yahoo.

The idea is laughable. What a traditional IT department regards as big, consuming, multi-year challenge (because of course steam-driven information plants are complex), Google doesn't even waste a minute thinking about. Picture Eric Schmitt, CEO of Google, thinking about a neat new thing. He writes a few lines of pseudo code, shows it to Larry Page or Sergey Brin and a little while later a user anywhere can plug into the information utility socket to get the new new thing. After all, the pumping plant is there, it's a matter of a little added pvc plumbing.

Ah, but the other 998 of the Fortune 1000 provide something physical, a product or a service. Could Google run an airline? Manufacture cars? Manage a hotel chain?

Of course that's not what Google and Yahoo do for a living. They just provide server-based information and communication services.

But wait, Google seems to be getting into the backbone business and Yahoo is solidly in the entertainment media business. Both of which of course are closer to traditional product-service categories.

Established industries and their IT departments believe that business information utilities are fundamentally different from consumer information utilities. And they are.

The design of a business information utility should be fundamentally easier to do than the design of the consumer version. A consumer information utility is like a consumer appliance: if you're going to have millions of low-price customers, the product or service had better work easily and flawlessly or the support costs will kill you. An industrial product or service or information utility, by contrast, can bear the cost of the handholding that makes up for lack of quality.

It would be easier to build and manage a business information utility (what SOA wants to be) if it weren't for customer loyalty to old relationships and commitments. Throwing out an old refrigerator is easy; throwing out a mass of entangling alliances is impossible, even if it is completely in the way of a competitive information infrastructure.

If effective information facilities are such a key success factor in any business, then what prevents those who have a way with code and packets and no entangling alliances from taking over the economic sectors that are now run by people who have a way with factories and distribution networks?

Answer: not much.

Posted on September 29, 2005 at 08:08 PM

TRACKBACK

TrackBack URL for this entry: