I once saw Jonathan Schwartz, Sun’s new chief executive, speak at a pre-recession trade conference. It wasn’t a conference to announce the recession, although Sun would have been the most appropriate outfit to sponsor that shindig. Schwartz uttered three words that stuck in my head and which I’ll always associate with him: “Choice is bad”.
Say what? I couldn’t believe my ears. His proclamation flew in the face of everything IT was doing and thinking at the time, and everything that the technical and mainstream media were talking about. We loved choice, didn’t we? “Let us choose!” was the rallying cry of Linux and open source, the flag under which grassroots battalions drove AMD into the market, and the mantra of the Macheads. We couldn’t have too much choice in search engines, content portals, appliances, ergonomic keyboards, Linux distributions, 3-D accelerators, smart phones, ActiveX controls, integrated development environments or venture capital opportunities. We were tech junkies constantly pushing for the new — anything new — even when we were happy with what we had. This created what looked like unlimited demand from thin air — there was unprecedented interest in every new thing, but only until the next new thing came along.
Choice can be bad, but not just because it distracts us and distorts the tech industry’s view of the market’s desires. Choice can also breed inertia. Some technology buyers must be deprived of the option of standing still so that the rest of us aren’t stuck behind them in line. For example, customers’ demand for choice forced Microsoft into indefinite support of every version of Windows ever sold, lest someone complain that Bill Gates pulled the plug on an application that they can’t, or would rather not, upgrade. Microsoft was still wasting support staff and engineers on 16-bit software and drivers, and Windows NT 4.0 servers, when Windows Server 2003 was about to be born. I was at a reviewer’s workshop when I learned that Microsoft had finally gathered the nerve to end-of-life its legacy OSes.
Choice isn’t bad. It just took buyers a while to remember how to use it properly, to cull the number of choices and the time spent choosing. Choice for its own sake, or for the sake of maintaining the pretence of a level playing field, is down the tubes. There isn’t a need for 30 equally regarded commercial Linux server distributions; Red Hat and Suse will do. There isn’t a need for six x86 microprocessor makers; Intel and AMD suit the mainstream commercial market’s needs nicely. We don’t benefit from the freedom to choose from a menu of competing standards, would-be standards and proprietary approaches.
A sagely editor of mine once gave me his take on tech industry economics: balance is maintained when there is more than one leading vendor, but fewer than three. I thought the message was that a couple of giants should be encouraged to run roughshod over the whole of IT. That’s not the message. Choice is a powerful force when used well. We need to keep the ranks below the market segment leaders well stocked because not everyone can afford to buy from the leaders. Also, leadership in any area of technology, whether it’s operating systems, enterprise databases, or microprocessors, is not static.