Baby, Can You Drive My PC?

John Perry Barlow

As you may recall, I concluded last month's column with the bold assertion that I thought NEXTSTEP might now make good practical sense for such entities as brokerage houses that trade entirely in the virtual world.

In such fluid terrain as they inhabit, relevance is constantly being redefined. It follows that one's methods of assessing the environment must be as rapidly mutable as the environment itself, and the electronic genome of NEXTSTEP can spit out new code quicker than a spreading virus.

For this reason, I find that NEXTSTEP is suddenly being taken seriously by people who might once have dismissed it. In recent days, I've talked with MIS types from as far away as Holland who are taking a serious look at it for the first time.

It makes an enormous difference that you can now experiment with this still-exotic OS for $800 a copy rather than $8000. There are also important accounting differences in software versus hardware acquisition that make NEXTSTEP much more attractive to potential customers as a soft product.

But at a certain point in these conversations, I'm always given a level look and asked something like, "But do you really think NEXTSTEP makes sense now? Don't you have any doubts?"

Hell, yes. It's my job to have doubts. First, of course, is the concern any rational person might have about the continuing survival of NeXT. But I'm not sure this should be a major consideration. MIS can always sigh and haul out that fat stack of Windows NT installation disks . . . whenever Microsoft gets around to shipping them.

Then there are compatibility problems. But in a conventional financial-services shop, the shrinkwrapped software that runs beside your custom apps usually consists of programs such as Lotus 1-2-3, for which native NEXTSTEP software like Athena Design's Mesa shows a high degree of compatibility.

But I'm worried about drivers. In the many years since I had a PC, I had forgotten that the price of a truly open computer architecture is a purgatory of configuration. While I have a strong philosophical agreement with the principle of open systems, it's another matter when you're actually down there among the DIP switches and the .SYS files.

This is not going to be an easy problem to solve. In the DOS, Windows, and, presumably, NT universes, drivers are often written by the device manufacturers themselves in the expectation that their hardware will be encountering one of these operating systems. It is unlikely they will create additional drivers for an operating system that will be found on only a small percentage of machines.

The driver problem is a liability that might become an asset over time. NeXT's object orientation should make it possible to produce drivers more rapidly than for the competition. Eventually, it shouldn't be necessary for each application to have a driver for every monitor and printer. Under NEXTSTEP, applications will deal with the OS rather than addressing hardware.

Still, it's going to be a mess. It's already a mess. I'm preparing for some character-building experiences.

Hey, you know what would be really cool? A system in which the hardware was designed from the bottom up to be integrated with the software that runs on it. Then we'd never have to think about these things. We could just plug 'em in and go.

Nahhhh, it'd never make it in the marketplace.

John Perry Barlow drives under a variety of influences.