GATES: One of the key characteristics of the software industry is that, because of incredibly rapid technological change, products must be continuously modified to reflect innovations. For example, software will need to change to support speech input, which will be fantastic for users. So development costs are ongoing. With the high-volume, low-cost model adopted by Microsoft and the PC software industry, such costs are spread widely, so consumers pay a very low price to benefit from billions of dollars of R&D.
The key is in value and utility-if consumers get both, they will be willing to pay for them and, if the software is good enough, it will be used widely from the outset. So the world you are describing already exists: Consumers already get an amazing amount of functionality from their software at a very low cost. Contrast the old proprietary computing model, where software accounted for a high proportion of system cost, with the PC model, where software is only a tiny percentage of overall cost. That comparison makes much of today’s PC software seem almost free.
DERTOUZOS: In the commercial period, when the software is still evolving, a successful strategy for maintaining revenue is, increasingly, the annual upgrade, which, incidentally, adds to the “feature shock” of users. This practice, together with an evolving Web, suggests that we’ll move from buying shrink-wrapped software to simply buying upgrades through periodic downloads at a monthly fee. Do you see Microsoft and other software developers becoming such “service” organizations?
GATES: Regular upgrades are clearly necessary in an industry that is changing as fast as the software business-just as they are in, say, the auto industry. I can’t ever imagine a time when software will not continue to evolve in this way. With the high-volume, low-cost model, you have to make the software as attractive as possible to as many computer users as possible, and that means lots of features. And clearly not all of them will be used by every buyer. But in general I think you are right that, in order to “hide” the complexity and adaptability of software from the average user, upgrades will increasingly be carried out transparently and automatically, without users having to do anything.
So rather than having to ensure that your software is always up-to-date, the software will do it for you-you’ll wake up in the morning and the latest version of the software will have been installed overnight. To that extent, software will evolve into even more of a service business than it already is, and in the long term there will probably be a move toward a subscription-style model.
DERTOUZOS: Browsers and operating systems will merge in functionality, simply because people need to have the same commands for dealing with information, regardless of whether it is local or distant. On this, you and I agree. However, we disagree on how to get there: I dream of a system built from scratch that gets rid of layers of old software and brings a new truly easy-to-use metaphor to the Web-centric world, as important as the desktop was earlier. I believe that you want to get there gradually, by upgrading Windows. Recall that the Web itself was created by a small team of people, yet ended up on millions of computers. Could something like that happen here, with a new system that might spring out of nowhere? Would you consider replacing your own baby, ahead of a competitive threat, with a brand-new, simple, super-efficient browser-operating system?