Click to win

W On Newsstands Now
Issue 3.06 | Jun 1995

Reality Check

The Future of Software
Raw computing power is doubling every 18 months, but software is still too high-priced, too difficult to use, and too buggy to trust for many purposes. To compound the problem, only an Žlite few are able to program useful applications. So computer scientists are developing new paradigms of program design, development, and distribution. But today's highly paid programmers may find the results unwelcome. Wired asked several experts to consider when software brains will catch up with computing brawn. - David Pescovitz

Provably Correct Application Software Superdistribution Interoperable Objects Offshore Programming Evolved Application
Scott Brown2010 2005 2005 unlikely 2100
Danny Hillis unlikely unlikely 2005unlikely1999
Derek Leebaertunlikely 1996 205020251998
Richard Mark Soleyunlikely 2010 2005 unlikely 2005
David Vaskevitch unlikely2000 2000unlikely2020
Curtis Yarvin unlikelyunlikelyunlikely20051998
Bottom Line unlikely2005 2019unlikely2020

Provably Correct Application: Few of our experts think a provably correct commercial application - one that has been mathematically shown to be bug-free - is a realistic goal. Brown thinks provable correctness can rarely be justified because of its cost, but he expects the technique to be worthwhile for some mission-critical applications within a decade. Several of those polled agree with Hillis, who says even a program "proven" to have no bugs "will sometimes show unexpected and undesired behaviors." And Leebaert points out that changing variables like "support environments, operating systems, and chips" could affect program correctness. Soley puts the question of correctness in perspective by quoting a friend: "The last bug isn't fixed until the last user is dead."

Software Superdistribution: Most of our experts think the hardware necessary for superdistribution of software - in which users are charged according to use, like electricity billing - will be standard in personal computers by the end of the decade (see "Superdistribution," Wired 2.09, page 89). Soley thinks computer manufacturers "need to be assured of broad compatibility standards" before lower prices associated with superdistribution will make "renting" software attractive to users. But Yarvin finds it hard to believe that computer owners will willingly pay extra for hardware that restricts their ability to copy software. Hillis goes even further, predicting that the development of common software applications such as word processors will be so easy by the turn of the century that a "great software price collapse" will occur and "people will pay you to use their software."

Interoperable Objects: According to those polled, we're approaching a day when most applications will be built from interoperable objects that will allow you to use one manufacturer's search function with another's word processor, for example. Vaskevitch points out that with current and proposed object-oriented programming standards, "this is already happening much more in real life than people realize." Soley agrees: "A surprisingly large number of applications developers have over the years committed to a set of bottom-line interoperability specifications," like Lotus's WKS file format. However, Yarvin argues that because code is so idiosyncratic, most of the software world is likely to remain incompatible.

Offshore Programming: Even with programmers' hours becoming longer and overseas labor becoming increasingly skilled and available, the majority of our experts thinks most commercial coding for American customers will remain in the US. Brown, how-ever, expects coding to become less labor-intensive, and hence less advantageous to move offshore, as "visual programming tools and libraries of commercially developed objects" become available. But Vaskevitch insists that "our novels will not be written offshore, nor will our programs." Leebaert thinks offshore programming will inevitably be cheaper despite the political and logistic impediments. Yarvin agrees that sheer numbers and smarts will bring code importation over the top. "There are already more programmers outside the US than in it," he says. "And these boys don't smoke dope."

Evolved Application: Our experts think a program employing techniques such as genetic algorithms to artificially evolve code based on a user's needs is on the horizon. Hillis believes the first programs of this type will be "hybrids of evolved and engineered parts." He says that in five years, a commercial application that at least partly depends on evolved code will be developed. However, while Brown agrees that evolved software is technically feasible, he asks, Will it sell? He argues that better programming techniques, such as visual programming, "will reduce the need for computers to write application programs."

Reality Checkers
Scott Brown

Manager of Advanced File Systems, Novell Inc.

Danny Hillis

Founder, Thinking Machines Corporation

Derek Leebaert
Adjunct professor, Graduate School of Business, Georgetown University; editor,The Future of Software (The MIT Press)

Richard Mark Soley

Vice president and technical director, Object Management Group Inc.

David Vaskevitch

Director of enterprise computing, Microsoft Corporation

Curtis Yarvin

Language architect, Xaos Tools Inc.

Wired Blogs

Bruce Sterling Read Wired magazine columnist Bruce Sterling's blog, Beyond the Beyond.

The Long Tail Read Wired magazine editor in chief Chris Anderson's blog, The Long Tail.

Gadget Lab Newsletter

Gadget Lab Sign up for the newsletter that gives you the 411 on the latest tech toys.
Gadget Lab Archive
Gadgetlab XML feed

Get Wired

2005 National Magazine Awards

Wired Blogs

Beyond the Beyond
by Bruce Sterling

Wired 40


Browse Issue Archive

Browse by cover