Having joined Microsoft as a programmer when it had 40 employees, Charles Simonyi has seen the software industry evolve. Now, as the founder of Intentional Software Corp., he thinks a great--and painful--change will come from mechanization. Excerpts follow from an E-mail exchange with Simonyi:

InformationWeek: You say there's room for a lot of optimism. I've talked to a lot of people struggling in the industry who could use some. Where do you see the trends driving outsourcing actually creating domestic opportunities?

Simonyi: To me, the outsourcing trend indicates that an ever-larger part of IT work has become routine, repetitive, and low-bandwidth--one might even say unexciting or boring. The high educational level of foreign workers doesn't contradict this--they are underutilized by these tasks just as much as the domestic talent has been historically underutilized, but in this arena the lower costs and the more-patient producers win. The opportunity has to do with redefining the arena, to play where the value is, where the excitement is, rather than chasing the lower costs and search for more tedium-resistant programmers.

Take the very simple example where classically educated people in India answer phone banks to tell customers how to solve installation headaches. At some point, would it not make more sense to fix the software so that the need for the calls would not arise in the first place?

The fact is that all IT people today spend some portion of their productive time doing tasks that are at the threshold of mechanization. For senior people, this may be a modest 10%, for junior people it may be as much as 90%. Outsourcing has been historically a prelude to mechanization, and mechanization is a high-value domestic opportunity.

This may take a long time, but just look at the steel industry. After decades of shedding smoke-stack industries, we now have again a vibrant high-tech domestic steel industry in the form of the minimills. Total employment has been reduced, but the value per employee is much higher, the jobs are safer and more rewarding. I foresee something analogous in software.

InformationWeek: Writing software has been a nice career for a lot of Americans. Will it continue to be so? If so, how will it look different than today's market?

Simonyi: Nothing will remain the same for long, especially not in high-tech. Moore's Law isn't predicated on keeping the key parameters and key technologies constant--we on the software side should take a hint from this. I was just reading Ed Yourdon's comment how the least-productive U.S. software workers ("the bottom 20%") are doomed [Wall Street Journal, Oct. 6]. He is right. The conventional wisdom says that their jobs will be exported. I say that may be true in the short run, but in the long run these jobs will be 90% mechanized, with the help of senior domestic talent.

InformationWeek: How does a person train for this changing market, as, say, a college student, or a mid-career professional? Are there ways programmers have traditionally prepared for their craft that just don't make sense anymore?

Simonyi: It will all depend on when Moore's Law arrives to software, that is, when the routine IT programming activities become mechanized. At that point a programmer's three-year-old experience will be like a three-year-old laptop is today: a quarter of capacity, a quarter of speed, ready for replacement. The basics of software engineering will remain important but the focus will move to the next metalevel: how to automate things. Languages and compilers will have the limited and narrow importance that order codes and assemblers have today.

InformationWeek: You use programmers in the U.S. and Hungary, I believe. How do these onshore-offshore trends apply to your own company?

Simonyi: We are certainly not saving any money by doing that. We may be very atypical in that we were looking for quick availability of exceptional talent rather than the other factors that have been mentioned.

InformationWeek: OK, now give me the straight-ahead Intentional pitch--how's Intentional fit into this new world of writing software?

Simonyi: Intentional Software Corp. is positioned squarely to promote the mechanization of routine tasks and thereby reduce the need for outsourcing--not only outsourcing from the country but even outsourcing from the organization that has the major stake in the software solution. So a diesel-engine manufacturer may not need to outsource the engine controller software to a programming house. ISC will do this by creating proprietary technologies and tools that address the necessary conditions of mechanization, namely that the subject-matter expert's inputs (their "intentions") can be faithfully recorded in a machine-transformable file, and that the programmers' expertise can be expressed by them in terms of a transformation from the intentions file into some product. So ISC's tools will help redefine the relationship between the subject-matter experts and the software engineers. Today this relationship is based on low-efficiency, person-to-person communications. While these communications may include high degrees of domain formalisms, they have to be acted on time after time by the human recipient, which makes the process expensive as well as error-prone. We need to augment and largely replace the low-tech person-to-person communications with high-efficiency person-to-transformer communications, whether it's a subject matter expert adding an intention, or a programmer changing an aspect of the implementation.

Return to main story: The Programmer's Future