Keynote Address: Developing on "Internet Time"

Summary by Idajean Fisher

The underlying premise of the keynote was the notion that as the Internet becomes more market and marketeer driven, developers' time lines are constrained to become increasingly compressed. He further asserted that this is a predictable outcome of what he described as the natural conflict between marketing and the engineer.

By way of example, he cited his own experiences with the early development cycle for Java-specifically, the fact that developers were able to benefit from three years in a pure development environment with little or no outside intervention.

In the presenter's opinion, the reality today is rapidly becoming that, on rumor of product, contemporary marketing engines gear up for action. This tends to drive down the time allocated to pure development. It also tends to destroy the traditional roles of alpha and beta test cycles in the development process. Whereas a few years ago new code would be sent to alpha test to a small, well-defined test user community, today alpha testing may be more rightly described as large scale software distribution (on the order of thousands or tens of thousands of users).

What are the potential impacts on traditional software development processes? The presenter suggested the "beyond the edge" development model which was described in terms of "design for today" and by delivery fall into "instant obsolescence" to be one possible impact. He also proposed a perceived impact on the way industry at large looks at risk taking, the supposition being, that taking risks at times demands failure. Therefore, it can be argued that insufficient failure can in turn be interpreted as insufficient innovation on the part of the developing entity.

Finally, it was suggested that these two changes in concert have produced an underlying change in the political environment in companies heavily involved in software development. Further, the supposition is that this in turn leads to increased numbers of spinoffs and dedicated business units to pursue single-track development not easily allowed by the newly imposed time lines in a single organizational structure. That all of this in turn leads to increasing parallelization in the development model employed by large software organizations was also implied. In other words, employing lots of individual business units with single-project focus and highly redundant development methods is an effort to try to speed up time to market.

Originally published in ;login: Vol. 22, No.2, April 1997.