day 88

digital darwinism 10: the reset button

Yesterday we learned about the concept of creep -- the rate of change in requirements between the time the coders start hacking and the time the product is released. Back to my point: whether or not a software product can be developed without chaos depends on the creep. As Capers Jones and Figure 2 show, extremely small values of creep (5% or more) result in chaos. Without a strong architectural basis, creep is inevitable. Without an enduring architectural plan, creep rates of 5% are more likely than 1%. Bingo! What is the consequence of creep? Does it matter that applications based on pervasive infrastructure software like ActiveX resemble the Denver Airport baggage-handling system rather than a Swiss Army knife? The computer industry collapses about every decade under the weight of its own chaos. We give these collapses names: the mainframe era, the minicomputer era, the personal computer era, and now the network era. This catharsis sweeps away the tangled webs we weave, and everyone starts over again. The old legacy systems are allowed to fade away and next generation systems spring up like a rising Phoenix. Our decade is nearing its end. The industry has pushed the reset button. Everyone is about to start over again. Soon, Windows API, DMI, MIB-II, OLE API, CORBA ORBs, OpenDoc, TAPI, TSAPI, CTI, and others will be gone. Like my old copy of Bill Gate's first BASIC manual, these systems will be little more than stories we tell our bright-eyed graduate students at boring faculty parties. In their place will be new systems based on Netscape plug-ins, HTTP/HTML, Perl/CGI, VRML 2.0, JavaBeans, JDBC, and other new pretenders. They will be mutations of today's Internet and WWW nuclear blasts. They will be tomorrow's troubling legacy systems. Like their predecessors, they will crawl out of the swamp of chaos, evolve into giants of the industry, and thumb their noses at architecture.

digital darwinism 1 2 3 4 5 6 7 8 9 Daily Dose Index