Comdex Keynotes, Java Developments Target IT Complexity

Peter Coffee: Mere price/performance isn't enough, if making it work costs too much.

Last weeks Comdex trade show, in Las Vegas, offered some startling sights to those of us who remember when this was the signal event of the IT year. The modest "Used Cisco" booth sitting next to the Nokia pavilion, the acupressure massage devices being hawked next door to the Hewlett-Packard and Sun Microsystems exhibits, and the almost total absence of Sony--in former years, always showing a spectacular array of innovative products and concepts--were particular items that caught my attention there.

Many of the keynote speakers, though, did not disappoint: AMDs Hector Ruiz made me look good by showing off the performance and the forward software compatibility of the 64-bit Athlon, for which Ive long been sharing my high expectations.

Suns Scott McNealy has finally learned how to give a speech about what his company can do, and why his customers should care, instead of wasting time with slams at Microsoft--and he offered a clear choice between the IT options offered by Microsoft, Sun and IBM. If the rest of Sun can learn to market its way out of a paper bag, the company is positioned to do great things.

National Semiconductors Brian Halla almost made the trip worthwhile for his keynote alone: How often does a person get to hear the first public presentation on a newly identified price/performance trend (dubbed Bahais Law) that may someday be as widely cited as Moores Law?

I also managed to catch up with Stephen Wolfram, creator of Mathematica and author of the controversial book "A New Kind of Science," on the day before his keynote speech, when we had a long conversation about the book that Ill share in my eWEEK column a week from today. In short words, discussion of Wolframs book has focused too much on the question of whether the universe runs on simple rules: Whats much more important are the tools that we now have for exploring entire spaces of rules, looking for new behaviors and for tools we can use to manipulate systems in much more systematic ways than ever before. Thats the "new kind of science" that most of the books many critics accuse it of failing to deliver; despite my own considerable initial skepticism, Im now leaning in Wolframs direction.

Wolfram and McNealy, moreover, had common themes in that both were talking about complexity: Wolfram in terms of mastering it by analysis, McNealy in terms of killing it with research and development. Both of them offer important messages of hope to enterprise IT builders who are finding that the complexity of IT offerings is threatening to offset, or even overcome, their improvement in gross price/performance.

Also attacking costly complexity is software vendor SAP AG, which as of last week is now a member of the Java Community Process programs Executive Committee for the Standard/Enterprise Edition of the Java platform. I spoke with SAP development manager Michael Bechauf about the companys goals as a committee member: "Enterprise application developers need a clear separation between system-level and application programming," he urged. If a platform doesnt help a development team do its job with high-level facilities, that team will use its lower-level tools. But then, said Bechauf, "youre in danger of writing non-portable code, and you need staff teams whose job is preserving portability--and then you have overhead."

Bechauf sounded a lot like McNealy, whose keynote included the comment that it should be "a court-martial offense" to write enterprise applications with platform dependence. Both of them sound as if they hope to refute Parkinsons Third Law, that "success leads to complexity and complexity leads to decay."

In this Thanksgiving week, heres wishing you the kind of success thats built on simple things.

Tell me how youre defining IT success