When I wrote last week about the future impact of Moore’s Law on platform economics, I didn’t anticipate the vigorous agreement of SAP co-founder and chair Hasso Plattner. That’s just what I heard, though, from the stage of the Software 2007 conference in Santa Clara this morning, where Plattner gave a segment of the morning keynote (and where salesforce.com chairman/CEO Marc Benioff will be keynoting the afternoon session at 2:00).
"It’s incredible how fast a modern CPU is when running software written for a modern CPU," said Plattner, talking about the changes that arise from massive memory enabling huge in-memory databases. He challenged the audience to think in terms of "Google speed for all information in the system," observing that Google’s search results are more valuable because they’re fast than because they’re exactly what you want (which they usually are not).
"You can use your brain to refine the search," he elaborated, and make four or five successively more focused queries using search terms suggested by initial results, "instead of taking a lot of time to compose an elaborate and specific query that returns nothing." An algorithmic approach to this problem, based on so-called "semantic Web" research, seems destined to keep multiple generations of graduate students provided with thesis topics.
I’d add two points to Plattner’s comments. First, it’s important that "Google speed for everything" should include unstructured as well as structured data. Second, there’s art in devising a search-centric user interface that serves people trying to do something, rather than merely entertaining people using the Internet to do nothing.