Opinion: Big research budgets beg relevance questions.
Our colleagues at Baseline released in November a survey of R&D spending by 20 tech leaders. Looking at their average "R&D tax" of 8 percent of sales, I found myself thinking about the shopping list that I would write if those budgets were mine to command.
If I had to allocate Microsofts $6.6 billion idea fund, Id be urging Microsoft Senior Vice President Rick Rashids shop (research.microsoft.com) to push even harder on generating metadata as a byproduct of the things that we do with information.
This would make metadata search far more useful in real life, instead of being great demoware that requires a library scientist to make it fulfill its potential.
Metadata generation would be an ideal partner for something else that Ive wanted for years: automatic versioning of all files, instead of leaving that function to be handled by individual applications.
As a counterpart to versioning, Id like to have the operating system generate and tag new file versions automaticallywith reference, for example, to events on my daily calendar.
This would let me issue a query for "documents I originated within two days of meeting with Andrew and Barbara regarding Project Charlie" or "documents I opened during my call with client Dave."
The system could track my use of data so that I could ask for "images I embedded in documents subsequently e-mailed to any customer engineer in the Eastern Field Group."
Would this require some cleverness in designing extensible data structures? Yes. Would it require new computer science? No, as any REXX or Lisp hacker can attest.
Admittedly, it would be one of the worlds least sympathetic situations to have the problem of allocating Microsofts R&D resourceswhich are not merely large but growing, with last years allowance up 6.4 percent from the prior year.
Its quite another thing to have the R&D funding task at Hewlett-Packard (which spent less last year than the year before, according to the Baseline study) or Dell (whose R&D budget is flat and less than 1 percent of sales). What should the R&D charter of these companies be?
Some people might be cynical enough to think that HP R&D is best focused on making printer color management even more complex than it is.
After all, the company would make even more money if better color quality were on the other side of a minefield of practical difficultyso that users wound up making more prints that werent quite right.
If black ink plus red, green and blue ink arent generating enough cartridge sales, why not move to six separate colors? Or eight?
Oh, wait a minute, Canons already there, with its Pixma Pro9500 scheduled for release next year and slated to use a 10-color pigment ink system, includingseriously, folksnew gray and matte-black inks to get just the right nuances in black-and-white prints. So HP has to find another path for differentiation.
Why not move in the other direction, with improved optimization algorithms that let you print a test pattern, scan the result back into the machine and automatically tune the printer for maximum accuracy in printing on that specific type of paper?
Ill even give HP the tagline for marketing this idea: "Let others lower the cost of what you print; we lower the cost of what you keep."
As for R&D at Dell, Im not persuaded by Baselines excuse that Dell "focuses on low-cost distribution, not new product development."
What Dell should be researching is both ends of the fulfillment process: rapid identification of what the customer wants and resolution of post-sale problems with maximal satisfaction at minimal cost.
Rather than making customers wade through lists of options, Dell should be exploiting rich media to let customers drag and drop options in a visualization of the custom-build process.
The company should be leading efforts in natural language understanding and expert-system engineering to make customer support its flagship feature.
What would you put on your key tech partners R&D to-do list? Tell me by commenting on the blog post "Making a to-do about R&D" at blog.eweek.com/petercoffee
Technology Editor Peter Coffee can be reached at firstname.lastname@example.org
Hey, big spenders
See who has the R&D mojo go.baselinemag.com/rd
Whats the meta?
Look over the high bar that Vista may not clear go.eweek.com/metadatamodels
More color for less money
Learn some photo printing tricks
Check out eWEEK.coms for the latest news, reviews and analysis in programming environments and developer tools.