One loser from this change could be Microsoft. The price per gigabyte would increase, so bloatware would be harder to tolerate. Restrictive licensing practices might not fit well, either. But otherwise, software would benefit. In particular, voice recognition could become more practical, because you could port the training you've done of the system from device to device. The same goes for handwriting recognition and other customizations.

In particular, improving "disk" speed on PCs by multiple orders of magnitude could have all sorts of favorable consequences. Object file systems and full-PC search could be much more usable. Autosave would work better, and PCs wouldn't freeze as frequently as a result of browser cache thrashing. And as far as security permissions are concerned, anybody from number crunchers to engineering professionals could work efficiently on huge, personalized data sets.

Ah, yes -- security. Without the disk-access bottleneck, it would become reasonable to encrypt or decrypt a PC's entire database each time you logged off or on. No more security breaches from stolen laptops -- or, rather, from picked pockets and busted lockboxes, since it would no longer be a good practice to store the data with a thief-attracting laptop at all.

In another security-related area, two-factor authentication would also be easier, since the flash drive itself becomes the "thing you have," rather than a smart card or an RSA-style clicker device. Meanwhile, one objection that's been raised to this idea is, in my opinion, somewhat bogus. Yes, badly protected diskless machines aren't really secure; at least theoretically, they're vulnerable to "man in the middle" attacks. And perhaps that's an argument for continued use of laptops and against the idea of terminals in every hotel room -- although if you fear that kind of attack, why not fear spy cameras and audio bugs as well? But in any case, it's not a persuasive argument against corporate or home deployments. If your janitors can't be trusted, traditional PCs are at least as vulnerable as diskless ones.

Diskless PCs are too futuristic to figure into most current IT plans, although they do provide another reason to move away from Microsoft desktops (indeed, their support for dual booting will eventually smooth such a transition). But it's interesting to think about them now. If you want to pursue the subject further, you can find more thoughts and links on the subject on my Web site at www.monashreport.com.