I like what I refer to as "Information Theory" (i.e. metaphysical musings on IT, which is the best definition this untutored writer can come up with) because it can predict answers and avoid wasting time on futile quests.  For example:

"Can I run an algorithm that will accurately predict whether a problem is solvable?"

• Non-theory based answer: ' You can tray method XYZ, but then you'd have to fall through to ABC for cases that it can't handle, which may slow it down, but if you... '
   
• Theory-based answer: ' No. '

Trouble is, certain mantras that are bandied about as advice or product designs are basically doomed not to work at a theory level, such as:

• "Just re-install Windows"
• Backup, backup, backup
• Human-applied passwords
• "optional" passwords
• Band-aid risk management
• Informal virus scanning
• We can patch it later
• Multiple user accounts
• No-maintenance file systems

That's if you define "work" to mean "work" rather than "cover my ass so when things fall apart, I can't be blamed".  That's also for 100% rather than fractional values of "work".

This is not to say there is no value in backups, passwords, host-based virus scanning, patches and (in very particular situations) re-installing the operating system.  It is to say that these measures are only partially successful, and may cause collateral damage - so they are no substitute for things that do work, where those methods can be applied.

I will write pages on each of these RSN (Real Soon Now); until then, you can consider this a "teaser" page and flame me accordingly    :-)

(C) Chris Quirke, all rights reserved; July 2004

Back to index