The challenge of writing software lies in figuring out how to do things. But sometimes, the best enterprise software is distinguished not for what it does but for what it doesnt. Rights management features in end-user applications, such as those apparently planned for a forthcoming version of Microsoft Office, are the kind of technical achievement that should be nailed up in a crate and placed in the back of the warehouse. Like the artifacts that get this treatment in the final scene of "Raiders of the Lost Ark," some things are so easy to misuse that were better off without them.
A company would not install padlock hasps on office doors, effectively inviting employees to use their own locks that company security personnel would be unable to open. Neither would a company rely on individual employees or departments to hire guards, or otherwise manage physical security,
on an office-by-office basis. Companies manage physical security from the top down, and that level of assurance and control belongs in the digital realm as well.
The alternative is inconsistent protection, combined with the building of fences between groups that should be sharing common data repositories. Weve already seen the tip of the iceberg of rights management running amok, just in the simple feature of password-protecting individual document files in applications such as Microsoft Word. These are a cure thats worse than the disease: Like inadequately tested vaccines, they dont give much protection, but their side effects are devastating. So far as I can tell, the primary function of passwords on Word documents is to be forgotten by those who apply them; the rare password-protected document thats actually worth something to a thief can be cracked by any of dozens of tools or commercial services.
For fees ranging from tens to hundreds of dollars, and service times ranging from days to weeks, you can find out whats in almost any Word file. Wouldnt you hate to incur that nuisance and expense, though, to recover documents that might have been left behind in password-protected form by an ex-employee? Theres a reason why youll get more than 1,700 Google hits on the search string "data hostage," and its not because poets find it a picturesque phrase. Its a growing problem, partly due to proprietary data formats but also due to users having too many tools that can lock up data beyond the reach of any legitimate master key.
Moreover, implementation is everything; thats the other reason that data protection tools often do a better job of keeping data away from the owners than of keeping sensitive data safe from attackers. Users dont know the weaknesses, but professional data crackers do.
One popular freeware file encryption tool, whose name I wont mention because attackers dont need more help in choosing targets, introduced a password- changing feature a few years ago. Due to a programmers oversight, the first release with this feature produced encrypted files that contained two data structures, one representing the current password, the other representing the first password ever assigned to that collection of data—and in a way that allowed the original password to be re-enabled. Its not hard to imagine a system administrator taking advantage of a loophole like this to create a master key facility as an administrative convenience, but a system integrator or a malcontent administrator could also exploit this undocumented feature to access data without authorization.
Putting content rights management tools in end-user applications is a challenge; to do it securely, a team must anticipate vast numbers of possible interactions between applications and underlying systems software. Deciding not to do it, though, seems to me an even more laudable act of good software design. Rights management features in end-user applications such as Microsoft Office, with end-user control and lacking policy-based management tools, could bring chaos to collaboration and gravely complicate enterprise data management.
IT buyers and administrators should insist that application vendors introduce such potentially disruptive facilities only when theyre accompanied by strong management tools that give enterprise managers, not individual users, appropriate control of information assets.
Peter Coffees e-mail address is firstname.lastname@example.org.