Are Security Bugs Special?

One famous programmer thinks they're crowding out a lot of other good development that needs to be done. Are they just regular bugs like all the others?

I spend a lot of my time writing about security bugs found in products and what is done about them. In fact, there are hundreds of blogs and news outlets that spend a considerable amount of time on such bugs. The news "markets" seem to think they're worthy of the attention. An industry has built up in the harvesting of these bugs and their sale to clients, both for good and bad reasons.

Not everyone thinks this is a good situation. Take one Linus Torvalds who, in a recent e-mail discussion, called the whole security bug management process corrupt. [Warning: Linus uses some very coarse language in this discussion.] His basic argument is that security bugs are just bugs like non-security bugs. The focus on them is taking away attention from non-security bugs that also need to be addressed.

Obviously there's a lot of truth in Linus's argument. First, there's something just plain wrong about companies paying people to find exploitable security flaws in other people's software, and then selling that information. But Linus seems to be talking more about the impact of embargos on development. There is a private mailing list among operating system distributors and vendors called vendor-sec used (citing Wikipedia here) to discuss security issues in distributions and coordinate release of fixes.

Such "coordination" often involves an embargo on the release of information about these bugs, and it's not hard to see how keeping such things secret would rub the wrong way someone who has done so much, more than anyone perhaps, to keep the development of software open. The secrecy culture around security fixes seems, to Linus, antithetical to the philosophy that made Linux not just possible, but successful. He has some really nasty things to say about vendor-sec in particular (this is the part with the worst foul language).

Linus isn't the only one p-o'd at vendor-sec. About halfway down this very long message (watch out for more sailor talk) the author (Al Viro, another Linux kernel developer) goes on about what a mistake working with vendor-sec was for him, and for much the same reason as with Linus. Anyway, the "coordination" doesn't seem to have accomplished much. The list is populated with Linux and BSD distributors who are notorious for being out of synch with each other, although perhaps less so for security bugs.

I have to think that this is another case of an ideological viewpoint running smack into the brick wall of market realities. Yes, there's a defensible philosophical argument that all bugs deserve attention and that security isn't everything. I can understand Linus's distaste for the culture of secrecy around security bugs, but I do think they are more important that most other bugs. A really serious security bug, like the recent DNS bug, really does merit telling everyone to drop everything else and fix this right now.

But it's a free country, for the most part, and I hope nobody thinks they can tell anyone, least of all Linus Torvalds, what he must and must not write. Security's important, but it's not the only thing, and someone's got to keep the other functions working well too.

Security Center Editor Larry Seltzer has worked in and written about the computer industry since 1983.

For insights on security coverage around the Web, take a look at Security Center Editor Larry Seltzer's blog Cheap Hack