A frequent pitch of the open-source movement—bordering on a cliche—is that closed-source companies, most famously Microsoft, are wrong in thinking that the unavailability of their source code will enhance the security of the product.
The open-source advocates belittle this attitude as “security through obscurity,” and insist that the holes in the products will come out anyway. Quite the contrary, the theory goes, the more secure product is the open-source one, which will be examined by anyone who cares to, looking to improve it.
Is this idea true? We may soon find out, thanks to the leak of parts of Windows NT4 and Windows 2000 to the Internet.
All day Ive been receiving analyst predictions that the release of this code will result in a large increase in the number of exploits of Windows in the wild. This prediction is an expression of utter faith in the idea of security through obscurity. If there was no security value in obscurity, then there should be no security to lose in the release of the source code.
Remember, we are talking about a code base of portions of two products several years old. According to Russ Cooper, Editor of the famous NTBugTraq mailing list, the NT4 source is from NT4 Service Pack 3, without the Internet Information Server and includes Internet Explorer 4.0.
The Windows 2000 code is a very small subset of Windows 2000 Service Pack 1. It includes Internet Explorer 5, Simple Network Management Protocol, Public Key Infrastructure, networking and some SDK code, along—strangely—with a large collection of empty mail messages.
Cooper said he would be very surprised if this leak results in any significant new risk. “Given how hard people have pounded away at the binaries in the past, pouring over 55,000 source file to find something new in old versions will likely, hopefully, be a very unfulfilling task.”
I have to agree with Cooper, although we are both guessing to a point.
First, anyone running a system with these particular builds of Windows will already be at substantial risk of attack from known problems that were discovered without the aid of source code, so its reasonable to assume that there are few such users out there to attack.
The question then becomes how much of the code in todays more-common configurations is common with the source code that was leaked? In other words, how many attacks can be built for todays systems out of the source from 3.5 years ago and longer?
If there are still meaningful attacks to wring out of the code, it will only be because there is security value in obscurity. Black hats and white hats have been throwing their worst at this code for many years now, and if an attack of significance comes out of this code leak I think it just goes to show that source code matters.
When you consider the number of institutions with licenses for Windows source code, its amazing that it took so long for this to happen. These licenses go a long way back too, and the program has been growing.
Given that non-disclosure agreements and other legal obligations are an important tool of closed source, Id say it speaks well of the practice that it took so long for this kind of incident to occur.
Another recent analysis claimed that this incident is like capturing a Russian fighter from the 1950s. Now, that may be an exaggeration, but it makes the point that a lot of the leaked code is obsolete already.
But the fear and excitement over all the presumed attacks to come against Windows—depending on who you were listening to—at heart betrays a lack of confidence in the derision of “security through obscurity.”
Theres no middle ground here, either closed source is more secure for being closed, or it isnt. While I tend to think it is, we may learn for certain now.
Security Center Editor Larry Seltzer has worked in and written about the computer industry since 1983. Be sure to check out
More from Larry Seltzer