Microsoft Refutes Reports of Windows Flaws

Microsoft is responding to a report published last week by London-based security intelligence firm Mi2g that claims the Apple Macintosh operating system and certain varieties of Unix are less vulnerable to attack than the popular Windows and Linux operating systems.

The report, a summary of which was released to the public by Mi2g, attributes 44 percent of the software vulnerabilities announced in the first 10 months of 2002 to Microsoft's Windows operating system and 19 percent to the open-source Linux OS. By comparison, the company attributes only 1.9 percent to Apple Computer's Macintosh OS.

In an interview Mike Nash, vice president of the security business unit at Microsoft, said that he feels those numbers are misleading.

"Essentially what [Mi2g] has done is look at a combination of vulnerabilities announced by vendors and new vulnerabilities reported by users," Nash says. "There's no way to determine if the same issue is counted multiple times, or if erroneous vulnerabilities are being reported."

More Users, More Problems?

Products with more customers, like Microsoft Windows, are bound to have more vulnerabilities reported under such a system regardless of whether those products are less or more secure than the competition, according to Nash.

However, Jan Anderson, a member of Mi2g's Intelligence Unit, says that the small size of the Macintosh operating system's user base--what Mi2g refers to as "security through obscurity"--does not entirely account for Mi2g's results.

"Our main point here is that although only about 3 percent of systems are running Mac OS, the proportion of attacks suffered by these systems is 60 times less than this, i.e., 0.05 percent. There are also relatively few known vulnerabilities of Mac OS, as stated in the news release," Anderson wrote in an e-mail response to questions.

Counting Flaws

The issue may come down to which vulnerabilities get counted and which don't.

In a statement, Mi2g said that the company is in touch with Microsoft at a senior level and that the two companies are working together to deal with the issue of vulnerability counting.

"Our methodology relies on collecting vulnerabilities and slotting them into two categories: confirmed vulnerabilities and candidate vulnerabilities. We collect our vulnerabilities from recognized, credible, and reliable sources, including CERT (the Computer Emergency Response Team at Carnegie Mellon University), eliminating all duplicates and discounting for multiple references. We do look at vulnerabilities which affect a particular operating system as a whole even though they may originate at a server or application level," Mi2g said.

D.K. Matai, chief executive officer of Mi2g, says that removing unconfirmed reports from Mi2g's numbers doesn't improve the picture for Microsoft.

"Even there, we note that Microsoft doesn't account for 44 percent of vulnerabilities, it accounts for 54 percent," Matai says.

According to Matai, Mi2g compares its data against data maintained by independent software-vulnerability tracking organizations such as CERT, and that the company's numbers are consistent with those maintained by CERT.

CERT declined to comment on the vulnerability data it maintains.

What Really Matters

Regardless, software industry analysts and security experts consulted for this story agree that looking at the number of reported vulnerabilities is a poor measure of an operating system's security.

"Comparing the number of vulnerabilities to shipments of the software is interesting, but not very useful," says Dan Kusnetzky, vice president of systems software research at IDC.

"All software is written by people, and people are fallible. The thing to look at that's more important is, when problems show up, of any kind, what is the response from the software vendor? How quick is the response? If the response comes six months after a problem was reported, that's not good."

Marc Maiffret, chief hacking officer of eEye Digital Security, agrees.

"All operating systems have vulnerabilities," Maiffret says. "The question is, 'How fast were they fixed?' and 'Is there a way to secure the vulnerability in advance of a fix--a guideline document that would have helped?'"

While Apple's OS may have fewer reported software vulnerabilities, Maiffret and Kusnetzky both say that there are also few incentives driving hackers--or companies like Aliso Viejo, California-based eEye--to scour that product.

"Breaking into Mac isn't something that gets a hacker kudos in his or her community. Breaking Microsoft gives that person the ego dollars that they depend upon," Kusnetzky says.

Asked what businesses and individuals should do to assess the relative security of the various operating systems, Maiffret and Kusnetzky both suggested that customers focus on securing their systems and assessing the ability of the software vendor to respond to problems as they arise, as opposed to worrying about the number of vulnerabilities.

But both Maiffret and Kusnetzky stop short of giving Microsoft a pass on the question of security.

"Microsoft does need to do a better job at being secure," Maiffret says. "There are too many trivial mistakes that you'd think a billion-dollar company wouldn't make."