Symantec has changed how it spells out Firefox and Internet Explorer browser vulnerabilities in reaction to complaints last September from Mozilla Firefox users and developers.
"How we did it before wasn't a fair comparison," said Oliver Friedrichs, the senior manager of Symantec's security response group. "It wasn't an apples to apples comparison."
Previously, Symantec's Internet Security Threat Report counted only vendor-confirmed bugs in the two browsers, which led to gripes from Firefox fans that the Internet Explorer tally was inaccurate, and too low.
In the newest report, which Symantec issued Tuesday, the Cupertino, Calif.-based security company has split the counts into two categories: vendor-confirmed and a combination of vendor- and non-vendor-confirmed flaws.
That gives the edge to IE in one tally, Firefox in the other.
In the last six months of 2005, Microsoft confirmed 12 vulnerabilities in Internet Explorer, down slightly from the 14 in the first half of last year. Firefox, however, sported 13 vendor-confirmed flaws, one more than IE, but also down from the 27 in the previous period.
In fact, when counting only vendor-confirmed bugs, Firefox appears to be significantly more vulnerable than IE over the last 18 months. During that period, the number of Firefox-admitted flaws easily topped 60. In the same period, IE posted fewer than half as many vendor-confirmed bugs.
Explaining the difference, Friedrichs said "In open source, more vulnerabilities will be acknowledged because of the transparency in development."
But the new counting methodology, which Friedrichs said was the "more accurate" of the two, combines all vulnerabilities, including those made public but not necessarily confirmed by the vendor.
In that count, IE comes out second-best: In the same six months, Firefox suffered from 17 total vulnerabilities, while IE had 24.
"The vendor- and non-vendor-confirmed numbers are the ones I'd recommend using," said Friedrichs. "For one thing, it removes the delay that can effect numbers because of long patch times by commercial vendors."
Symantec, said Friedrichs, won't make claims that one of the two leading browsers is more secure than the other. "We just stick to the facts," he said. "But the number of vulnerabilities are legitimate, so we can say that Firefox has fewer vulnerabilities."
Among the other data in Symantec's report are new "time to compromise" figures that try to gauge how long an unpatched, unprotected computer would last before it has snatched by a hacker.
Windows XP Professional, said Symantec, stays safe just one hour and 12 seconds, while the Windows 2000 Server (with SP4) made it an hour and 17 minutes. An unpatched Windows Server 2003 system lasted somewhat longer.
In contrast, unpatched Linux installations of both Red Hat Enterprise Linux 3 and SuSE Linux 9 Desktop were never compromised during their month-and-a-half exposure to attackers.
Patched Windows systems, however, remained untouched throughout the test, backing both its and Microsoft's advice to patch regularly, and patch promptly. "Applying patches in a timely manner is an important component of an effective security strategy," the report read.
Overall, patches are rolling out of vendors faster than ever before, the report noted (although it was mum on whether those patches had been applied). During the last half of 2005, what Symantec calls the "window of exposure," the time between the appearance of exploit code and the release of a patch, was 42 days, more than two weeks shorter than the 58 days of 2005's first six months.
Not that that's good.
"During this [window], end users and administrators may be forced to implement security 'workarounds' without an official fix and networks could be vulnerable to compromise," the report concluded.