Hosted by:
Created with:
|
********************************
Anti-adware misses most malware
By Brian Livingston Now that 80% of home PCs in the U.S. are infected with adware and spyware, according to one study, it turns out that nearly every anti-adware application on the market catches less than half of the bad stuff. That's the conclusion of a remarkably comprehensive series of anti-adware tests conducted recently by Eric Howes, an instructor at the University of Illinois. Howes, a well-known researcher among PC security professionals, collected 20 different anti-adware applications. He then infected a fresh install of Windows 2000 SP4 and Office 2000 SP3 with several dozen adware programs in separate stages. Finally, he counted how many active adware components were removed by each anti-adware product. (Note: I use the single term "adware" in this article to refer to both "adware" and "spyware." Since it's not necessary for a spyware program to "call home" to be disruptive, the distinction between adware and spyware is meaningless. All such programs display ads or generate revenue for the adware maker in some other way. ) Howes's tests were conducted over a period of weeks in October 2004. His results were mentioned at the time in several places, including Slashdot and eWeek. Unbelievably, however, none of these commentators bothered to print a simple chart showing which anti-adware application did the best job at removing the unwanted components. Even Howes himself hasn't posted such a summary. In a telephone interview, Howes exhibited both modesty and perfectionism, implying that his work wasn't yet done to his satisfaction — despite the fact that his tests are some of the most extensive I've ever seen. Howes's test results sprawl over six long Web pages, with no overall totals or summary of the figures. It's a daunting body of data, but its bottom line is explosive. Adware seems to be evolving much faster than anti-adware, and the battle is so far being won by the adware side. For this issue of the Windows Secrets Newsletter, therefore, I've complied Howes's figures into a straightforward chart, shown below. I removed five products that didn't complete all of Howes's tests for a variety of reasons. What's left is a revealing rating, from the top to the bottom of the anti-adware heap. Each anti-adware application, according to Howe, removed a certain percentage of "critical" adware components. These are executable .exe and .com files, dynamic link library (.dll) files, and Windows Registry entries (autorun commands and the like). Almost all the anti-adware programs that were tested removed fewer than half of the hundreds of adware components Howes cataloged. The best at removing adware was Giant AntiSpyware, but even that program removed less than two-thirds of a PC's unwanted guests. Giant AntiSpyware catches 63%, tests say Howes's tests were conducted before the Microsoft Corp. announced in December that it was purchasing Giant Company Software outright. For that reason, the tests use the version of Giant AntiSpyware that was available in October and not the newer Microsoft beta version that's currently available. Even so, with Giant's application removing 63% of a PC's adware components, and its nearest competitor, Webroot Spy Sweeper, removing less than 50%, it's clear that Microsoft has a potential winner on its hands. In the following table, which was reviewed by Howes himself before its publication here, the Adware Fixed column represents the percentage of critical components successfully removed, not just detected, by each product (higher percentages are better). The False Positives column shows the number of benign Windows files that were incorrectly reported by a product as adware (lower numbers are better):
Howes didn't test the anti-adware programs in the above list against a program called CoolWebSearch (CWS). This little bugger mutates every few days, it seems. CWS actually requires a completely separate anti-adware program, CWShredder, which is constantly evolving along with the nuisance. This is explained in more detail later in this article. The fact that anti-adware products fail to remove all or even most adware components has been an open secret among security professionals for some time. For this reason, tech writers often say, "You should install two different programs and run both of them for maximum protection." To test this assertion, I compiled Howes's raw data into a new table showing the removal rate of the best app, Giant AntiSpyware, with every other tested product. According to this analysis, combining Webroot Spy Sweeper with Giant AntiSpyware did the most to remove unwanted components. But the combination of the two apps increased Giant's 63% success rate only 7 percentage points, to 70%:
Finally, the computer press often recommends that the two anti-adware products that should be used together are Ad-Aware SE Personal and Spybot Search & Destroy. That preference may have become the conventional wisdom because both of these products have low-end, freeware versions. PC World, PC Magazine, and other publications have recommended this combination as recently as June and August, respectively. Ad-aware and Spybot may have been a great combo back then. But adware apparently moves much faster than these two companies do. According to Howes's data, the two programs together barely removed half the adware components on an infected PC:
I found no combination of any two anti-adware programs that removed more adware components than Giant AntiSpyware and Webroot Spy Sweeper, based on Howes's data. Removing only 70% of adware, unfortunately, isn't good enough. A much better strategy is to prevent adware from getting into your systems in the first place. I'll cover that next. How to defend yourself against adware First, let me make my opinion clear: The installation of adware should be illegal and harshly punished. Adware has exploded because it offers big economic incentives for its sponsors. They'll never adequately inform PC users about their software before it's installed. This troubling aspect of adware will never be wished away. Only software that a PC user specifically consents to should legally be able to install — and "end-user license agreements" that stretch off the screen should never be counted as consent. (This isn't a knock on "ad-supported software," such as the Opera browser. Such legitimate software is clearly integrated with its advertising and makes it easy to shut off the ads by registering.) In reality, today's tech-illiterate legislatures will never ban adware — if they could even think of an effective legal approach to do so. We need to engage the battle on a technical level instead. To understand adware, you first need to know how PCs get it. The ways that Howes obtained the adware he used in his tests provide us with some perfect examples:
The underlying reason that adware has compromised the entire Internet is that there's big money to be made. The best analysis of this I've seen is by Benjamin Edelman, a Harvard Law School student. He's documented almost $140 million in recent investments by Silicon Valley venture capitalists in just four of the largest adware makers. See list of adware angels For those who are interested in deeper research on adware, links to Eric Howes's raw data on his comparative tests are posted on his anti-spyware testing page. To send us more information about adware, or to send us a tip on any other subject, visit WindowsSecrets.com/contact. You'll receive a gift certificate for a book, CD, or DVD of your choice if you send us a comment that we print.
|