- Is Microsoft preparing us to move beyond Vista?
- Why Google wanted to lose wireless spectrum auction
- iPhone shortage fuels rumors of imminent 3G phone
- XP for cheap PCs: a second crack in the wall
- Darts into data: Leveraging random action to competitive advantage
- Most iPhone buyers are existing Apple customers
- AT&T's so-called open network principles
- Mono dev tool offered
- ActiveState upgrades IDE
- Serena plans SaaS products
September 28, 2006 | Comments: (0)
MS antisphishing tool wins MS bakeoff
Microsoft sponsored a study comparing the effectiveness of antiphishing technologies and, surprise surprise, the company's IE 7 anti-phishing technology came out on top, according to a post on the IE Blog.
The study, which was conducted by 3Sharp, compared antiphishing toolbars from Microsoft, NetCraft, Google/Firefox, AOL, EarthLink, eBay, Geotrust, Netscape and McAfee. The study used a "standardized set of 100 known phishing Web site URLs and 500 known good URLs to see how well each anti-phishing technology flagged both phish and legitimate URLs."
Antiphishing toolbars were evaluated by "how well it did two things: warn or block the user from actual live phishing Web sites, and refrain from incorrect warnings or blocking on legitimate Web pages."
Toolbars were rated on a scale from 0 to 200, where "0" is the equivalent of having a browser with no antiphishing technology and 200 is a perfect antiphishing product that "caught all the known phish without making any mistakes by falsely warning or blocking any good URLs as phish."
Microsoft's Phishing Filter (MPF) in IE 7 Beta 3 received the highest "composite score" at 172, followed closely by NetCraft's toolbar with a composite score of 168.
But when you dig into the numbers, another story emerges. First of all, IE's MPF antiphishing toolbar doesn't top out any of the individual tests that make up the composite score. It finished second to GeoTrust's toolbar in spotting known phishing URLs (%89 percent catch rate, compared to %99 percent for GeoTrust). True, it didn't misidentify any known-good Web sites, but neither did five of the other toolbars tested.
So how did MPF end up on top? It boils down to how 3Sharp calculates that "composite score." As the group says, it intends the composite score to represent how good the phishing toolbar was compared with no protection at all. So the survey assigned greater points for blocking malicious sites than just warning about them, and more for warning about them than for doing nothing. On the flip side, blocking or warning on a good site cost more than doing nothing about a known good site.
So in the end, Microsoft didn't do the best job of spotting phish sites, but it did do the best job of blocking the ones it did spot, and blocking was what garnered the most points. In contrast, GeoTrust found almost all the phishing sites that were thrown at it, but doesn't have a blocking capability, and only warns users. GeoTrust also stepped in it, big time, when it came to false positives: a whopping %32, compared with just %1.6 for the next highest contender, EarthLink, and %0 for everyone else. That's outrageous!!
Moving on...NetCraft blocked all %84 percent of the sites it correctly identified, and that was better than Microsoft's %83 block rate, but Microsoft warned on another %6 percent that NetCraft didn't so...Microsoft WINS!!!!
Unfair? Possibly. Blocking a phishing Web site earned you twice as many points as just warning about it in this test, but is blocking really twice as effective as just warning users? That's reasearch that needs to be done. It certainly seems like 3Sharp's study may have been an outcome in search of a method -- but that wouldn't be anything new for vendor sponsored studies. After all, we can assume that if the numbers hadn't come out favoring IE7, we probably would never have seen this study. But hey, it's Microsoft's money.
But while we're at it, why not talk about the stinkiest antiphishing technology. According to 3Sharp, that honor belongs to McAfee's SiteAdvisor, which recorded a composite score of...hold your breath: "3." That's right: 3.
Shane Keats, of SiteAdvisor cries foul on that. "It's silly and wrong. We don't claim, anywhere, to offer phishing protection. In fact, we're pretty explicit that we don't."
True enough. SiteAdvisor is more of a malicious Web site detection service that can spot "fishy" rather than "phishing" Web sites -- developing a kind of reputation service for Web sites of all sorts -- will they misuse your personal inofrmation, have they distributed spyware or other malicious code, etc. Phishing and ID theft are part of that...kind of...but its just a piece.
Posted by Paul Roberts on September 28, 2006 09:55 AM
RATE THIS ARTICLE:
-

- COMMENTS
Paul, thanks for highlighting our study. A few points:
- We came up with the composite accuracy score model because it captures two important dimensions: "how good is product X at catching phish?" and "how good is product X at not interfering with legitimate sites?" This is explained in some detail on page 10 of the report.
- In the FAQ (and the accompanying podcast) that I posted this morning on my blog, I explained why we scored a block as worth 2x a warn. Preventing someone from loading a bad site is a better security decision than providing them a warning that they might miss or ignore. You can certainly argue that 2x is the wrong coefficient to use, but I suspect you'll find general agreement in the anti-phishing community that blocks are better for end users than warns alone.
- You didn't point out that GeoTrust's TrustWatch was heavily penalized for generating incorrect warnings on 32.3% of the 500 URLs we tested with. This was far worse than any other product; 6 of the 8 products didn't make *any* mistakes with the known good URLs.
- McAfee certainly *does* position SiteAdvisor as having anti-phishing functionality. See http://www.robichaux.net/blog/2006/09/mcafee_siteadvisor_sure_looks_like_an_an.php for a detailed list. Heck, even the product's own FAQ says it does anti-phishing.
Posted by: Paul Robichaux at September 28, 2006 11:23 AM"Microsoft sponsored a study comparing the effectiveness of antiphishing technologies and, surprise surprise, the company's IE 7 anti-phishing technology came out on top, according to a post on the IE Blog."
Ahahah.
Did MS also choose the 100 + 500 URL?
Posted by: Maps at September 29, 2006 11:53 PMThe CMU Usable Privacy and Security Lab conducted a similar study, but we didn't test the Microsoft toolbar. See http://lorrie.cranor.org/pubs/toolbars.html
Posted by: Lorrie Cranor at September 30, 2006 02:34 PMMaps: no, 3Sharp chose the 100 phish and the 500 known-good URLs.
Dr. Cranor: thanks for the pointer to your study; I didn't realize that it existed. Makes for interesting reading!
Posted by: Paul Robichaux at October 2, 2006 01:18 PMI recently (yesterday) signed up for McAfee Siteadvisor (and am also a tradedoubler affiliate for McAfee products though am yet to market any) I myself was a little concerned at some of the obviously impartial reviewing posts on siteadvisor and was looking for the best,most impartial and professional advisor to recomend to customers and to use to professionaly clear my own site of recent insinuations by a supposed jealous music review site online-preferably a trustworthy free to use service as we operate on a non-profit basis and plough back any revenue into supporting indie artists and musicians. Though McAfee is certainly helpful at times it isn't allways possible to guage just how qualified some of it's reviewers realy are .
Posted by: Mairtin O'Riain at January 24, 2008 05:49 AMTOP STORIES
ADDITIONAL RESOURCES

- Remote Access: Maintain Security and Decrease the Burden on IT
- Beyond AntiVirus: Symantec Endpoint Protection
- What Every Enterprise Needs to Know About VDI

- Disaster Recovery in Minutes
- Protecting Microsoft(R) Applications
- Reduce Recovery Times and Tape Costs





