[Home] [Headlines] [Latest Articles] [Latest Comments] [Post] [Sign-in] [Mail] [Setup] [Help]
Status: Not Logged In; Sign In
|
Science/Tech See other Science/Tech Articles Title: Google and its "safe browsing" database About 1.2 percent of all scrapes of Google's results done by Scroogle show at least one "safe browsing" interception by Google. This is consistent with the 1.3 percent figure in Google's own report, "All Your iFRAMEs Point to Us," dated February 4, 2008. The way that Google handles these interceptions is by prefacing the link in their search results with www.google.com/interstitial, which sends the searcher to Google's page for more information. On Google's results page itself, it identifies such listings with the words, "This site may harm your computer." Scroogle has always respected this format for such links, and now we also show "Google intercepts this link" next to these search results. We could have ignored these links and stripped the interstitial from the URL, on the grounds that there is not a high correlation between Google's "safe browsing" database and other similar databases. It really depends on how Google chooses to define the word "unsafe," just as the ranking order of any search engine depends on how it defines the word "relevant." There are too many variables between the user's browser and how it interacts with the web, and the numerous techniques available to webmasters, advertisers, and con artists. How good is Google's quality control with this database? On January 31, 2009 Google accidentally labeled every single link as "harmful" for nearly an hour. That raised a lot of eyebrows. Later we saw an item at Stopbadware.org that was posted by manager Maxim Weinstein on January 20, 2009. He said that Google was now reporting 183,000 badware sites, whereas it was 145,000 "a couple months ago." He had no idea why: "Google has been known to tweak its systems, sometimes leading to a significant increase or decrease of reported hosts without any change in external conditions." For anyone like us who has had close experience with Google's search-engine rankings since 2001 or so, it's no secret that massive fluctuations can occur even when everything external to Google is fairly stable. Now we know that this is also true of the "safe browsing" database. If Google marks a site as one that "may harm your computer," at least you can get details on the reasons behind their analysis at www.google.com/safebrowsing/diagnostic?site=www.example.com (substitute www.example.com with the URL of the target site). And Google, in conjunction with Stopbadware.org, also has a procedure that allows webmasters to appeal and get their site unlisted. That's the good news. The bad news is that the quality control issue is bothersome. Google also seems a bit stingy with its database. It's easy to find out about a site if you already have one in mind, but as far as we can determine, it's not easy to get a useful random sample of Google's current listing of malware sites. This makes it difficult for independent researchers to evaluate Google's quality control. Our guess is that Google doesn't want competitors to grab major chunks of its "safe browsing" database. Someday it might be profitable to license access to this data to other companies. On an issue as important as malware, the public interest demands that this information be shared openly, and evaluated openly by independent researchers. If Google disagrees with this notion, then it might be better if they discontinue their research. Their motives are suspicious. Why should they bother with all this research on harmful sites, when adding the option to disable JavaScript in the Chrome browser — an option that's always been in all other browsers — would probably do more for safe browsing in the long run than their entire database will ever do? For these reasons, Scroogle is recording the first instance of an interstitial URL on every search that produces one, and is making this data available for download. We are doing this in small chunks of one hour at a time. If you want to build up a list for your own use, you can download this page once per hour around the clock. At the same minute of each hour our latest list overwrites the previous list, and we do not archive old lists. This keeps our data current, and it's also the easiest way for us to do this. Anyone who is experienced enough to handle serious malware research will find it trivial to automate these small downloads and combine, sort, purge duplicates, and parse the URLs into something useful.
Post Comment Private Reply Ignore Thread Top • Page Up • Full Thread • Page Down • Bottom/Latest Begin Trace Mode for Comment # 1.
#1. To: X-15 (#0)
goodsearch.com searches for me, and donates to the Salvation Army at the same time.
There are no replies to Comment # 1. End Trace Mode for Comment # 1.
Top • Page Up • Full Thread • Page Down • Bottom/Latest |
||
|
[Home]
[Headlines]
[Latest Articles]
[Latest Comments]
[Post]
[Sign-in]
[Mail]
[Setup]
[Help]
|
||