Now more evil than ever
February 20, 2005
It's always been evil. For starters, Google's toolbar phones home with every page you land on, and tattles to Google what you're looking at, while offering up your globally-unique Google cookie ID that expires in 2038.
As if that wasn't enough, the toolbar updates automatically, without asking. Once installed, this happens no matter how restrictive your Explorer security settings might be, and regardless of whether you have ZoneAlarm or some other firewall installed. Basically, this means that Google potentially can install anything at all on your hard drive, at any time that you connect to them. If you have broadband, it happens so fast that you won't even notice. One day you may see some new features on your toolbar, and then you'll realize that something happened.
Now Google has introduced a version 3.0 of the toolbar. It is still in beta, so this one won't be auto-updating for another two months or so. This new toolbar introduces AutoLink, which sounds friendly when Google describes it:
The online review of a great new restaurant has the place's address but no map. You could type the restaurant's street, city, and ZIP code into the search box, but why bother, when clicking the Toolbar's AutoLink button will automatically create a link to an online map (US addresses only)? AutoLink can also link package tracking numbers to delivery status, VIN numbers (US) to vehicle history, and publication ISBN numbers to Amazon.com listings.The reason this has webmasters upset is because Google is essentially changing their pages by adding links where links did not exist before. These links take the toolbar user off of the page. Potentially they are a new source of revenue for Google, because the same technology can be used by Google to link to ads.
Anyone who thinks that Google wouldn't use this to link to ads hasn't studied what Google is doing with Gmail. But there is a bit of a problem with changing a webmaster's page, because altering the content and/or the linking of a copyrighted page would invite lawsuits. Therefore, Google is going more slowly for now, to see if they can start off on this particular slippery slope without too much clamor from webmasters and bloggers. Who can object to more information about a book, for example?
Load up your new Google toolbar and go to Barnes & Noble.
Search for a book and go deep enough so that you see the price
and the ISBN. Google's toolbar now says "Show book info" because
it detected an ISBN on the page. When you click on "Show book info"
the ISBN is highlighted. Click on the highlight and you end up at
Amazon's page for the exact same book, with Amazon's price for that
book. Convenient for users? Maybe. A threat to B&N? Definitely.
There was a uproar about this when Microsoft tried to introduce SmartTags back in 2001. This was a "feature" that placed a squiggly line under words on a page that Microsoft selected, and added a link underneath that took you to Microsoft. SmartTags was withdrawn before it could be introduced into a production version of Explorer. The person at Microsoft who was behind this abomination, Jeff Reynar, now works at Google (evil knows no boundaries). Even Microsoft came up with a page header tag for webmasters to disable this "feature," which is something that Google still hasn't bothered to do.
Barnes & Noble is no doubt consulting with their attorneys already. Fortunately, an existing link behind an ISBN will override Google's new link (adding links to unlinked text is one thing, but overriding the webmaster's own links would be too arrogant even for Google, Inc.). This behavior of allowing the original link to stand provides a way to shield your pages from surfers who use Google's toolbar.
Webmasters can construct a link which does nothing except stop Google. It usually does not change the appearance of the text, and it breaks the toolbar's ability to form its own link. The toolbar will still highlight it, but nothing at all will happen when the user clicks on the highlight. This link, for those who need a "toolbar shield," is simple:
<A>0-14-004007-2</A>
Addresses will work as well as numbers. You are defining a null anchor and turning the text into anchor text. The link is invalid so it doesn't appear as a link, but it's valid enough to keep Google from hijacking it. Google will still recognize it as "hot" and will highlight it when the user clicks on "Show book info," but nothing at all happens when the user clicks on the highlight.
(On pages where the number appeared inside of tables or other types of complex coding, it was blue and underscored like other links, even though the link remained dead. If this is true on your pages, you can force the font color inside the anchor, but the underscore might be a problem. If you do any testing, you should know that the last digit of all ISBNs is a check digit. If you change any digit, then it won't compute and the toolbar won't recognize it as an ISBN.)
The user will think that Google's toolbar is broken. That, of
course, would only be wishful thinking. But it's better than
letting evil Google guide your visitors into the deep Amazon, from
whence they may never return.
|
Top Secret leak from the Googleplex: Our game plan for mission creep on link hijacking 1. Currently with the toolbar you can search on Google, Yahoo, or MSN and include the keyword isbn with one or more keywords from the author or title of a book. Out pops the ISBN number in the snippet, for easy linking straight to Amazon, or wherever we decide the user needs to go. This last point is crucial. Cut a deal with Amazon to publicize this particular user convenience, under threat of switching our redirect from Amazon to Barnes & Noble. Amazon must carry our water on this because it would be embarrassing if we pushed it. In the meantime we can still emphasize that we're not receiving any money from Amazon. 2. Once the uproar over the toolbar subsides, we have to migrate the link hijacking to those who don't use our toolbar. The first step is to serve up our cache copies with our own linking inserted into the copy. We already highlight the search terms, and we already have an opt-out NOARCHIVE option for webmasters, so this move should be easy. The fact that this opt-out is required on every page on that webmaster's site means that few use it. Rumors that our indexing algorithms are less enthusiastic about such pages can be encouraged by planting posts on various forums. 3. For years we've shown maps to a person's house when their phone number is placed in the search box, and only a few pundits were upset. All we had to do is point out that this information is available from other reverse directories and map makers. Besides, we already have an opt-out, which conveniently ignores the fact that the same number embedded in web pages is still produced by searches. These pages are often more interesting because only web search engines can find this information. The next step is to determine whether any of the snippets for any search result have full names or telephone numbers, so that we can highlight these and link to maps and/or another search. Lots of local ads will fit on one of our map pages. 4. After the above three have been implemented, it won't even raise an eyebrow if we start linking keywords inside of Gmail text. If those privacy advocates get upset again, we can just open up Gmail to everyone, and the stampede to sign up for Gmail will drown out the critics. Notice how we've already stopped deleting Gmail in the trash folder after 30 days, now that the critics have dispersed. 
5. The Gmail Social Networking Team that's working on that
Echelon contract says that we cannot hope to keep the Gmail
invitation scheme going forever. We have to find other ways to
harvest email addresses. Google Alerts has given us tens of thousands
of additional addresses, plus excellent profiling information. But
our competition is using RSS in this area, which is much more
efficient and much more private. Sooner or later both the Gmail
invitation scheme and Google Alerts will be too obvious. If and
when we are pressured to provide opt-outs for controversial features
of any type, we should require an email address like we used to for the
Google phone number removal.
6. Already when our crawlers see a robots.txt disallow for directories, and we know from external links the URLs of pages in those disallowed directories, we go ahead and list this URL in our index. This is one fat foot in the door, and only a couple of posts on a couple of forums have even brought up the issue. Now is the time to sneak the other foot in. Stop honoring the disallows in robots.txt, but do it over a period of two years by tracking it on a per-site basis. Webmasters won't know when to jump, like some idiot frog sitting in a pan of water heating up on the stove. In two years we'll have a much more compelling index than Yahoo or MSN. Provide an opt-out form that requires an email address from webmasters who object. 
Keep our billions coming, GoogleGuys and GoogleGirls! |
