Souring the Honeypot

Brian in Opinion | 2 Comments February 11, 2011

The web is all a twitter (pun intended) over claims by Google that Bing is stealing their search engine results.  I won’t re-hash it – this link has plenty of reading material.  Google’s honeypot trap got me thinking.  What if, after installing the Bing bar and accepting all the “would you like us to track your links to improve Bing” options like Google did, you clicked on incorrect links to make the wrong associations?  Would Bing learn those associations?

Let’s look at Google’s example.  Let’s say the Google engineers typed one of their nonsense words like “hiybbprqag” into Google with the Bing bar installed, but instead of clicking a results link they scrolled to the bottom of the page and clicked the “About Google” link.  After a few weeks of doing this, if you typed the nonsense search term into Bing what would you get?  My suspicion is that Bing would deliver Google’s about page as its #1 search result.  Of course this is impossible to test with this particular term because it’s now all over the internet, but anyone with a week or so of extra time on their hands could try the same experiment with their own nonsense term.  They don’t even need a honeypot page in Google search results – they already have the About Google page.

If you tried this same trick in the Chrome, or the Google or Yahoo toolbars, would Google and Yahoo search results end up delivering the about page too? 

I think all of these toolbars do the same thing:  they create a feedback loop between the stuff you type, and the stuff you ultimately click on, and use that data to improve the search index.  This happens on all pages, strengthening links based on what people actually choose.  If you happen to be clicking on Google search results with the Bing bar installed, Bing learns what you chose as the most relevant from Google’s index.  If you are using Bing in Chrome, Google adjusts its index based on Bing results.  Just don’t do both – the internet might crash in an infinite loop.

Comments (2) -

Not Important, Wednesday, February 23, 2011 at 2:13 PM

Hello! And welcome to 2009!

Over the last two years I noticed how Wikipedia steadily moves up in the search results I get from Google. I am assuming this is because most of the results I click on are from Wikipedia. This kept happening even though in interim I erased cookies on my computer on multiple occasions and switched to a new browser.

Google probably keeps tabs on my computer somehow (some unique identifier on my network card maybe?). The feedback loop you mention probably takes place in Google's data centers, not on my computer.

Brian, Monday, February 28, 2011 at 11:19 AM

I agree -- although I bet you're not the only one clicking on Wikipedia results.  Google's data centers are learning from all of us, as are Bing's and all other search engines.  It's only interesting when you combine a search bar from one company with results from another engine, so you're teaching one engine about the other's results (or, more accurately, your click preferences about the other engine's results).  I'm surprised Google has made such a stink about this, since it's obvious what's going on.
Comments are closed