It was interesting watching Bing and Google squabble like fashionistas over the last purse on sale yesterday. Each claiming to have the high ground. And, as always the usual interest and excitement over a bit of drama slowly gave way to the ‘so what’s’ as the day went on, and so what would be right, as much of a distraction as it was, it really had very little relevance to actually making money. Until you look at it from a slightly different angle.
There are two possible scenarios, according to yesterday’s news, for ranking in Bing. The first and considerably less interesting is that Bing is copying Google’s search results. This is a nice easy one, because it means that there is potentially one less search engine to optimize for.
The second scenario is the one that caught my interest though, allowing me to entertain my grey side for an evening as I went over the possibilities. If we believe Bings version of events, Google’s sting results were due to Bing seeing people, using its toolbar, visiting those sites from a certain referrer string, and so (quite logically) determined that those sites were relevant to that query.
What if we applied this in situations where the site really was relevant? If we follow the logic to its conclusion, enough people performing a search, with the tool bar enabled, and clicking through to a single site, would cause that site to improve in rankings on Bing. How simple, to be able to hire hundreds of low cost workers to simply install the Bing toolbar, search for your head term on Google, scroll through as many pages as it took to find your page, and click. Suddenly Bing is swamped with information suggesting that your site is the one most relevant to the term.
Google’s sting results would also suggest that Bing may not be analyzing these results for relevance, so would the site with the most clicks win? It’s impossible to say how big an impact this would have on a competitive term, weighted with all of the other elements of the algorithm, in fact this is possibly as close to a test of a single factor as could be achieved anywhere, but it’s certainly an interesting area for further investigation.
This is such an old trick I’m not sure Bing could really be so naive as to include something so gameable in their algorithm, but, if it doesn’t work they show themselves as having copied Google, if it does, they have shown they have a long way to go before they truly understand what it’s going to take for them to be a viable competitor.
Now that is interesting. Nice article, Sarah. I wonder if it wouldn’t be in Google’s interest to systematically train the Bing Toolbar to find irrelevant results, so that Bing’s “decision engine” becomes useless on important search terms. I mean, according to your “grey side” they’ve not only left themselves open to being played, but to being played by their biggest competitor. I bet they’d stop copying search results if it started messing up their “decision engine”.
It would certainly be an interesting move for Google to make, but I can’t imagine at this point they would want to do anything that would take away their ability to cry foul any further, and playing that sort of dirty game wouldn’t gain them any fans from anywhere
Ha it’s nice to see talking to me for a while appears to have rubbed off on you!
I think that whilst I’d love that to work I think the cost of the outsourcers v the cost of buying links to support the site might not add up. Having said that running a test with mechanical Turk (or crowdflower for a better api) might be an amusing distraction.
I’m off to set a test…
Yes, apparently you way of thinking is contagious! You have a good point re cost, although the original Google test didn’t seem to use many clicks at all, I guess it depends how much of a shift would need to be seen in order for it to trigger. Can’t wait to see if your testing brings any results!
Sarah is a pretty much a SEO genius. Love the post. Thanks!
I would guess that the possibilities for gaming Bing’s algorithm in this way are limited to very obscure long tail searches in which Bing is less able to rely on the other elements of its algorithm. So it probably has a limited application. But for those who can find the right niche …
Sarah-
Completely agree with you. Bing is in a losing catch 22. And let’s hope they’re not going with the clicks- very gameable!
since google has become the new ‘micosoft’ now, i think a little (or big) competition from Bing would be healthy, especially when it was stated here that “Bing is copying Google’s search results” and me thinks Bing is using google’s old (and easily manipulated) algorithm of which the one with the most clicks get to be on top of the list.
Re: “If we follow the logic to its conclusion, enough people performing a search, with the tool bar enabled, and clicking through to a single site, would cause that site to improve in rankings on Bing..”
Google Instant, which is driven by Google Trends, is subject to exactly that spam attack. Sometimes a long search string for a product will appear near the top of Google Trends for 15 to 45 minutes. Google had to average trends over a longer period to stop such momentary spams.
Most “crowdsourcing” systems are subject to such spam attacks. Recommendation systems are very vulnerable. Yelp, Yahoo, and CitySearch are full of obviously bogus reviews. Google Places gets hit by this constantly. From October 2010 to January 2011, Google seemed to use recommendation counts for ranking. The spam there got completely out of control. Offshore services and attack software were posting phony recommendations. Publicity in the New York Observer finally made it just too embarrassing for Google, and now, high recommendation counts no longer seem to improve rankings.
This is why Blekko’s “slashtags” won’t work for long. If Blekko gets enough market share to attract SEO attention, it will be spammed.
The recommendation systems that work are ones where the recommendation is known to come from an actual customer. eBay and Amazon, which process transactions and know who bought what, do that.
What also works is looking at the real-world business behind the web site. Blekko does that manually for a small number of businesses, filtering out content farms and whitelisting legitimate health-related sites. We (as SiteTruth) do that on a larger scale, using SEC files, Dun and Bradstreet ratings, and similar hard data.
Re: “This is why Blekko’s “slashtags” won’t work for long. If Blekko gets enough market share to attract SEO attention, it will be spammed. ”
This is not correct. Blekko slashtags resist spam in several ways. First off, you have to be invited to edit a slashtag, and we aren’t inviting anyone to help us edit slashtags like /health, which only chnage slowly and re very commercially valuable. Second, even on more general slashtags which have a lot of editors, we expect the system to be self-policing much like Wikipedia is.
In the early days of Gmail, one had to be invited. One once had to have an email account at a major university to get onto Facebook. As those scaled up, the restraints were removed, and the quality went down. Blekko will face the same problem.
Expecting the Blekko system to be be “self-policing like Wikipedia is” is naive. It takes a lot of organization, hard volunteer labor, elaborate procedures, and extensive discussions to make Wikipedia work. Blekko doesn’t have any of that. “Crowdsourced search” has been tried. Remember Wikia Search?
Also, for a profit-making company, you’d have to pay those people. Not only is it tough to get people to work for free for a profit-making company, it’s illegal. See U.S. Department of Labor vs. AOL.