tedster - 3:14 am on Jan 11, 2010 (gmt 0)

One improvement I can envision for the Suggestions algorithm would be to require a relatively high threshold of content for the [company scam] search, figured as a percentage of the [company] search results. That should be measured as rather high before the Suggested word combination gets rolled out to the search boxes across the world.

I have a suspicion that, just a few weeks back, something like a raised threshold DID happen for "Searches related to" -- but not for Suggestions. At that time some "Searches related to" links that I was puzzling over for a few companies just vanished.

However, those same Suggestions are still showing up right now, and I wish they would be zapped as well. I mean, what can you do about a few UGC pages that say something like "I'm sure there were some scam accusations about Company, but they've been hushed up." In the meantime, we're cranking out some new content so that the companies involved can at least show up in the results with their point of view -- if someone does click on that particular Suggestion.

The problem I see is a kind of subliminal contamination of reputation that occurs just from seeing the derogatory word coupled with the Company Name - just the "suggestion" registered in peripheral vision that there is a scam associated with Company. And that's especially problematic when the user doesn't even click to see that there's no fire behind that suggestion of smoke.