http://www.sistrix.com/blog/985-google-farmer-update-quest-f... and http://searchengineland.com/who-lost-in-googles-farmer-algor... are other links I've seen. Bear in mind that much of this third-party analysis compares e.g. US queries vs. queries against Google from Canada, Italy, or India, and geolocation can change the results. Also, different people are running different sets of queries and that subsampling can skew things depending on the query sets. Please bear those disclaimers in mind with any third-party analysis.
Matt, I've seen some high quality smaller sites (Internet versions of traditionally published books with original content) drop as well. Is there any place to submit such sites in order to help this algorithm improve?
Looking through this list of larger sites that dropped, I see one that is similar to the sites I'm talking about, so more people might be able to say what happened: findarticles.com. I have no relation to this site (it is actually a competitor), but I believe the vast majority of its content is licensed versions of articles published by real books, magazines, and newspapers (http://findarticles.com/p/articles/an_1/?browse=A&tag=co...). It seems to have dropped just as much as content mills with low quality articles, such as ezinearticles.com or articlesbase.com. It is possible that I don't know something about findarticles, but this makes it appear like this algorithm change is looking at some superficial common factors and is unable to really distinguish high quality sites from low quality sites.
> licensed versions of articles published by real books, magazines, and newspapers
That makes it sound like Find Articles is quite likely to have a lot of duplicate content, one of the things that Google's update was meant to penalize:
> This update is designed to reduce rankings for low-quality sites—sites which are low-value add for users, copy content from other websites or sites that are just not very useful.
Maybe, but this quote talks about punishing sites copying content from other websites, which is not what they are doing. Most newspaper sites will have the same versions of AP or Reuters articles as well, but this new algorithm didn't affect them. I believe that many older newspaper articles on their site are not republished anywhere else on the Web.
Matt, part of our site is a blog-like service. We'd love to have a Google service where we submit the content our users publishing and reject them based on the rating returned.
I'm sure this can also be used for evil purposes by content scrapers, but as a startup with limited resources, an automated (and free :) way to stop such content would be great.
Some high quality user generated content sites seem to have been impacted by this change too. My main site (see my about for details) has lots great car reviews, which are unique to the site, yet my Google US traffic has fallen off a cliff.
If my site is a content farm, then surely so are sites like Stack Overflow and Trip Advisor, as I'm using the same model of moderated and curated user generated content, and while I don't think my site is quite as useful as Stack Overflow (what is?), I've been running this site as a labour of love since 1997, and I've had countless emails from people who've found the site useful, so I must be doing something right.
We tend not to pre-announce ranking changes, because priorities and timing can always shift. But broadly speaking, anything I mentioned in my blog post a few weeks ago at http://googleblog.blogspot.com/2011/01/google-search-and-sea... is still open for improvement. The topics of that post included scrapers/copiers, spam, and low-quality sites.
