Alexa confirms that this wasn't something isolated to a few websites, but rather a change in the search results shown by Google that is shifting traffic across the web. The exact nature of the change is still under investigation, but it is possible that Google made an Algorithmic change in how they rank search results. This is very serious for many sites.
This shifting of web traffic is best illustrated by looking at the daily Reach graphs for different sites. A website's reach is determined by the number of unique visitors to a site, shown as a percentage of the total number of people on the Internet that day.
First, here is site that was showing nice growth over the past three months, and then took a sudden hit in traffic between the 21st and 22nd. I've only shown one example here, but this is happening across the web.
Search traffic isn't quite a zero sum game, but it's close and some sites have experienced a significant increase in visitors due to the Google change. Among the biggest winners, oddly enough, are file sharing and torrent sites. It's tempting to speculate why Google might be sending more visitors to file sharing sites, but it's still too early in our analysis to say anything definitive.
So far the changes in traffic to affected sites have been sustained, so this appears to be a deliberate change on the part of Google instead of a transient glitch in their system. There were some indexing issues that happened prior to the change that caused some sites to report problems as early as the 19th, but according to our data the change to Google's search results went live on October 21st around 3PM Mountain View time. The change may be related to the "Mayday" change, but given how quickly web wide traffic shifted it seems unlikely this was something that had been slowly percolating through Google's indexing system over the past six months.
Our analysis of the October 21st event is ongoing, and I will update this blog as we uncover more information. I am personally checking one site's update about Alexa rank.