AlyssaS - 9:16 am on Jul 14, 2011 (gmt 0)
After reading this thread, I went into the hubpages forum (which anyone can look at) and there is a thread about this from Paul Edmonson
They've been testing a number of volunteer accounts, not just his account. Some of the testers have posted up graphs of their experience
They're using rel=author to distinguish different authors on different subdomains.
Edmonson made this comment in response to someone who asked about subdomains based on topic rather than username:
After much deliberation we've have chosen a path that aligns subdomains with usernames as closely as possible. This move is about letting high quality content get relief from Panda and segmenting content in a clear way for Google to judge each users Hubs. Your main subdomain page will replace your profile page.
I'd encourage people to think of their presence on HubPages as a personal brand similar to a twitter handle.
Therefore note they are NOT isolating based on topic, but on author - and most authors have a myriad of topics in their account. In other words they are betting that it is the quality of prose that is the issue, rather than the fact that their sites covers a massive range of topics.
They are 301-ing from the original pages so people don't lose their backlinks.
It will be interesting to see what happens when Panda does it's next run. It could work, as Hubpages was heavily spammed with the sort of badly spun junk that was literally un-readable (I mean serious junk) and they had lost control in that they weren't able to take down the stuff fast enough. That probably sank the site. So isolating the good authors probably helps a lot.
I'm really not sure how the mixing up of topics on each subdomain will work out. When people make blogger blogs, they usually write about one topic, say Harry Potter, rather than mixing in anything that comes into their head.
This could be a solution for Pandalised sites with UGC content - not sure it will help other sites. If you make the same mistakes in terms of content and layout, your new subdomain is likely to get pandalised again on the re-run.
But the hubpages experiment is very interesting nevertheless.