Yahoo! Search sorts out redirects
Yahoo! Search came clean last week at the SES Conference in Chicago. They finally admitted to not updating the index when their spider encountered a 301 redirect. They said something to the effect of: “We were following the redirect, but indexing the source instead of the destination.” Any logical person, or spider in this case, would index the destination, but since this is Yahoo! Search’s first stab at actually running a search engine, we’ll let ‘em slide.
Not only did Yahoo! Search admit to this, but they then proceeded to inform us that they fixed it and there was much rejoicing. There was a catch, however; instead of just following the redirects and updating their index, they threw a monkey wrench into the way that they handle them. Here’s a rundown of what they said:
If you use a meta refresh to redirect your pages, Yahoo! will follow it. If the time is greater than 1 second, then the redirect will be treated as a 302 Temporary Redirect. If it is less than or equal to 1 second, then Yahoo! will count it as a 301 Permanent Redirect. Please note that every other major search engine does not follow this rule, and meta refreshes are still a bad idea.
A 301 from one domain to another will drop the old page and put the new one in. This is good; it works like every other search engine.
A 302 from one domain to another will keep the old page and not put the new page in. Once again, like every other search engine. Good job, Yahoo!
A 301 from one page to another in the same domain will drop the old page and put the new one in the index, unless it is the homepage. If the homepage 301 redirects to another page on the same domain, then the Yahoo! index will keep the homepage and not count the new page. Weird, but sensible.
A 302 from one page to another in the same domain will keep the old page and not put the new one in, much like it does for pages between domains. Again, excellent.
Once you sort it out, these approaches makes a little more sense then most redirect handling, except for the meta refresh thing. I’m still trying to figure out why count those.