Google could give better search results if pages use link rel=canonical

[Paul Sladen]

Launchpad's Google-fu is sometimes reduced as of late, often failing to appear on the frontpage. This is likely due to the incredibly-large number of alternative URLs available for the same content, giving the appearance of a content farm. Introducing rel="canonical" on all bug report pages in the form:

  <link rel="canonical" href="https://launchpad.net/bugs/NNNN"/>

should hopefully mitigate this, and provide cleaner, shorter search result snippets at the same time. This also provides protection against the project that the bug report is allotted to changing:

  'About rel="canonical"'
  http://www.google.com/support/webmasters/bin/answer.py?answer=139394

[Martin Pool (mbp)]

Hi Paul,

For the sake of describing problems rather than possible solutions,
can you give some examples of searches that ought to find Launchpad
pages but don't?

[Paul Sladen (sladen)]

I had a search this morning where Launchpad didn't appear in the first page (should have taken a screenshot), but that could have been Google's monthly shuffles.

For the all the trial searches I've done now, LP is appearing in the top-three. I do however frequently search for people's names in connection with Launchpad and those do tend to be lowly-rated from the search engine requiring explicit hand-hacking to get to.

[Martin Pool (mbp)]

Using canonical URLs for each page could also help avoid duplicated
items in search results.

[Martin Pool (mbp)]

see also bug 869609

[Martin Pool (mbp)]

This could be arguably more elegantly solved by getting rid of the per-context bug pages, but that's much larger.

Another case, suggested by that talk, is that we could use rel=canonical from edge etc to lpnet, but edge is blocked from robots, and generally speaking not turning up so much. I think perhaps that can be done through the google webmaster console.

I'm going to add this for bugtasks, the specific case suggested, and add a mechanism for it, and then we can consider others as new bugs when they come up.

[William Grant (wgrant)]

edge no longer serves webapp pages -- it redirects to production.

[Martin Pool (mbp)]

On 20 November 2011 20:58, William Grant <email address hidden> wrote:
> edge no longer serves webapp pages -- it redirects to production.

That is good. The same thing applies to eg staging, but since staging
is much less used than edge was it's not a major problem. At any rate
it can wait until there's evidence it is a problem.

--
Martin

[Launchpad QA Bot (lpqabot)]

Fixed in stable r14382 <http://bazaar.launchpad.net/~launchpad-pqm/launchpad/stable/revision/14382>.

[Martin Pool (mbp)]

qa: view-source:https://bugs.qastaging.launchpad.net/launchpad/+bug/808282 looks good to me.

[Martin Pool (mbp)]

@paul if you find more cases where google (or bing etc) returns multiple results for what are really the same page, please file a new bug pointing back to this one. thanks.