User:Rossami/On redirects

Recently, another editor asked me to explain the reasoning behind some of my comments at WP:RFD. Specifically, he/she asked "how much time is needed for a redirect to be considered sufficiently old to be useful, or vice versa (sufficiently new to be considered fit for deletion)?" I got a little long-winded and tried to answer a number of implied questions about the history of our redirect policy and traditions. That answer is pasted below.


 * You asked specifically about age as an indicator. I was taught that Wikipedia values history and that we ought to keep the vast majority of good-faith redirects regardless of age.
 * If a redirect documents a pagemove (and many of them do), we should always keep the redirect (unless it was pagemove vandalism) because:
 * the page title itself is content and GFDL requires that we preserve the attribution history of content changes
 * Yes, you can carry out a history merge or document the change on the article's Talk page but most times, the redirect is easier and more reliable.
 * editors returning to the page need to find their contributions. Content that's been moved can easily get lost.  When new editors can't find the correct page for their content, they frequently assume that it was some mysterious computer glitch, say bad things about Wikipedia and add the content back - only to have it deleted yet again and still without explanation or, worse, with a hostile note.  Redirects point them to where their contribution should have been made in the first place and tend to preempt these misunderstandings.
 * Redirects also preempt future editors from making the same mistake and creating a page or list that the community has already decided is inappropriate.
 * The history of the redirect often holds clues as to the reason for the move to the current title. This allows others of find and review that decision and to see if the original reasoning still holds. Or, more often, to agree with the decision and save everybody the trouble of re-running the whole debate.
 * Many of the pagemove redirects predate the creation of the various namespaces or naming conventions.
 * Redirects help readers find content that they've looked at/worked on before.
 * Some have argued that the search engine is sufficient to help you find your contributions. Unfortunately, we know that is not true.  The search engine has some inherent weaknesses that we have never been able to overcome.  I still remember when Stan Shebs challenged me to evaluate the search engine results for "142 AD".  The search engine failed to return many relevant pages (and also returned a high proportion of false positives).
 * Others have argued that the watchlist is sufficient to help editors find their prior contributions. Not everyone uses the watchlist (and some misuse it, watchlisting every page they've ever edited so they can't ever find anything useful).
 * Even if the editors can find where their contributions went, people who merely read the prior version may not be able to. They are far less likely to have watchlisted the page.  We just end up with frustrated readers who are sure that the article used to be there and, again, conclude that Wikipedia must be unstable.
 * Even if you update every inbound link to the redirect title, the old inbound links can still exist in the history of other pages. Any of those other pages may have to be reverted (for example, to clean up vandalism), restoring the old link. In a perfect world, the reverting editor would find the intermediate edits and again repair the link. In the real world, that step is often overlooked.
 * Worse, we have no reliable way to determine who outside the project has created a link to a particular page. If you delete a redirect, you may be deleting a link that someone else depended upon.  We want Wikipedia to be a reliable reference for our readers.  Breaking the links in their papers and cross-references will not endear us to them.  (Yes, google Advanced Search has a links search feature but no, it doesn't work well enough.  Tests have shown that far too many inbound links are not found.  It will not, for example, find any off-line links.)
 * Deleting a harmless redirect does not actually give anything back to the project. We still keep the deleted versions of pages in hidden history so it's still consuming server space.  Redirects don't cost the project anything in response time.  In fact, the act of deletion adds a record to the logs.  To the extent that you care about system resources at all, harmless redirects should be ignored, not deleted.  (See  below.)


 * That last point is a factor that I believe is widely misunderstood. Many people think they are doing something good for the project by "cleaning up the unnecessary redirects".  Even if it does no harm, it has no benefit either.  At the risk of sounding trite, redirects really are that cheap.


 * Re-reading your question, I now wonder if I may have misread your intent. I answered the question "do old links ever stop being useful" to which I believe the answer is "no".  Were you instead trying to ask "when do new links start becoming useful"?  If so, the policy implies "immediately".  My presumption is that the editor would not have created the redirect if he/she didn't think it was useful.  Remember that not everyone navigates through pages the way you do.  What's useful to you could be completely obscure to me - and vice versa.  Links that are made in bad-faith or which are potentially confusing or harmful should be deleted.  But links which are "useless" strikes me as an unmanageable deletion criterion.  Since we can't know how everyone uses the encyclopedia, we can never prove the negative.


 * All that said, redirects that are younger than a few months (that were not created as a result of a pagemove) are unlikely to be in high use in my opinion. When the pagehistory shows that to be the case, I almost always support the deletion nomination (though I rarely bother to comment on those since I have nothing to add to the discussion).  Cross-namespace redirects created since the creation of the different namespaces can usually be cleaned up without confusion - though in the case of redirects to policy pages, I rarely see the point.  It's awfully hard to confuse "Wikipedia Deletion Policy" with anything but the project's policy page.


 * Those are a few of my thoughts on redirects. I hope some of it is a little bit helpful.  Rossami (talk) 13:12, 22 September 2007 (UTC)

Additional resources

 * Tools/Navigation popups/About fixing redirects has some interesting statistics on why "fixing" redirects solely for the sake of changing them is not an efficient use of Wikipedia resources.