I have a feature request that I'd first like to put on the mailing list, before putting it at SourceForge.
When I use the search feature, I am frequently annoyed by the large number of pages showing up that are simply redirects. When I type Zimbabwe, I might also be interested in "History of Zimbabwe", which shows up. I am, however, not interested in the redirect "Zimbabwe/History", which is a previous location of that same article. I know it is important to keep that redirect for linking (both internal and external), but it is not useful for searching. The same holds for all other pages that were moved because of some naming convention or change in policy, or moved across namespaces (as a side note, I'd like to be able to search in Wikipedia: namespace articles as well, there are getting too much of these pages).
My proposal is to make two kinds of redirects. The first type is the type we currently have, the second one is different in that it still redirects those visiting it, but the page does not show up in any searches. I think this can be implemented relatively easy (a "flag" in the database might do the trick), but I'm not sure if that is true; it may have some performance impact. However, I do think this is an important feature with respect to those that are using Wikipedia as an encyclopedia: you want to find thing fast, and not get 231 article title matches if searching for "Wikipedia" (which is only the 19th hit, anyway).
Of course, it would be necessary to check all redirects over time to see if it should be a "silent" redirect or not, but that's a one-time thing.
Jeronimo
I have a feature request that I'd first like to put on the mailing
list,
before putting it at SourceForge.
[...]
How about excluding all redirects from the search results, that lead to an article that is already listed?
An yes, I'd also like to be able to search in all the other namespaces, like the many people that have asked before me. Can't we have an "advanced search" feature on some special:page?
Kurt
Jeroen Heijmans wrote:
I have a feature request that I'd first like to put on the mailing list, before putting it at SourceForge.
When I use the search feature, I am frequently annoyed by the large number of pages showing up that are simply redirects. When I type Zimbabwe, I might also be interested in "History of Zimbabwe", which shows up. I am, however, not interested in the redirect "Zimbabwe/History", which is a previous location of that same article. I know it is important to keep that redirect for linking (both internal and external), but it is not useful for searching.
well it is if I want to see if there exists a page "surrealist" so I don't need to write a pipe for [[surrealism|surrealist]].
As for redirects like "Zimbabwe/History", we don't want people to link to those. I am of the opinion that we should simply delete them -- the only argument against is that Google has indexed them.
tarquin wrote:
As for redirects like "Zimbabwe/History", we don't want people to link to those. I am of the opinion that we should simply delete them -- the only argument against is that Google has indexed them.
Also there are links from external sites and personal bookmarks. Maybe nothing links Zimbabwe/History, but there's at least one link to Zimbabwe/Geography at this page: http://www.geometry.net/basic_z/zimbabwe_geography.php
How would it help our project if people follow links to what they are told are articles only to find "Describe the text of the new page here."?
Do not delete redirects just because you find them un-aesthetic.
-- brion vibber (brion @ pobox.com)
Brion VIBBER wrote:
tarquin wrote:
As for redirects like "Zimbabwe/History", we don't want people to link to those. I am of the opinion that we should simply delete them -- the only argument against is that Google has indexed them.
Also there are links from external sites and personal bookmarks. Maybe nothing links Zimbabwe/History, but there's at least one link to Zimbabwe/Geography at this page: http://www.geometry.net/basic_z/zimbabwe_geography.php
How would it help our project if people follow links to what they are told are articles only to find "Describe the text of the new page here."?
Do not delete redirects just because you find them un-aesthetic.
That is a fair point which I'd forgotten about. But other sites do not feel compelled to maintain old links.
I have suggested before that instead of just "Describe the text of the new page here.", we have the system run a page-title search and say:
"This page does not exist in wikipedia. Pages that match your requested title include: [[History of Zimbabwe]], .... etc"
then again, we could just leave redirects, it's easier :-)
tarquin wrote:
Brion VIBBER wrote:
Also there are links from external sites and personal bookmarks. Maybe nothing links Zimbabwe/History, but there's at least one link to Zimbabwe/Geography at this page:
[...]
That is a fair point which I'd forgotten about. But other sites do not feel compelled to maintain old links.
I like to think we're better than other sites. ;)
I have suggested before that instead of just "Describe the text of the new page here.", we have the system run a page-title search and say:
"This page does not exist in wikipedia. Pages that match your requested title include: [[History of Zimbabwe]], .... etc"
Wouldn't hurt to do that _in addition_ to keeping explicit redirects.
-- brion vibber (brion @ pobox.com)
tarquin wrote:
well it is if I want to see if there exists a page "surrealist" so I don't need to write a pipe for [[surrealism|surrealist]].
As for redirects like "Zimbabwe/History", we don't want people to link to those. I am of the opinion that we should simply delete them -- the only argument against is that Google has indexed them.
I happen support the view that some of these useless old redirects should simply be deleted. That Google indexes them is really no excuse for keeping them. We keep them because Google indexes them; Google indexes them because we keep them. How can it end that way?
The risk that we might lose some potential follower because the link to the article he is searching is broken is so low as to be ridiculous. As Wikipedia gets bigger that risk keeps getting smaller. If we keep things just long enough for Google to index the article under its updated name that should be more than enough. Many of these old format articles are in such an irregular format that they aren't likely bases for searches. A search for /Zimbabwe History/ is going to give "History of Zimbabwe" anyway.
Keeping garbage on a site just because the space is cheap on the hard drive seems like such a wasteful attitude.
Ec
Jeronimo wrote:
My proposal is to make two kinds of redirects. The first type is the type we currently have, the second one is different in that it still redirects those visiting it, but the page does not show up in any searches. I think this can be implemented relatively easy (a "flag" in the database might do the trick), but I'm not sure if that is true; it may have some performance impact. However, I do think this is an important feature with respect to those that are using Wikipedia as an encyclopedia: you want to find thing fast, and not get 231 article title matches if searching for "Wikipedia" (which is only the 19th hit, anyway).
My proposal is to have "#REDIRECT" and "#DEPRECATED". The latter can also be rendered as nonarchivable, so that they can be deleted after some time in most cases, since they'll fall out of the memories of search engines. They can also be used as you suggest in searches. (But they should still be found in searches if the article that they point to *doesn't* show up.)
-- Toby
wikipedia-l@lists.wikimedia.org