Feature requests

Top priorities: really important features we still don't have

• A 'power search' page. For example, a search for 'demon' will list all pages that use the word 'demonstrate' (case-insensitive substring match?). That's okay now, but as the wikipedia grows the lists of false matches will get longer and more searches will be affected by it. A power search would provide options for boolean expressions, exact-phrase, case-sensitivity, etc. --Atlas 2091
  • I second that motion. Attempts to outsource OddlyCapitalized links would be helped greatly by a case-sensitive search, so one does not, for instance, have to search through 149 pages with the word "liberal" looking for the ones that are printed LiberaL. Of course, the point is moot if 0.92 will include tools to rename pages (and references to them) automatically. --KQ
  • A regexp-like search option might be handy...
  • I would like the search to return those pages more prominently whose title matches the search string. First give me the title matches, then the article body matches.
  • Searches currently turn up only the exact phrase entered, which should probably be one option among many. For an example why, search for "John A. MacDonald," "John A MacDonald," "John MacDonald," "John D Macdonald," and "John D. MacDonald."

• Could we move the search box higher up the screen? It is hard to find right now.

• We need a way to let ordinary folks upload files to the Wikipedia server--perhaps the uploads would first have to be hand-processed, to make sure they don't carry viruses and aren't too large. But eventually, it seems to me we're going to need something like this. (Perhaps it wouldn't have to be closely integrated with UseModWiki software, though. We could just link to a Wikipedia upload page, which would run our own unique software.) --LMS
  • I'm quite surprised you don't have a file uploader, actually. I can send you the one used in the WorldForge website: http://www.worldforge.org/website/tools/upload_a_file. I'd be willing to collaborate on adapting this for Wikipedia if there is interest (I will need a similar tool for work soon). -- BryceHarrington

• Each Wikipedia page could and should have a Talk subpage why not making it default for every new page. Kpjas
  • Seconded. I like the idea of making a default /Talk? on each (non-sub)page.

• Since the text within Wikipedia is covered by the GNU FDL there is a requirement that it be available in the most "transparent" form. I would suggest that since the publically viewable form of is HTML but the underlying dataset is stored in the wiki source form it is currently not available in a transparent form. I guess that it is possible to gather this data by using the "edit" button of all the documents and culling the information from the text field, but again this is not very transparent. Even if you do this, you still have lost the version information which could be argued to be a fundamental part of the data. Is there a plan to provide a download format from which a new copy of the data in wikipedia can be generated? -Phillip2
  • At present, no plan, but we would like to do this, of course.
  • I also think this would be a nice service. -- BryceHarrington

• Clicking the name of the article at the top of the page searches for occurrences of the page title on other pages, but it would be possibly more useful for that to point to a list of all the pages that link to the page in question. There doesn't seem to be any way to do that at present. --Larry Sanger
  • This is likely to be added eventually, and may even become the default behavior of the page-title link. This was not much of a problem with the old WikiName? links because those titles were automatically links. Now with "free links" there may be several pages that use a page title but do not link to it. --CliffordAdams
    • Eventually, a feature to find non-linked mentions of pages could also be useful. --CliffordAdams

• Missed searches - user searches that returned no results (from Wikipedia-L)

• Similar to the What Google Likes page and the request above: a report for pages visited most often (aside from, say, the RecentChanges and HomePage pages and other usual suspects). (So busybodies like me can know frequently-visited areas possibly wanting improvement.)

• Generate and place on an automatically-generated R/O page a list of the most popular Wikipedia searches. This would encourage people to write articles specifically on those topics that people search for. --Larry Sanger
  • This would probably be easiest to do outside of the wiki script by processing the server's logs.

• A script to autogenerate the What Google Likes page once a day or so
• There are several pages that have comments like "as of (date) we have NNN pages / MMM comma pages...". This information becomes stale pretty quickly. Would it be practical to provide a [Wikipedia Statistics]? page that could be updated fairly often with things like page counts, web server stats, user stats, etc.? --loh (2001-06-21)

Report features

• Automatically-generated NewTopics page
  • Still considering this, but not for version 0.92. --CliffordAdams

• Count and publicly display (somewhere) two or three different numbers of Wikipedia pages that are longer than various particulars lengths (to give a more realistic idea of how many actual pages are on the wiki, as opposed to redirection pages, tiny stubs, etc.).
  • A good idea. The only quibble is that it requires opening every page to generate the report, but it shouldn't take any longer than the current search function.
  • If wikipedia articles are stored in flat files, you could just approximate the length based on the file size, right?

• How's about an automatically generated Wikipedia Contributers page that lists contributers and the number of pages they created or changed? --SoniC
  • Unfortunately, not all contributors use a username or have a single login. Also, not all of the requested information is kept permanently. --CliffordAdams

• It would be great to have a feature that would identify all orphaned pages. There's a small but growing problem of our having many useful pages that are not linked-to or underlinked-to. It would be great if we could list all pages to which there are no links, or only one link. --Larry Sanger

• I would like a page that is a top list of the most frequently linked-to non-existant page, more or less the opposite. LinusTolke

• How about a feature that lists all places where the current page is linked from?

• Provide a batch job that searches for and lists all links along with the page the link is contained in so that we can look for duplicate/slightly different versions of the same link and fix them. Please make this a weekly batch job so that it does not take too much time.

• Is it possible to invent something to sort lists alphabetically? Take a look at pages like Actresses and you know why this would be helpful...

• Plagiarism-bot. A bot that goes through Wikipedia pages on a regular basis, does Google searches, and flags pages that it suspects might be plagiarized.

• Spell check batch job for newly created articles once a week with a link at the end of the article to the incorrectly spelled words. Even better would be one that worked as the "preview" button, before saving, and that marked somehow the suspicious words.

• Let's have contributors of a given Wikipedia page listed either in the header or in the footer. IP's are skipped of course and if there are more than, say, 10 contributors make a link to a special subpage "Contributors".Kpjas
  • If people wish to be listed, can't they simply include their name at the end of the text?

• Provide a way to search for all the changes made by one user in Wikipedia. For example, if I contributed to numerous pages over a long period of time, then 2 years later, I would need a way to find out whether someone else added to the subject I once touched.

Naming conventions (this is all done, but not yet implemented on Wikipedia)

• The wiki software could be written so that, for example, if I write [[andy jewell]], [[Andy jewell]], [[andy Jewell]], or [[Andy Jewell]], the page linked to will always be Andy_Jewell. In that case, one would be able to capitalize or not in the text of wiki articles at one's pleasure, and only the titles of articles would look strange (in some cases, e.g., Ich_Bin_Ein_Berliner). But that situation would be preferable to the current situation, in which some people are capitalizing second and third words of article titles, when, in referring to the subject of the article, one doesn't ordinarily capitalize those words. (For discussion, see naming conventions.) -- Larry Sanger
  • This has been done in version 0.92. The canonical form of a page will have all words capitalized, and links can have arbitrary words capitalized. This will be a little strange sometimes, but it seems to be the best solution for now.

• Now that we have Free Links, can we have the canonization software allow parentheses, so we can use titles with short glosses rather than awkward compounds, e.g., [[Nirvana (musical group)]] rather than NirvanaBand? The downside of this is a little more typing to make a link; the upside is that we get to have titles in plain English. --Lee Daniel Crocker and Larry Sanger (see Naming conventions/Disambiguating)
  • This is in version 0.92. I was surprised to see that parentheses are allowed in URIs/URLs? according to RFC 2396, so it shouldn't be hard to add them. --CliffordAdams

• There's a good reason why we can't use single quotes in titles, right? It sure would be handy to have the use of them, though.
  • After more thought, I've given in on this request also. Single quotes are allowed in 0.92. Please don't ask for any more punctuation, however, or I may scream. :-) --CliffordAdams
    • How about backticks and semicolons? (joke!)

Cookies, logins, and privacy

• In addition, given that the certain cookies have stopped working, can't I have a "log in" to confirm my id? Everytime I hit "preferences" otherwise, I get a new user ID.

• I noticed last weekend that I can set my ID in Preferences to anybody else's, certainly, if they are not using it. I would not care if people had mutlyiple IDs. But, I do see a problem in my setting my ID to, say Larry Sanger and entering content all over which will wrongly be identified as his in recent changes. And possibly elsewhere. RoseParks
  • There is a secure user-ID already--it is the "User ID number" shown on the Preferences page. If you are using recent versions of MSIE or Netscape you can see the ID number and the IP address by moving the mouse over the user name. (A popup saying something like "ID 1622 from 165.79.13.xxx" should appear.) It should be non-trivial for anyone else to use that ID number, since it protected by a random number in the user's cookie (which is compared with a copy on the server). Eventually a more conventional username/password combination may become an option, but this is unlikely to happen soon. --CliffordAdams (or somebody with ID 1675 :-)
  • I second the request for a more conventional username/password combination. This is one thing necessary to make the website fully scaleable. --LMS

• The ".xxx" bit in dynamic IP addresses is commendable, but is there a way to do something similar for people logging in from institutions with static IP addresses ending in letters? The last one I saw listed the entire address. Certainly anyone with a static IP should have a firewall, but I think we've seen enough mischievous people wander through that the precaution of masking the IP still has merit. And no that's not a rhetorical question, as I'll be logging on from a static IP starting in early June 2002. --KQ

I'm not sure "privacy" is a good thing here. The price one pays for freedom is accountability: anyone should be able to edit anything, but I don't think they should be able to do so without being identified (at least with their chosen pseudonym--they can still hide behind whatever anonymity their Internet connection gives them). When I find some vandalism, for example, I look back in the Recent Changes list for that same IP and check those pages--I usually find more. Likewise, when I get to know certain posters, I get to know what might be interesting or important to read. If someone is concerned about privacy, that's his problem, and he should take the burden of using software to anonymize himself (and there's plenty of software available to do that). --LDC

I think you've misunderstood. What I'm talking about specifically is not having the entire static IP address shown of someone logging in from an address ending in letters. All it would take is to .xxx a portion of it so that the firewall is not constantly fending off requests. This is the same courtesy already extended to people logging in from any IP address ending in numerals. My concern for it stems from the fact that it identifies where the person works or attends school, which if you'll pardon my saying so, is not necessarily anyone's business. If it's not workable then you can expect to have me not continue contributing. --KQ

I usually use a static IP, but RecentChanges (hover over my name) only shows a masked IP, not my domain. Did you see this on the main Wikipedia, or one of the internationals? UseModWiki can turn off name lookup, and I'd assumed from the RC display that it was turned off.

You're asking that mybox.frobnuts.com be masked as mybox.xxx.xxx, (which is a domain name rather than an ip address, to be unnecessarily picky) so that FrobNuts?'s network (thus their firewall) won't be identifiable?

Even the current method really doesn't provide that level of protection, or much privacy of any kind beyond a casual glance. For example, 111.222.333.444 would display as "111.222.333.xxx", and thus "hide" a specific host, but the network can likely still be identified via a simple whois lookup on "111.222.333". With my usual IP address, the two leading numbers are sufficient to tell you where I'm logged in from; you don't even need the third one. In general, only a residential poster would have some anonymity, simply because the ip address (whether static or dynamic) usually maps back to the ISP, not a specific business or school.

To achieve the equivalent level of privacy you're asking for, even the ip addresses should be masked as xxx.xxx.xxx.444 rather than as they are now. --loh

Interesting. I don't remember if it was on the English wikipedia or the Spanish one; it's been some months past. I had two concerns; one was indeed not giving the entire IP address (or domain name); the other was not revealing immediately where contributors post from, leaving it instead to their own discretion. I haven't seen it again; but if it was originally from the Spanish wikipedia that would be because I've contributed nearly all I can in Spanish. :-) I expect that I'll try it from another computer to see how it displays and will be using anonymizer or somesuch if it's not to my satisfaction. Thanks.

Other feature requests

• Disallow self-referencing wiki links. It is counterproductive to the idea of a cross-linked encyclopedia to link the page the user is currently on to the page the user is currently on. One can easily kill these links when one comes across them, but perhaps it would be easier to simply disallow it in the wiki code?
  • I think the idea here is not to parse a link on the XYZ to the XYZ page as a link.

• For every topic, go through and grep every other topic (maybe just the root topic) and automatically create a link to the topic we are looking for. So eventually every word could possibly be a link. This could be great for people interested in the origins of words and would save topic creators much time too.
  • This is a bad idea. First of all, there are the obvious technical difficulties. More importantly, human beings are much better at deciding when a link is relevant to a given bit of text. Patience--it'll be perfect in the end, trust me. --LMS

• HTML comments in Wiki pages. I'd like to be able to add comments to a page for authors and editors that are permanent, not like discussions on a Talk page. Things like "Note the layout of this table; if you want to add an item, do it this way..." or "Someone included this example here, but I think it's better on this other page, so I moved it there", and so on.
  • Why permanent? What would be the point of that?

• Anchors and internal links [[like#this]].
  • Anchors within a single page would be great, but inter-page anchors are probably not a good idea. One possible implementation could be: [#foo] ... [$foo], where the hash is the reference and the dollar is the target. I'd really like to have that on the Wikipedia FAQ page.

• Automate making old-style links into Free Links
  • This will probably happen soon after the 0.92 release. The wiki code to rename pages is already present, although it is available to wiki administrator(s) only.

• InsertPagesIntoOtherPages
  • Not likely to happen soon. Opens too many new issues for minimal gain. --CliffordAdams
  • The KeheiWiki? has an include functionality. If you ever wish to explore this let me know and I can give pointers. -- BryceHarrington

• I would like to have a CANCEL Button on the Edits page (Page in Edit mode) added to the two existing ones SUBMIT and PREVIEW. Just to have a clear understanding what is going on and what the system/database is up to. StefanRybo
  • How about a "SPELL CHECK" button, too?
    • It would be helpful if the spell checking also check for variations of Wiki links, such as "Recent Changes" vs. "RecentChanges". It can eliminate branching of similar pages.

• Automatically update the present (server) date. Link the date to that day's page.

• When I'm editing a topic, I'd like to have the search box available at the bottom of the page. Janet Davis

• Making the "redirected from" comment at the top of each page an option which is by default off, not on. This feature would clean up wikipedia pages for casual users, and leave the nuts and bolts visible to people who wish to track down the offending OdDlyCapitaliZed links, etc. Of course (if I may say so) ;-) for this option to work, the caching problem would first have to be resolved. --KQ
  • Under this scheme, how could we ever visit the redirection page? By switching a preference? I dunno, seems like we have to edit redirection pages too often. This would just create confusion, wouldn't it? --LMS

• A way to do doublecolumns and/or sidebars (so we can have a list 'o links to the right of the text)
  • Double columns are only natively supported by Netscape. However they could be done by using tables. chuckr30
  • Maybe, if such sidebars could be created automatically and for all articles. But what's wrong with having the links at the bottom of the page? If that's reasonably acceptable, why go through the confusion of having links in a sidebar? --LMS

• I was trying to find a good way to change the links for the entire UnitedStatesConstitution section so that all the subsections would automatically migrate to United States Constitution. Instead of copying and pasting everything on the web, I figured I could just copy the files on the server from one directory to the other. I tried this and the site broke, so now I think I will just ask if there is a way to do this. So: When we are changing links to the much nicer free link style, is there a good way to get subsections changed as well without having to manually edit each page in the subsection?
  • The wiki admin(s) can use the page-rename commands to move pages. Unfortunately, each individual subpage needs to be moved by this method. I'll consider this to be a feature request for renaming all subpages when a page is renamed. --CliffordAdams

Old feature requests (completed!)

• Free Links
• Renaming/deleting function
• BackLinks?: see which pages links to the one you're at (Click the title of the current page (at the top of the page)--it will do a search for links to the current page)
• a way to allow non-English characters in article names (in newest version)

Really ambitious and fanciful feature requests

• For famous writers: include pages of the books they have written, taken from the Project Gutenburg Library. PG is a library of books that have fallen into the public domain. Include them as links to the specific .txt files on PG's servers, so we don't have to deal with the endless defacement of Romeo and Juliet. Imagine being able to look up Kipling, read a description of his life, and then read all of his works.
  • This can't be done automatically in any easy way. It's something we're doing by hand. See [public domain resources]?.

• Make the category an article belongs to a property of the article. So one can look for all Mathematics articles, etc. This can already be done to some extent: you can always create a category page and add the article to it; but if someone forgets to do this the article will be missed. An automatically generated page could exist for each category, listing what articles belong to it; and also an automatically generated page for articles that do not belong to any category, so they can be assigned to one. Also:
• At the top of each topic page, I would like to see a link to the parent topic. chuckr30
  • These aren't just software feature requests, because it's not a programming task to determine what "category an article belongs to." (What does that mean?) Please see this column. Re: "This can already be done to some extent: you can always create a category page and add the article to it; but if someone forgets to do this the article will be missed." I reply: So? So the page will be (temporarily) imperfect. Better that than a totally controversial hard-wired category scheme of limited value. The content is the category scheme. --LMS

• I was just thinking about e-mail interface to Wikipedia. What for, you might ask. Sometimes it is for some people hard to get onto the Wikipedia server. I also think it would booster authors' productivity. Off the top of my head I can think of three commands that can be placed in subject line of an e-mail :
  • POST
  • GET - raw text of the page
  • RECEIVE - formatted text of the page
• of course there should also be 'an article a day' service. POSTing should be reserved to active memebers of this mailing list. There should be some way to safeguarde against accidental overwriting of existing pages. When POSTing raw text in body before '-- ', and author's id from e-mail before '@'.

• I would like a way for the author of an article to "lock" the article. If someone wanted to add something to it, they would email the changes to the author. If the author approves, the author can post the new article. I think you will see tons of poor quality articles here eventually, because some 12-year old thinks it's cool or funny to deface articles. I think it would be wise to be able to lock the articles. chuckr30
  • There is no such thing as "the author of an article" on Wikipedia, unless it's just an accident that one person has happened to work on an article so far. So this is not going to happen anytime soon. It is totally anti-wiki, for one thing. Eventually, some such solution might be necessary to make Wikipedia fully scaleable--but I doubt it. Right now it is clearly unnecessary: people are our protection against vandals. --LMS