Jump to content

NamingConventions

From Meta, a Wikimedia project coordination wiki

This page is solely of historical interest. Current English Wikipedia naming conventions are described in en:wikipedia:Naming conventions.

Recommend naming one-word pages like AfghanistanCountry rather than the eminently ugly AfghanistaN.

In this same manner, every noun can have Person, Place, Idea, Invention, etc. tacked on to its end to identify its type. You might even want to hold this convention for words that already meet the wiki naming conventions, like ThomasEdisonPerson. You might name the "letter" pages things like IndexA, which might link to IndexAaAl and IndexAmAz (once you get big enough :-).

For example, say I want to link to a page from my site. I have an InterWiki prefix set up for Wikipedia on my own wiki. Without going through the rigamarole of actually looking up something on Wikipedia, I have no a priori way of knowing whether to link to

See also : NamingConventionsWikiPedia:AlAska,
See also : NamingConventionsWikiPedia:AlaSka,
See also : NamingConventionsWikiPedia:AlasKa, or
See also : NamingConventionsWikiPedia:AlaskA. Lather, rinse, and repeat for longer names like AfghanistaN. If I knew there was an AlaskaState naming convention, then I wouldn't have to give it a second thought.

The arbitrary naming convention used so far is very inconvenient and doesn't contribute at all to AccidentalLinking. At very minimum, you might investigate other wiki engines that support arbitrary link patterns like

See also : NamingConventions[Alaska] and
See also : NamingConventions[Ward Cunningham]. -- ScottMoonen

There is, however, a search form at the bottom of every page. Isn't it both more convenient and more liberating in the end to let people do whatever they want, and just search for a string of characters whenever you think there might be related page already in existence, to which you might link?


I respectfully disagree:
  • AfghanistanCountry matches every search AfghanistaN matches.
  • AfghanistanCountry additionally matches every search for "country".
  • AccidentalLinking is vital. When I'm typing an article on the SovietUnion and want to link someplace I don't want to waste the time to go off and search whether it's AfGhanistan, AfgHanistan, AfghAnistan, . . ., AfghanistaN. I want to know right off the bat what it is.
I submit that there is nothing in the way of convenience or liberation that AfghanistaN can do for you and which AfghanistanCountry can't. And it seems to me there are serious usability deficiencies in AfghanistaN. Consider especially those persons who haven't experienced a wiki before. OTOH, as I suggested above, you might research wikis that allow you to create one-word links that arent UgLy. -- ScottMoonen

I (Jimbo Wales) am the one who started the bad naming convention. I simply plead total ignorance. I realized when I started that it was a problem, but I didn't know what to do, so I just kept plugging away.

I think that ScottMoonen's suggestion has great merit. Unless someone beats me to it, I will work on making that change over the next few days.


I believe the above-proposed convention is, in many cases, too restrictive. It creates titles that sound strange when used as parts of sentences. E.g., if I want to say that Afghanistan is a country in Asia, I am forced to say, "AfghanistanCountry is a country in Asia," which sounds very awkward.

The argument that this convention makes it easier to search is obviated by the fact that, after all, people will organize all countries, all countries in Asia, all Muslim countries, etc., each on their own pages (in the near or distant future anyway), and more importantly, that a full text search is possible. Why impose a category scheme in the naming convention?

So I would prefer not to include categories in most (not all, I suppose) topic names because that tends to restrict the discussion of the category. For example, if I were to write "FiddleMusic" instead of "FidDle," then the tendency would be to discuss, well, fiddle music instead of all things fiddle. I could write "FiddleMusicalInstrument," but then the tendency would be to discuss the fiddle qua musical instrument, and there are many other aspects of the fiddle apart from those directly related to the fact that it's a musical instrument. Generally, I'm opposed to imposing pre-set category schemes: things do not wear their categories on their sleeves (as though they had only one supercategory) and we should not name articles as though they did.

Finally, I think it makes sense simply to capitalize the internal letter of one-word topics which begins the strongest syllable (or next-strongest syllable, if the strongest syllable is the first). Thus: FidDle; AlAska; RusSia; DoneGal (yes, in fact that's the strongest syllable, the way the Irish use the name). In cases where there is both one word and one syllable, I'd propose capitalizing the last letter: BasS; JazZ; SoaP. This is the convention I've been following, anyway. And, of course, for multi-word topics, capitalize all the words (except, perhaps the conjunctions and articles?) and mash 'em together.

That's my take, but go and do whatever you like! -- Larry Sanger


Well, Larry Sanger has good points, too. I think we can all agree that alphabetical page names like 'AaA' are bad. I want to organize all the CountriesOfTheWorld, but someone else will want to organize all the SportS or whatever.

I have changed my index pages, but so far I have not made any alteration to the country names.

One problem with Larry Sangers proposal, for multi-syllabic words, is that people may not know what is the strongest syllable.


We could, should we so desire, make only two very simple rules. First, multiple words get capitals at the beginning of each word. Single words (and we could choose, here - and spread the word to attain consistency) get caps on the first and third, or first and last letter. Hence, GoD is accurate in either scheme by default, and hell would either become HelL or HeLl. Likewise, AfGhanistan or AfghanistaN. We would not do both, but one or the other. At least such a rule would be simple and easy to apply in all cases. (but then how would we link to "id?" heh)


Wouldn't it be a good idea to add a tool on each page helping us to find names with incorrect or missing links. My suggestion is to modify the preview page to include a list of all words or combination of words present in the text that have entries (independant of capitalization) in wikipedia. In that way it would be simple to pin-point errors in capitalization and the need for rules diminishes. --LinusTolke


The first and last letter approach requires the least thinking to apply, for single word topics. In any case, it seems kinda plain that we ought to try to standardize a bit here, since otherwise "accidental linking" is going to be a lot less accidental.  ;-) -- Bryce


Thank you, Bryce. I, too, would be fine with the "first and last" rule on single word topics, to go along with the "first letter of every word" rule on multiple word topics. All the rest, what say ye? -- AyeSpy


On the other hand, it wouldn't be too hard to change the wiki to allow links like word for single-word topics. See WhichWikiShouldWeUse for further discussion of this idea. Otherwise, the last-letter capital is probably the best idea that fits within the wiki conventions. --CliffordAdams


A request: for an encyclopedia, the usual Wiki link-naming scheme seems not quite right. How about patching the Wiki server code so that any string that ends in a special character (such as "~") becomes a Wiki link.

I second this request. I propose that wiki links should be enclosed in double braces, Template:Like this. Of course, we'll have to do some serious re-naming of pages when the new standard is implemented--that might not be for a while, too. Don't hold your breath.


The discussion so far has left out a class of single words that I would like to bring attention to. Those being the words that have something of a natural way of being broken up that reflects the structure, history, and/or root meaning of the word. Like PhiloSophy, terms that end in -logy, -ory, etc.

I rather like the reminder of the roots of a term. Sort of an antiNewSpeak type thing. That having been said, seeing -ism capitalized all over the place has become a bit unnerving. Further, many people don't want to bother learning the roots or be reminded of them.

However, I would like to offer this as an alternative (for some words) to the last-letter idea and read what others think of it. Oh, and the word seems the best solution since it lets me play with roots and frees the wiki of an arbitrary constraint --PhillipHankins

In most cases this can be codified fairly easily, and so applied without an etymological dictionary. Usually the letter immediately preceding the first O would get capitalized: GeoMetry, DemoCracy, PhiloSophy, BioLogy. Words that end in I stems would have the last I capitalized: MarxIsm, PolitIcs, ChristianIty, SkiIng. Failing those, you could just look for some common stem words: Al-, -Lysis, -Archy, etc.


Various animal articles are coming in, so I'd like to fill in some of the hierarchy thet fit into. Does anyone have any idea about how groups should be named? Some possibilities:

  • PhylumPorifera (one difficulty is that not all groups have clear ranks)
  • PoriFera
  • SpongE

Note also that high-level taxa could get named differently from low-level taxa: KingdomAnimalia, WolF. Does anyone have any thoughts on how they'd like groups named?


Don't it go to show that these wikinames just won't work in the long run? It's fun - and easy - to add new pages, but what if I want to look up a subject? Will I find America under AmeRica or AmericA (both look equally silly, if you ask me)? And why should I have to memorize a gazilion CapitalizationRules in order to find what I want?

Nah, in my opinion we should be using a method that will allow us to create a link regardless of, but not excluding, the capitalization - much like they do on Everything2, where you create a link by enclosing it in square brackets [LikeThis]. And once this is established we should clean up this wiki for all the obscure and ugly capitalizations we can see today.

My advice is that we implement and enforce this before too long, otherwise it will be a tremendous task to correct the pages and to look up things on Wikipedia. --KlausSeistrup

The code for "free" links like
See also : NamingConventions[[Democracy]] is currently being tested on a test wiki. (Feel free to try it out on that site.) This new linking feature should be available "soon". I used double-brackets for links so that people can continue to use brackets in their ordinary editorial sense [like this] without creating links.
Conversion to the new form of links should be semi-automatic--I'm planning on building a script that will list all links and a suggested new name (like "DemocracY=Democracy". After editing this list for special cases the list will be used to automatically rename the pages and all links (within the wiki) to those pages. --CliffordAdams

Sounds great, Cliff. Cheers, mate. --KlausSeistrup


Links with spaces:

The new free-linking code can allow links with spaces and limited punctuation, like John F. Kennedy. While I think that the current wiki names (squishing words together with capitals) are often useful to writers, they can make pages difficult to read and explain to users new to wikis.

Should links with spaces be allowed or even encouraged on Wikipedia? Should the existing pages be converted to this form? (Should this page become Naming Conventions for instance?)

In the end, the preferred link style may become a site policy decision. Before that happens I hope to have some discussion of the issues. The biggest issue I see is whether the Wikipedia should try to follow the wiki-culture from other sites, or whether it should develop its own cultural norms (which may diverge from existing wikis). --CliffordAdams


I'm personally for links with spaces. Maybe it's not very "wiki" (I didn't know what a wiki was till I came here through nupedia), but it's certainly more "encyclopedic" --AstroNomer


I'd appreciate other parties to jump in on the discussion Cliff and I are having on http://usemod.com/cgi-bin/mb.pl?FreeLink. What do you think about the canonicalization (c14n) concept? That would make a link like

See also : NamingConventions[[John F. Kennedy]] into john_f_kennedy which would match correctly against [[John F Kennedy]] and JohnFKennedy as well. -- SunirShah
See also : NamingConventions