Wikipedia talk:Category types

Discussion originally posted at Overcategorization
I've also been thinking about some general purpose templates for categories that might help in their management. I hate the catdiffuse template, and would like some that are much more user friendly, not blatantly self-referential and move us towards the category structure discussed at Category intersection. I'm not sure of all the details, so I wanted to bounce the idea off you before I do anything. (Samuel then wrote the first version of the proposal on the project page)

What do you think? -- Samuel Wantman 08:43, 25 November 2006 (UTC)


 * Perhaps this thread should be moved to Wikipedia talk:Overcategorization, but I tend to agree with the sentiment expressed at Wikipedia talk:Categorization. As currently implemented categories have no formal meaning and no formal structure (other than directed graph), which I think means any attempt to attribute meaning to categories is ultimately doomed.  I suspect there's not much point in doing anything along these lines until we have something like semantic mediawiki. -- Rick Block (talk) 18:13, 25 November 2006 (UTC)
 * Rick, thanks for commenting. Yes as currently implemented categories do not have a formal meaning and a formal structure. That is why I'm trying to give them more structure.  The classes I came up with are based on the structure that most categories seem to have, and by putting these labels on categories we would be giving categories more structure in a wiki-like way.  Categories seem to be at a point that they will either degrade into meaninglessness or take form.  I'm not quite ready to give up on them yet. -- Samuel Wantman 20:24, 25 November 2006 (UTC)
 * I agree with Samuel's types of categories outlined above. These are exactly the sort of categories I've been seeing develop. I've always tried to add a few lines explaining a category to the editors using it, and directing them to more appropriate categories if needed, and directing readers to the correct place as well. See what I did at Category:Tsunamis and Category:Tsunami. Carcharoth 00:41, 26 November 2006 (UTC)
 * Oh? I've visited this page before? I don't remember it having the snazzy looking symbols back then... :-) Carcharoth 10:40, 11 January 2007 (UTC)
 * It was just an outline at WP:OCAT then. --Samuel Wantman 10:50, 11 January 2007 (UTC)

New comments

 * Very strong oppose Samuel Wantman wants to convert his own views on how particular articles should be categorised into mandatory rules that everyone must follow, taking  away their discretion to use their own judgement. This would be inappropriate regardless of what outcomes he advocated, but it is made especially so by the fact that he has very eccentric views that command very little support, as is demonstrated by the currently unanimous rejection of his attempt to get the category diffusion template deleted. Choalbaton 14:47, 4 January 2007 (UTC)
 * Opposed to what? This isn't a vote. I'm not even done with creating these labels and the text on the page.  These labels could be tailored to our categories no matter how they are populated.  While there is objection to removing Catdiffuse, there was clearly support for fixing it and helping people understand how to use categories.  That is all that I am trying to do, and you can help in that effort. -- Samuel Wantman 02:51, 5 January 2007 (UTC)

Personally, I find the page interesting. A couple thoughts: As this is a guide to category use, I think it should be a guideline, once it's more complete, and I think it should be made clear that other uses for categories are possible, just that these are the most common (embracing WP:IAR). - jc37 06:13, 5 January 2007 (UTC)
 * I'm designing the templates with several parameters so that they can be tailored to different situations. If it gains wide acceptance it would be a guideline.  Until then it is just category tags. -- Samuel Wantman 06:24, 5 January 2007 (UTC)


 * I like the general scheme here. However, I still find the notion of "index categories" problematic, as it seems to push us (even further) away from having any coherent ontological scheme, and multiplies the number of categories likely to be present in a given article.  Unless something like category math really is implemented -- in which case index categories would become standard -- I don't see how these add value.  For general browsing purposes, lists are far more maintainable.  -- Visviva 10:02, 5 January 2007 (UTC)
 * See Category intersection. Developers are working on how this can be implemented.  There is widespread support for something like this.  The only impediment to it is figuring out how to constrain intersections that will hog up too much server time.  There do seem to be ways to deal with this.
 * The bigger question for me is whether we want the categorization system to be a classification system that finds a small pidgeon hole for every article or an indexing system that makes it possible to browse through larger groupings. With the current system, I don't understand the rationale for having things like Category:Battles of the American Civil War depopulated when it is broken into subcategories.  It assumes that users have some a priori knowledge of the topic before they start browsing. That seems backwards to me.
 * In the film example, maybe I'm the oddball, but I really think there is a value to having an index of all the films in Wikipedia along with more specific subgroupings. I don't think that diffusing "Comedy films" into dozens of smaller categories would make browsing very rewarding.   I don't think there is all that much difference between browsing through a large category or a small one.  Size isn't the issue.
 * I also don't think what I'm proposing adds many categories. I'm suggesting that there usually be a single general category and several specific subcategories. Currently, we have just the subcategories populated, so I would be adding a single category to each film for each hierarchy, but this is only for articles that are in index categories.  Most of these articles do not have entries in more than a few hierarchies.  The multiple categories for a film would just add a single category.  People would add a few.  It would be a rare case where the number of categories would multiply.
 * As for lists, it is much easier to maintain several small lists than a large comprehensive one, so if anything should be a list it should be the pidgeonhole. If it is not worthy of an article or a list, why would it be worthy of a category?  As for intersections, it becomes un-manageable to intersect more than two or so traits, and this has been commonly recognized as an overcategorization.
 * As for classifying an article, I'm not sure what the value is for a general user. Each article is classified by the intersection of all the categories it is in.  For many articles this is a unique set of categories.  The bottom line of all this is that there seems to be a common desire to make categories into a way of doing database searches, and I understand why that is attractive.  However, I don't see our categorization system doing a very good job of this, so I think we should wait for category intersection.  When that happens, we will be need to repopulate all these categories the way I've described.  In the interim, we can work in that direction, and also create a system that is very conducive for browsing through general topics.
 * We already have index categories. Even if they are chopped into tiny pieces they are still index categories.  I'm hoping that these tags could be used the way things are now.  Perhaps you can think of a better name for the labels. There are two issues here.  One is the tagging of categories based on function, and the other is if larger indexes get depopulated and turned into navigation categories.  Perhaps I'm not understanding your concerns.  -- Samuel Wantman 22:20, 5 January 2007 (UTC)


 * Terrific idea - let me join your minority. (My abiding feeling is that all categories should be deleted forthwith and we could all do something useful.) roundhouse 13:02, 9 January 2007 (UTC)
 * Let me join what might be a majority now! :-) I don't agree with the deletion of categories, but I do agree with this formalisation of how the best categories work in practice, and I do agree that this will help focus categorisation and hopefully help prevent overcategorisation as people start to think about the overall structure of how they categorise things. Carcharoth 10:38, 11 January 2007 (UTC)


 * It's an interesting idea, to have some information explaining types of categories on relevant pages. I agree that it might help people understand categorization better.  I'm refraining from commenting on the specifics of the proposed tags until i've read them closely & thought about them a while.  However, a couple of quick comments on discussion above.
 * At one point, SamuelWantman says there's little difference b/w browsing a large & a small category. That may be so, for him; but people use the categories for a lot of different things.  some people use the categories just to find an entry point, or to scan for something specific; scanning a *lot* of stuff is a lot harder than scanning a *little* bit of stuff.  The "list of all films" is a good example to compare with the "US civil war battles'.  Subcategorization of US civil war battles presupposes some knowledge, as SamuelWantman says; but it doesn't have to be a lot of knowledge, just enough knowledge to navigate the hierarchy.  No subcats, as in the "index of all films", presupposes that you know the title when, in fact, that may be the very thing you're not remembering.  The point is that both are useful features & functions.  What one wants in those situations is most specific categorization + the ability to show an "exploded" listing of the parent cat + the contents of all the parent cats' children cat, in a tree formation. --lquilter 14:53, 26 January 2007 (UTC)
 * One problem with applying any "this is an X kind of category" tags is that in wikipedia at present there are few (if any) purely "X kinds of category". Most categories & cat trees are multi-purpose.  So, for the few purely taxonomic trees, it might be helpful to include a "tax cat" tag that explains that proper subsets only are appreciated; but for most cats, I think it would just add to confusion. And I think it would actually be wrong to try to be prescriptive at this point -- neither the software nor wikipedia community practice is supportive. --lquilter
 * I agree with lquilter's point that these tags should only be applied to mature and well-developed (I hesitate to say 'stable') categories. Firstly, if the categories in question change a few months later, the tags would need changing. If they are not changed, the tags will be misleading. The tags might also discourage people from subcategorising or reorganising a category structure - in some cases this would be good, in other cases it would be bad. Carcharoth 10:35, 1 February 2007 (UTC)
 * My second point relates to lquilter's point about wanting an "exploded" listing of the parent cat + the contents of all the parent cats' children cat, in a tree formation. This is already available at the Special:CategoryTree tool, though that will only show the first 200 articles in a category, so is not suitable for very large categories. Carcharoth 10:35, 1 February 2007 (UTC)
 * In fact, I suggest that for taxonomic categories, a templated link to Special:CategoryTree using "PAGENAME" be provided at the top of the category page. Carcharoth 10:39, 1 February 2007 (UTC)

These are a bad idea
These categories are naive and inaccurate, an attempt to pass of an eccentric and inaccurate set of definitions as fact rather than opinion. All of them are ridiculous. For example one cannot say that certain things are topics and others are not. The level and type of category that one designates as a "topic" is purely a matter of opinion. For example, why is it "film" that is the topic? Why not the broader "culture" or the narrower "cinematography"? These templates are a personal project and should not be imposed on anyone else. They will not help readers, but simply jar against their own way of thinking, which is likely to be just as valid at Samuel Wantman's way of thinking. Cloachland 06:32, 6 January 2007 (UTC)
 * Why is film a topic? Because it has articles about film instead of articles which are a film.  The only distinction I am making is one that is already made in most categories.  Culture, and Cinematography are also topic categories.


 * It has been customary to include subcategories in topic categories until the topics get so large that you cannot display everything on one page. Then it is quite normal to leave the articles in one category and create a navigation category for all the subcategories.  Typically all topic categories include some navigation.  It is only when there are no articles that I call something a navigation category. So basically all I'm saying is that if there are virtually no articles the category is for navigation, if the articles are all examples of the category name ("x is a y", bridges, people, films, etc...) it is an index, if articles are not about the category name, or are an instance of the category name, but just relate to the category name they are subjects.  I've just tried to pick names for these types that I thought would make sense for the common user.


 * X is a Y = index
 * X is about Y = topic
 * X relates to Y = subject
 * No X (just subcategories) = navigation


 * Small categories that combine all of these are topics. As they get bigger the category gets divided up. Sub-topic and index categories get created.  When there are lots of them navigation categories help organize.  I don't think this is my own personal way of looking at it, but a description of what actually happens. -- Samuel Wantman

Hiding the actual content
It is already a problem that the top section of category pages takes up so much space, and adding multiline boxes would make things far worse. When I go to the category I want to see what it is in, not all sorts of introductory comments and links to other things. These boxes assume that people aren't smart enough to develop their own understanding of the category system, but actually readers can be trusted to do that just fine, and will not benefit from this patronising intrusion. Cloachland 06:35, 6 January 2007 (UTC)
 * I agree. There are links to parent categories at the bottom of each category page, and these boxes partially duplicate their function while adding very little benefit. They also potentially introduce an element of controversy, as they take a POV on which links are most relevant. That could be dangerous in hotly disputed subject areas. Wimstead 13:04, 9 January 2007 (UTC)

I think you would need to survey the readers before you can conclude that they understand the category system. From my experience, editors don't agree over what the category system is meant to be, so this will help reduce conflicts like that. You might see it as making things too strict, but some order for the category system is long overdue. Carcharoth 10:30, 11 January 2007 (UTC)
 * As I see it the category system has plenty of order already and is steadily improving. Osomec 19:09, 17 January 2007 (UTC)

This is a nice flexible system
From my own experience of categorisation, I recognise the category types you are describing here, and I support the system that is being set up. To address some of the concerns people have raised, maybe it would be a good idea to demonstrate that a category that seems to lump lots of things together can actually be separated out into several of these types. Find an example that covers all these types, and then give the category name used for each type. You already have film for three of them. Try and find an example where the differences between all five types can be demonstrated in the same subject area. For example, if you have a long list of earthquakes (the index category: Category:Earthquakes), some people may want to divide it up by date, some by location, some by type, and so on. These are the navigation categories (Category:Earthquakes by country; Category:Earthquakes by century; Category:Earthquakes by type - the latter hasn't been created and populated yet). The topic category would be articles on the science of earthquakes (Category:Seismology). If an earthquake spawns lots of subarticles, like the 2004 Indian Ocean earthquake, then that would have a subject category, in this case Category:2004 Indian Ocean earthquake. Am I understanding this correctly? Carcharoth 10:27, 11 January 2007 (UTC)
 * Yes. This is exactly correct.  There seems to be a natural progression to how categories develop.  They start as topic categories and pretty soon, specific instances get divided out into an index (often by just adding an "s", like "film" and "films").  The indexes get broken up into several indexes and then the topic category is also a navigation category (see; Category:Bridges) for an example).  When this gets too big, the navigation category could get split off as well.  This doesn't seem to happen much these days, instead the topics get split into subtopics. -- Samuel Wantman 10:50, 11 January 2007 (UTC)

I'd like to see what this would make of Category:Women... - jc37 10:44, 11 January 2007 (UTC)
 * Oops, nevermind... It seems that User:Lquilter just got finished cleaning up the category yesterday : ) - jc37 10:47, 11 January 2007 (UTC)

This is a bad, inflexible, prescriptive system, patronising and a waste of space
The underlying idea seems to be that there is one correct way of thinking about categories, and users should be guided to follow it. This negates one of the greatest strengths of wikipedia, ie that it is gloriously flexible with multiple methods of navigation, and plenty of flexibility in the way they are used to account for special circumstances in each field. It is all very well for a user to clarify in his own mind how the category system works for him, but the proposal here is that this should then be displayed across Wikipedia in what would amount to tens of thousands of screens worth of text and colour. If this was implemented actual articles would rarely be visible when a category first loaded, one would have to scroll down past this largely useless information every time. When a user visits a category they usually know that they want to see what is in the that category; it is completely inappropriate to give make a set of suggestions for other places they might look the first thing that they see. Osomec 19:08, 17 January 2007 (UTC)
 * I too have problems with the "see this and that" bits. They are too formulaic. I think it is best to just have the "this is an X type of category" and leave it at that. Note that you can do this already, as the "see this and that" bit is optional. Carcharoth 02:32, 18 January 2007 (UTC)
 * Also, "see this and that" bits are often already in a category. Just putting the "type" label on a category is enough. Carcharoth 02:32, 18 January 2007 (UTC)
 * My intent was to try and standardize the way the "see this and that" bits are done, not necessarily to incite more of them. -- Samuel Wantman 10:43, 1 February 2007 (UTC)

Category types = Good
First of all, I really like the guidelines you've put together here Samuel. Whoever said "This is a bad, inflexible, prescriptive system, patronising and a waste of space" is being intellectually lazy and simply doesn't want to try and improve on the current system. I'm trying to leverage the power of all the information stored in the category links but without these distinctions it is extremely difficult.

I think that this simply standardizes the way that categories are already being used without ruling out new possible uses or even requiring a category to have a type and any point in time. I'm sure concerns over the display templates can alleviated; I agree that as is, with the "see this and that" bits, they are probably too large for the top of some pages.

Froggy 000 04:37, 6 February 2007 (UTC)

Too confusing
I don't specialize in categorization, but there seem to be a lot of categories of categories here, and it might do more harm than good to artificially place a level on each category. Any given category will have some subcategories and/or supercategories, and people can navigate through. Also, it will be time-consuming to implement, and it creates one more backlog for untagged categories. But it's not a bad idea - I just question its practical benefit. YechielMan 19:56, 20 February 2007 (UTC)
 * Yes, the way to find out what use a category might contain is to look at what it does contain, not to read a standardised description which may or may not address the relevant points. Wimstead 18:44, 1 March 2007 (UTC)

Great idea
I think this is a great idea. The category types make sense and the suggested templates will help visitors to navigate Wikipedia. The only downside I can see is that this should take a relatively low priority and should (as was suggested above) be a guideline rather than a firm policy. Categories that already have a lot of information in their headers either don't need the template added or should be tidied up to make room for it. Waggers 09:57, 28 March 2007 (UTC)

Consensus - Rejected
It looks like this proposal has is not proceeding and consensus for acceptance is not evident. This should be marked as rejected ASAP. --Kevin Murray 01:38, 27 April 2007 (UTC)

With no objections or progress it is marked rejected. --Kevin Murray 12:39, 18 May 2007 (UTC)
 * That is ridiculous. This is ongoing.  I just haven't had time to do anything about it recently.  -- &#x2611; Sam uelWantman 19:04, 21 May 2007 (UTC)

Index categories on controversial topics
As we all know, there are a good many controversial categories. There are a number of guidelines that address this, although their intent is a matter of interpretation. Examples include:
 * Categorization
 * &sect;2.1 Some general guidelines, point 8 &mdash; "Categories appear without annotations, so be careful of NPOV when creating or filling categories. Unless it is self-evident and uncontroversial that something belongs in a category, it should not be put into a category. A list might be a better option."
 * Categorization of people
 * &sect;1 Biographies of living people &mdash; "Category names do not carry disclaimers or modifiers, so the case for the category must be made clear by the article text. The article must state the facts that result in the use of the category tag and these facts must be sourced." and "Caution should be used in adding categories that suggest the person has a poor reputation."
 * &sect;4 General considerations &mdash; "For some 'sensitive' categories, it is better to think of the category as a set of representative and unquestioned examples, while a list is a better venue for an attempt at completeness. Particularly for 'sensitive' categories, lists can be used as a complement to categorization." and "Exert extra precaution with regard to the categorization of living people"
 * Categories, lists, and series boxes
 * &sect;1 Categories &mdash; "Categories appear without annotations, so be careful of neutral point of view (NPOV) when creating or filling categories. Unless it is self-evident and uncontroversial that something belongs in a category, it should not be put into a category."
 * Biographies of living persons
 * &sect;6 Use of categories &mdash; (Same text as Wikipedia:Categorization of people, &sect;1 Biographies of living people.)

Given the frequent mention of lists in these instances, it's clear that what's really being discussed is what this proposal terms an index category. I think that the best way to satisfy these portions of the guidelines &mdash; and eliminate a huge amount of conflict between editors &mdash; would be to severely restrict (or outright prohibit) index categories in the case of controversial topics. This is actually already the case for categories involving religious belief or sexual preference, as applied to biographies of living people. I suggest that we simply generalize the restriction in a way that makes sense, rather than continue to deal with these issues on a case-by-case basis. In order for this generalization to make sense, we would need to establish a taxonomy of categories similar to that described in this proposal.

Note that most of these categories could either be modified or merged into a different topic category, so the intent of the category could be preserved. For instance, everyone listed under Category:Pseudoscientists could uncontroversially be recategorized under Category:Pseudoscience, which is a topic category. I claim this would be uncontroversial because anyone even accused of being a pseudoscience (as long as the accusation is notable and referenced properly) is related to the topic in virtue of that accusation. At the very least, it would completely eliminate the problem of each side trying to "prove" one way or another whether the person really is a pseudoscientist, and focus the editorial debate on the more appropriate issue of cited references.

I would like to modify this proposal to be more in line with the desires of the other editors who have expressed an opinion here, and to emphasize the problematic nature of controversial index categories. I'd like some feedback here before proceeding, however. --Sapphic 01:13, 10 June 2007 (UTC)
 * I agree with what you are saying about index categories and controversy. The impetus for these templates was to try and distinguish between the different ways that people are using categories so that we can make decisions about when it is appropriate to use one type of category over another.  It might be very helpful to explain how typing categories might help avoid some of these controversies.  -- &#x2611; Sam uelWantman 06:20, 10 June 2007 (UTC)

Not all categories need this label
One thing that might help is to point out that the labels are not intended to cover all categories. There may be categories that don't fall neatly into one of these types, or are still developing a clearer structure, so it should be emphasised that these tags should only be used in clear-cut cases. Carcharoth (talk) 13:07, 7 February 2008 (UTC)

An active proposal again
There is related discussion currently at Wikipedia talk:Categorization. -- &#x2611; Sam uelWantman 06:47, 8 February 2008 (UTC)

Many months later, it is again being discussed on the mailing lists:. Just fyi. -- Quiddity (talk) 22:30, 4 November 2009 (UTC)