Wikipedia: Wikipedia subpages pros and cons/Evaluation

Feel free to offer evaluations of the arguments on Wikipedia subpages pros and cons here.

I didn't want to add my evaluations of some of the alleged advantages and disadvantages of subpages to the main page, so I made a subpage. ;-) Frankly, I wouldn't mind seeing subpages for non-article pages, such as the "Wikipedia:" namespace. But we could easily do without them there, too. I deny that their convenience for purposes such as this discussion constitutes a very good argument for thinking they're a good idea for encyclopedia articles. (Why would it?) Below are my evaluations of the "pro subpages" arguments. --Larry Sanger

Pro subpages

Ease of linking related data together

Helps link together related data: subpages can be used to divide an otherwise long article into sections; so can ordinary pages, but with subpages, the sections are connected automatically by being subpages. Britannica uses subpages for this purpose.
- True! The original version of this point did not acknowledge that ordinary pages could be used for the same purpose--that in itself does not constitute an advantage of subpages. Therefore, the only advantage subpages have over no-subpages is the fact that sections are "connected automatically." Now, is that actually an advantage? Maybe sometimes, yes, a weak advantage. Sometimes it isn't, and that's one of my points. In many cases, the subpage creator notwithstanding, we shouldn't have a "hard-wired" link from the subpage to the main page. This is inherently prejudicial, really, in that the subpage mechanism leads people to contextualize the subpage as part of, or somehow dependent upon, the main page. This is explained in "contra subpages." --LMS
- Moreover, it is highly misleading to say that Britannica uses subpages. Britannica has one large article and divides it conveniently into sections, instead of putting it on one page. That is not how subpages are (have been) used on Wikipedia. It's hardly as though anyone has written a twenty-page article and used subpages to break it into bite-sized chunks. In fact, mention of Britannica's use of subpages is more obviously a point against subpages: to find information about a specific topic, I've observed that often one has to search through a long article, with a lot of information not immediately relevant to the topic in question. Why should we want to emulate that? Wikipedia is not paper, and Britannica obviously chose their article chunking feature in order to facilitate the inclusion of long paper articles in electronic form. --LMS
Subpages can also be used to facilitate linking to individual sections and between sections.
- This, again, assumes that subpages are being used to create sections of one long article. If they were always used for that purpose, my objection to subpages wouldn't be quite as strong. But we can't dictate that people will use them only for that purpose. More importantly, Wikipedia isn't paper--it's hypertext. Why not let hypertext (i.e., internal, intelligent, human-created links) specify the relationships between chunks of information? There are very few information chunks that belong in an encyclopedia that are longer than our page length limit. Long articles should be broken into chunks. --LMS
Subpages can also be used to create automatic links from the child to the parent and from a parent to the list of children; these links, appearing in a linkbar or other special place on a page, stand out and provide a useful, yet non-obtrusive, reminder to the reader of what "main" connections of the current page, in some useful sense of the word.
- Again, I have serious doubts that this is always actually an advantage. Sometimes, it's precisely my point that we don't want such links, because they are positively disadvantageous (as explained in "contra subpages"). --LMS

Other advantages

Provides a useful home for data that wouldn't make sense on its own: subpages can be used to store small or large amounts of data about a subject that could be useful but would clutter the main page about that subject.
- I'm not sure how to evaluate this without further elaboration of what "data" means here. But if "data" includes ordinary text of the sort one finds in an encyclopedia, we must bear in mind that Wikipedia is going to be huge in the end. There are going to be articles on the most recherche of academic topics, which necessarily cannot be understood without huge amounts of other knowledge. Should we, on those grounds, make those many articles subpages of some slightly-less recherche academic topics? No; the decision in many cases would be totally arbitrary what topic to place them under. I can think of plenty of examples from philosophy where this is true, but it is also evidently true in many other subjects, such as history and physics. I think it is essential that each article begin with a definition that explains what the article is about; in most cases, that definition will, if correct, include something like a genus and differentia (see genus-differentia definition) and therefore point the reader to pages about more general and thus hopefully more accessible topics. Similarly, when an article is about a person or a place or an event, broader, more encompassing categories will be mentioned in the description of the person, place, or event, therefore leading the reader back to more accessible topics. --LMS
- Now, compare that system to a system that contains a lot of parent pages and subpages. In that case, quite often, one must visit a parent page in order to put a subpage into context (as I have had to do with Britannica articles), when this contextualizing can and should be done in the article about the topic I'm interested in. Since it's a subpage, the contextualizing is left to the parent page. In hypertext, that's a bad thing. Let people find the information they need efficiently. --LMS
Similarly to the foregoing, subpages can be used to create small sub-articles that are puzzling as stand-alone encyclopedia articles, but which make sense qua encyclopedia articles as subpages of a main article.
- Understood correctly, this is a point against subpages. It's bad that subpages have been written so that they can be understood only in the context of a main article. What if someone could have written that same article, now on a subpage, so that it was a nice stand-alone article, with just a paragraph of stage-setting? Our readers would thank us for that, I think. This gives the reader freedom: the reader can, if desired, get more background information from articles on more general topics. Look, there's background to be given about everything. That's what an education is all about, giving people background about everything. It's arbitrary to arrange articles so that some of them (the subpages) require that one get background from some specific article (the main page), while others, which could just as easily be made subpages of some other article, are not so arranged.
Established habit: they're known and used in the wikipedia community, removing subpages might cause confusion among those who have used them and who have not practiced writing pages without them
- This is a very weak advantage. I'd say there are rather few people who have been writing on Wikipedia who can't think up all sorts of ways to avoid using subpages. --LMS
Makes for concise titles: subpages convey the most information most concisely: for instance [[Algeria/Government]] vs. [[Government of Algeria]] or [[Algerian government]]
- This is no advantage at all. We get conciseness at the expense of clarity, as argued in the "contra subpages" section.
Useful for fictional universes and some other topics: subpages are particularly useful for collections of articles that have complex interrelations but very few if any relationships to topics outside the collection. For example Dungeons and Dragons and Lord of the Rings; perhaps poker.
- This is true, but everything is related--even nonexistent things! (In ontology, I'm of the view that there are things that do not exist. :-) ) Hence, we could relate Gandalf to other fictional wizards in an article about fictional magic, and wouldn't that be a great article indeed. But to make that article, we would be ripping each individual wizard out of his own main page-subpage universe to compare them. Many other such cross-universe discussions are possible. Anyway, for me, this is the least controversially positive advantage of subpages, because usually, when we speak of Gandalf, it's in relation to Frodo and Sauron, rather than in relation to a Wizard of Earthsea. As such, it's not a very strong advantage. Moreover, if we allow people to use subpages for this purpose, they'll use them, as they have used them, for a lot of bogus purposes as well. --LMS
Can be used to create standardised organisation of the same kind of relationship; for a trivial but by no means exhaustive example, consider "X/Childhood" in a biographical article versus competing schemes "Childhood of X" and "X's Childhood" creating confusion and unnecessary complication. (It seems however that all three schemes are equally arbitrary and one could standardize on either one.)
- This just doesn't make any clear sense. Put differently, I could just as easily have made it a "contra subpages" point. There are going to be competing schemes with or without subpages. E.g., we can just as easily imagine "X/Childhood" as "X/Upbringing" and "X/Childhood and Youth," etc. Besides, we shouldn't make this decision based on what can be easily standardized: we aren't standardizing yet and nothing about the software or our habits militates against some future standardization. --LMS
  - I disagree and believe it does make clear sense. Put simply, pick any 100 personalities on whom we'll have large biographical articles. In absence of any standartization (and I'm not saying these things can't be standartised, see below) and in absence of subpages, perhaps half of them will have "Childhood of X" articles and the other half "X's Childhood", or some other proportion perhaps. I think we can agree that that would be undesirable? With subpages, this particular confusion does not arise: "X/Childhood?" it will be, "Childhood/X" is clearly absurd and will not be used. Now it's true that someone may use "X/Childhood? and Youth" or whatever, but that's not an argument against subpages: the very same confusion may arise without subpages, with "Childhood and Youth of X", "Upbringing of X", etc. etc.
  - To sum up: whenever a page's title needs to transmit the idea of "the aspect B of A", where A is clearly the primary object and B is clearly its aspect, as in X and their Childhood, subpage-based naming allows us to unambiguously present the relationship without leading to likely confusion between different schemes in the English language, such as "B of A", "A's B", etc.
  - Now, in absence of subpages these things may of course be standartised. I'm not saying they can't be, but it's additional work for something we already have for free with subpages (which of course has other advantages in my opinion). And this standartization may prove difficult because of the multitude of different kinds of relationships we'll have to standartize on. Suppose we agree that "X's Childhood" is better than "Childhood of X"; but what about when X is country -- "Geography of X" does seem to be slightly better than "X's Geography". We'll have to consciously decide on any such issue and wade through existing pages, fixing their titles - one-choice-for-all is unlikely to work. Retaining subpages, on the other hand, eliminates this particular problem completely. --AV
    - Re "perhaps half of them will have "Childhood of X" articles and the other half "X's Childhood", or some other proportion perhaps. I think we can agree that that would be undesirable?" First, a strictly speaking irrelevant point: it's slightly undesirable, but not nearly as undesirable as an attempt right now to impose a standardization. Now more to the point, this is clearly not going to be an advantage of subpages or of not having subpages. Subpages do not require any particular standardization scheme, nor do they make any particular standardization scheme any easier. The fact that, in your example, there are more viable options (X's childhood" and "childhood of X") than the same variables "X" and "childhood" in combined to make a viable main page-subpage combination ("X/Childhood?"), is not very convincing. In the long run, subpages make standardization harder because one must come up with standards for making subpages, and as I've argued, subpages are inherently arbitrary, since everything in an encyclopedia can be made a subtopic of something else. --LMS
    - Re "whenever a page's title needs to transmit the idea of "the aspect B of A", where A is clearly the primary object and B is clearly its aspect, as in X and their Childhood, subpage-based naming allows us to unambiguously present the relationship without leading to likely confusion between different schemes in the English language, such as "B of A", "A's B", etc." The problems are precisely that "A/B?" does not always convey "the aspect B of A" in the context of Wikipedia--contrary to what you said, and your specific example not withstanding, the slash does not unambiguously present the relationship. I propose to use English, for clarity. "X/Childhood?" is understandable because of the meanings of the words involved (people have childhoods). In the titles of other possible subpages, such as Electromagnetism/Charge?, the slash has no such clear meaning, and is unacceptable precisely because it is ambiguous. --LMS
Can be used to separate out meta-pages from the contents of the encyclopedia proper.
- This is not an advantage specific to subpages. In Magnus's PHP wiki software, theoretically, we could get rid of subpages entirely while still, as we are planning to, using a "Wikipedia:" namespace for Wikipedia-related articles. --LMS
Autogenerating subpage lists: a nice feature would be to autogenerate a list of links to subpages on each page that has them. We haven't done this yet, but it would be nice. (It's in the PHP script; try [1])
- This is an advantage only if subpages per se advantageous to have; since they're not, it's not. --LMS

I offer here my evaluation of the contra-subpages arguments.

First, I submit that they can be usefully divided into the following two groups:

1. Arguments which argue that subpaging is an inherently arbitrary choice, and can be easily abused.

2. Arguments that argue that subpages are wrong for other reasons.

It is my belief that the arguments of the first kind are generally correct, but the danger presented by the arbitrariness is, first, greatly overstated, and second, not unique to subpages anyway. The arguments of the second kind, in my opinion, apply only to particularly bad choices of subpages, in other words, to subpages which shouldn't have been subpages in the first place and which therefore actually present evidence for arguments of the first kind. I'll argue with examples that most or all of the arguments of the second kind are powerless when confronted with actual examples of good use of subpages, some of which are even linked from this discussion.

Now, to the arguments.

Decisions on when or where to create subpages at all is necessarily arbitrary: every encyclopedia topic can be regarded as a subtopic of another encyclopedia topic. There is no good reason for us to regard some topics as subtopics of other topics when all encyclopedia topics can be so regarded.

In this paragraph, only the first clause is perfectly correct: Decisions on when or where to create subpages are necessarily arbitrary. This is an important point to keep in mind. However, are they the only kind of decisions that are necessarily arbitrary in Wikipedia? What about decisions on how to divide large text into multiple pages (subpaging or no subpaging)? What about decisions on how to name articles when several perfectly fitting alternatives exist? What about decisions on how much to quote from historical accounts, on whether or not to include relatively short original sources, or relatively long original sources for that matter? What about decisions on where to use italics and where to use a bold font?

These are all examples of issues that are not standartised in Wikipedia and are being addressed by contributors as they work on articles. Consensus slowly emerges on some of these issues as more people see various ways of dealing with the problem and come to agree on the best ways. The issue of subpages isn't that different. It's also a feature of Wikipedia, one that can be used and abused, and one that is widely used by Wikipedians and appreciated by at least a fair number of them. The argument that subpaging is arbitrary will only apply if its proponents show that this arbitrariness, unlike other cases of arbitrary choices in Wikipedia, has really harmed the project, for example, by creating large controversies and edit wars.

So far this hasn't happened, however. There are occassional arguments about whether some page is better off as a subpage or as a separate page, but no more arguments arise about that than about other choices Wikipedians are confronted with when editing articles. Moreover, very many articles, some of them famously (September 11, 2001 Terrorist Attack comes to mind) exist as conglomerates of many subpages, and seem to be extremely useful; noone seems to suggest that these particular articles are better off being refactored as single pages. There're calls for general abolition of subpages, sure; but noone seems to justify these calls by practical consideration of existing successful subpages (such as the one I just referenced) and showing how it would be better if it were refactored as a collection of pages without subpaging. It appears that such a demonstration is needed for the argument to be persuasive.

The rest of this argument, beyond the first clause:

every encyclopedia topic can be regarded as a subtopic of another encyclopedia topic. There is no good reason for us to regard some topics as subtopics of other topics when all encyclopedia topics can be so regarded.

is unconvincing because it sets up a false black-and-white picture. It is simply not true in any useful sense of the phrase that "every encyclopedia topic can be regarded as a subtopic of another encyclopedia topic".

Consider the following three pages: Chess, Chess/Rook (this page explains what a rook is in chess and how it moves), and [History of Austria]?. It strains credulity to suggest that the contents of Chess/Rook can be considered a subpage of the contents of Chess in the same degree that the contents of [History of Austria]? can be considered a subpage of the contents of Chess. It is simply not true in any but the very scholastic and tautological point of view.

In fact, what we see is that some pages lend themselves very strongly as natural candidates for subpages of other pages (such as "Rook" for Chess, or "Missing Persons" for September 11, 2001 Terrorist Attack), while other pages don't let themselves at all as natural candidates for subpages of other pages (such as 48 for Emacs or Charlemagne for [Electric Toaster]?). Now, to be sure, certainly there is a grey area in-between! But existence of such a grey area does not mean that "every encyclopedia topic can be regarded as a subtopic of another encyclopedia topic", nor does it mean that we should abolish subpages. It merely means that policy will slowly emerge through common work regarding what is and what isn't good subpaging -- just as such policy has arisen in many other cases with grey areas in Wikipedia.

Next argument:

The particular choice of a subpage hierarchy is arbitrary: [[Algeria/History]] might be used when [[History/Algeria]] would be as just as appropriate; both [[Film editing/Star wipe]] and [[Digital effects/Star wipe]] refer to the same thing and would be equally appropriate. There's no clear principles on which to make the decision, and the decision does have consequences.

I agree that a particular choice of a subpage hierarchy may be arbitrary, but this merely means that in such a case subpaging may not be appropriate (see above for explanation why existence of such cases doesn't constitute an argument against subpages). In fact, of two examples presented here, one seems very unconvincing - would anyone really entertain the possibility of [[History//Algeria?]]? - of course not! - gathering histories of all countries as subpages of one page is evidently an absurd idea, and it is clear, I trust, that every experienced Wikipedian won't think a second before choosing Algeria/History as the correct hiearchy (and while a very inexperienced Wikipedian may make a wrong choice, they make wrong choices in all kinds of other issues as well - that's why more experienced Wikipedians are there to correct them!). The second example is convincing, but it merely shows that Star wipe should be an article unto itself, that's all. If it can be considered a subpage of two quite different pages equally well, by all means don't make it a subpage at all.

Now let us examine this argument on some real and successful uses of subpages:

Is it September 11, 2001 Terrorist Attack/Missing Persons or [Missing Persons/September 11, 2001 Terrorist Attack]?? I think the answer is clear. Even more absurdly, is it [September 11, 2001 Terrorist Attack/Full Timeline]? or [Full Timeline/September 11, 2001 Terrorist Attack]?? It's silly to even ask the question.

Is it King/Chess? or Chess/King? Would anyone claim that there will arise any substantial disagreement over the decision to adopt the latter but not the former?

We see, by examining actual uses of subpages, that the issue of dubious hierarchy almost never arises; when it does arise - and that, I repeat, almost never happens in practice - it should simply be taken as evidence that this particular subpage should be converrted to a page of its own.

The argument would be an argument against subpaging in general only if it could be shown that in a substantial number of existing uses of subpages the hierarchy isn't clear and is arbitrary. But that is manifestly not the case.

To be continued tomorrow -- AV

AV, I'm not particularly interested in the subpage debate, I think we can write good encyclopedia articles with our without subpages, so the whole debate is not as significant in my mind as it seems to be for some other folks. So, I really don't want to get enmeshed in a debate which takes up time I could be fixing up the List of philosophical topics. I'm not going to get into the debate here, but I just want to point out that I think you have misunderstood at least one of Larry's arguments. When Larry says, "any encyclopedia topic can be considered as a subtopic of another encyclopedia article," he is not arguing that any encyclopedia article could be considered an subpage of ANY other article. To use your example, could not chess be under Board Games/Chess?. Larry wants hierarchy's, but he wants to allow for as many of them as possible, and to avoid setting up "hardwired" hierarchy's that keep you from putting Chess under Board Games, Strategy Games, Intellectual Competition, and wherever it happens to belong. Once chess is placed under Board Games, everybody and everything has to know about that specific hierarchy. Now before you get upset, I know that what I've said is only an argument for using subpages responsibly, and not necessarily an argument for removing the possibility of subpages. To do that, you'd need to couple it with another argument that either that 1) it is very difficult or even impossible to use subpages correctly, or 2) that using subpages correctly is significantly more difficult than the alternatives. And I'm not prepared to do that... MRC

: You're right, I did misunderstand the argument. For some reason I kept seeing "any other" where it was written "another". I'll rewrite that part of my argument. My apologies to Larry and many thanks to you -- AV