"Subcategories" are not subcategories

+5 votes
202 views
Definition of a subset: "A is a subset of B" means that all members of A are also members of B.

Wiki so-called "subcategories" aren't like that at all.  If category A is a member (ie a so-called "subcategory") of category B, the members of A aren't members of B at all.

Unless they're explicitly made members of both.  But that's banned by the lowest-level rule.

So, wiki so-called "subcategories" aren't remotely like real subcategories in the normal meaning of the word.  They're just member pages that happen to be of category type.  We need to be careful not to confuse the two meanings.

If we had true subcategories, it would make sense to have a hierarchy and a lowest-level rule.  But in the system we have, these things don't make much sense at all.

For instance, Bill Smith lived in Richmond VA, so he's duly added to Category: Richmond, Virginia.  But then we find he was buried in Dunromin Cemetery in Richmond, so he goes in that category.

But that category is a member (so-called "subcategory") of Richmond, Virginia.  So the lowest-level rule says Bill should no longer be a member of Richmond, Virginia.  But he could stay if he were buried outside city limits.

This means there's no sensible or fixed or useful notion of who should be a member of Richmond, Virginia.

It also means it's harder to find Bill, because there might be a lot of subcategories and maybe sub-subcategories to look through.
in Policy and Style by Anonymous Horace G2G6 Pilot (568k points)
It's really harder to find Bill because his name is Bill Smith.

(More serious answer below. I have Smith challenges in my tree, so I find that more of a challenge than categories are.)
I also have a feeling that this rule has something to do with technical aspects and server space, but I'm only guessing on that.
"Lowest level" is subject to some interpretation. Bill shouldn't be categorized in United States or Virginia, because placing individual people at that level would clutter those categories and make them useless for people trying to find anything specific. But at lower levels, there's room for some flexibility. The individual cemetery is the lowest level of the Cemeteries category structure, and the cemetery category logically also belongs in Richmond because the cemetery is there. However, some of us are more interested in people than in cemeteries, and the geographic categories have to accommodate people for reasons other than where they are buried. Since Richmond is logically the lowest-level category in the geographic structure, then Bill belongs in that category.

I believe Ellen's lowest logical level approach for the event is right. Bill lived in Richmond, Virginia so that category is appropriate. He was buried in Dunromin Cemetery so that sub-category is appropriate too. However, just because someone is buried in the cemetery doesn't mean that they should automatically be added to the Richmond, Virginia category too. For example, some people are buried in family plots but they may always have lived in a neighbouring city or township, or even much further away.

Bill is not a category (or a subcategory). He's a profile. See my answer below.

And you're defining a subset in the formal mathematical logical sense, and saying "look, a formal logic subset is different from a WikiTree category". I'm sure you'll find that no one will disagree. That's kind of not the point of categories. Ellen explained it with admirable clarity.

Unless we know what hospital he was born in then we might put Bill Smith in, say, the Chippenham Hospital, Richmond, Virginia category;  laugh

Problem is, people are talking about wiki subcategories as if they were mathematical subcategories.  They're talking about directed acyclic graphs and stuff.  That's the other sort.  It's not applicable to the wiki sort.

For the wiki sort, cycles do make sense.  Eg, Mis-spelled Categories and Maintainance Tasks could be members of each other.

I totally agree with your point, RJ,  that these aren't formal mathematical categories. And that there is no logical reason that (e.g)  Mis-Spelled Categories, and Maintenance Tasks can't be even described as sub-categories of each other. 

:) But as someone who did refer to directed acyclic graphs, I (nondogmatically) disagree that cycles are ever a good idea. I think, for pragmatic reasons (not mathematical) that having a clear, directed, hierarchical (acyclic) ordering across all the (official) categories will make everyone's life easier in the end (this is a discussion for another time and place, but I'd absolutely love to start using categories for smarter searches than WikiTree currently allows).

edit: There are plenty of easy navigational workarounds, that you can use so you don't need to have cycles. You probably know already that we can have links between categories in the explanatory blurb, like "See also several common maintenance categories, like [[:Category: Mis-spelled Categories| Mis-spelled Categories]], etc". But again, this is sort of beside the point.

5 Answers

+8 votes
 
Best answer
In the hyperlinked world of the Internet, information no longer needs to be organized in a linear hierarchical structure. And us older folks will have to get over the notion that the way things were 50 years ago is "normal," because the world has been changing around us.

A single category can have multiple parent categories, and different users of the category structure can view the categories via different category relationships that I will call pseudo-hierarchies. That Dunromin Cemetery has a parent category of Richmond, Virginia, making it part of a geographic pseudo-hierarchy, and it also has a "Cemeteries" category as a parent, making it part of a pseudo-hierarchy for cemeteries. And nobody should feel a need to argue about whether the geographic category structure or the cemetery category structure has primacy, because both can co-exist as functional structures.

And there's no need to insist that Bill Smith can't be in both the Richmond, Virginia, category and the Dunromin Cemetery category, because a profile can be in multiple categories. For many people, his role as a citizen of Richmond may be more important than the location of his gravestone. Placement in multiple categories should make it easier to find Bill, not harder.
by Ellen Smith G2G Astronaut (1.2m points)
selected by Martin White

Ellen, you’re killing me! “And us older folks will have to get over the notion that the way things were 50 years ago is "normal," because the world has been changing around us.“ 

Do I gotta? Really? sad

LOL. A lot of us here are "older folks". I was just reminded of that by the size of the fire on my birthday cake a few weeks ago.

Did you at least have the Fire Department on standy? angel

winkKids these days!!!

Aint it the truth, Natalie?!
Hey Pip! Look on the bright side. We only gotta get used to the world changing around us for another 50 or 60 years, and then we can relax. (this, according to my boss Sergey is "Russian optimism").
+5 votes

For instance, Bill Smith lived in Richmond VA, so he's duly added to Category: Richmond, Virginia.  But then we find he was buried in Dunromin Cemetery in Richmond, so he goes in that category.

But that category is a member (so-called "subcategory") of Richmond, Virginia.  So the lowest-level rule says Bill should no longer be a member of Richmond, Virginia.

This is incorrect. You are confusing two separate structures that happen to be linked to each other. Bill would be fine as both a member of both Category:Richmond, Virginia and Category:Dunromin Cemetery, Richmond, Virginia. There is no rule in place saying that a profile cannot be members of multiple categories.

The policies state that profiles should be added to the lowest category possible. So, Bill would be added to Category:Richmond, Virginia and not Category:Virginia. By being added to Category:Richmond Virginia, he is in fact, also a member of Category:Virginia, and can be accessed through navigation through Category:Virginia. Just because he is not linked directly to Category:Virginia, does not mean that he is not associated with Virginia.

So what you are implying is that because I live in the city of Houston that I cannot also live in Texas?

by Steven Harris G2G6 Pilot (503k points)
Texas is too broad a category to be meaningful for an individual person. (For that matter, Houston is also too big, but that's a different issue.) So, if you live in Houston, you can be in that category, and if you have affiliations with other Texas topics (maybe you used to live in Waco and you went to Texas A&M), you could be in their categories, too.
I agree Ellen, that is the point I was making.
+5 votes
As Steven noted, these are two different category streams.It's true that your cemetery is subcategorized under the location [[Category:Richmond, Virginia]] but this is how we categorize US cemeteries, both under the county cemeteries and the city/town location. If Bill had lived in another city at some point, that could also be a category on his profile.

The category heirarchy is more like a web, in my mind's eye. If you start at the [[Category:Categories]] and start poking around from the top level, you can see it.
by Natalie Trott G2G6 Pilot (831k points)
+4 votes
The "lowest level" rule really only means "lowest level of a type."

If one is dealing with type of category called location categories, Bill Jones, who was born, married and died in Richmond, should only be categorized under Richmond, Virginia, and not also under Virginia and not under United States.

But cemetery categories are a different type of category.  Bill is buried in the Dunromin Cemtery, which is part of a cemetery hierarchy of categories. The Dunromin Cemtery is also categoried under Richmond, Virginia because it is there, even though it is a cemtery and not a person.

We have other types of subject matter in categories.   Many are distinguished by geographical location.  There may be Notables in Richmond Virginia, and if Bill Jones is Notable, he would be categorized there, too.  The fact that the Notables category for Richmond is also shown as a subcategory of the location Richmond, does not change anything regarding lowest level category.
by Jack Day G2G6 Pilot (363k points)
+2 votes

You're completely confusing pages, profiles, and categories. 

So, wiki so-called "subcategories" aren't remotely like real subcategories in the normal meaning of the word.  They're just member pages that happen to be of category type.  We need to be careful not to confuse the two meanings.

Categories are not pages or profiles. Categories are a totally different data structure. They are simply a means of defining relationships among things. Categories are meant, with some exceptions, to be directed acyclic (to crib a term). That is, they should have a clear ordering, from less to more specificity, and no category should ever be in a circular relationship with itself. (Analogous to a family tree, where no one should be his own grandfather.)

The graph of categories, in practice, can be a little more complicated than a simple tree, but still, it should always be directed (higher to lower) and acyclic (no loops). This ordering by specificity is the key to having a functioning set of categories. This is why we can speak of subcategories, or child categories, or supercategories, or parent categories.

Profiles are not categories or pages. They should (we hope!) be in directed acylic graphs with each other (again, no time travelling grandmothers). But they can be categorised under any number of categories. That is, people are complicated, and can be in a lot of bins, based on where they are born, where they die, how they die, the religion they practice, what chess club they belonged. This is why Bill might be classified by the category 'Richmond, Virginia' (because it's the city where he was born), and 'Dunromin Cemetery, Richmond', because that was where he was born. It's completely consistent with the guidelines.

Space pages are not profiles or categories. They are similar to profiles, but again, they have very different uses. They might be used to explain groupings, or organize profiles, or describe things. But they are effectively just all-purpose explainers that we tack on to categories and profiles in an arbitrary way, to add value.

I think most of the disagreement from the discussion last night, and here, stems from this misunderstanding of what categories, pages, and profiles are.

by Brad Foley G2G6 Mach 4 (48.6k points)
edited by Brad Foley
There's also another, related, issue about category structure, in that it would be nice if there were a single, logical path from the highest-level categories to each end-point... And I thank everyone in the categorization project for trying so hard to make this happen. Probably, in practice, we'll end up with a category like "Methodism in Newfoundland" at the end of a geographic and religious branch. But I think a lot of the debate (again from yesterday) is to what degree there should be multiple categories like "Phillips Name Study", "Phillips Family Study", "Phillips ancestors to me" scattered around willy nilly, under personal, geographic, one name study or other branches and personal category trees.

Not that again!laugh

Also with the caveat that you can do whatever you like with personal categories, and impost whatever structure you want. As long as it doesn't create database errors, and can't be mistaken as an official project category. That's why it needs to follow personal category naming conventions.
:D What do you mean, Natalie? I hope I didn't go on for too long in my answer ... but this discussion seems to have spilled over from yesterday. I'm hoping that explaining the terms clearly will help clarify why the rules are what they are.

:'( I might have gotten too stuck into it, though. I'm studying for data science job interviews, so I'm probably a little *too* passionate about data structures and graphs right now..

yes It's a great answer, Brad.

(phew. crisis averted)
No I am not confusing anything.

There are categories, and there are category pages.

Normally they're linked by having the same name.  But either can exist without the other.

A category is just the set of (member, category) pairs.  If it's empty, it disappears.  It can't have a parent or anything.  But it can exist (ie have members) without a page.  Those that do are in the misnamed WantedCategories list.

A category page is just a normal vanilla wiki page like any other.  You put wikitext on it, which might include the code to make it a member of a category.

The difference is, when it's displayed, if the corresponding category exists (has members), a list of them is pulled out and added to the page.

But there needn't be any members.  Pages without members don't appear in Special:Categories.

The disagreement was about the notion that a personal project has to use personal categories.  Personal categories are essentially private categories, in principle, though not technically.  But a personal project isn't necessarily a private project.

But nobody will discuss that.  All we've had is a cosy clique closing ranks and ganging up to produce a barrage of irrelevant and imaginary technicalities.

 All we've had is a cosy clique closing ranks and ganging up to produce a barrage of irrelevant and imaginary technicalities.

What we've had is a bunch of people who've worked hard coming up with a workable and logical system of categories, trying in five different ways to explain why enforcing a simple, hierarchical structure of categorization makes sense for wikitree projects as a whole. I don't understand why you keep bringing up technicalities as if they matter to the point you're making, and then get mad when we try to explain why we think these technicalities are irrelevant to the argument.

You're right that the abstract notion of a category is different from a WikiTree category page. But that's a technical distinction that has no bearing on whether official WikiTree project categories should have a hierarchical structure, and whether personal project categories that don't conform to the official guidelines should clearly labelled as unofficial.

But a personal project isn't necessarily a private project.

Absolutely. But if I want to leave a personal project open to general collaboration, I should be expected to follow the overall (naming and structure) guidelines for projects. I, and other people, have said this repeatedly. There are very good reasons for clearly flagging personal projects that don't conform to these conventions as "personal projects". I've repeated these reasons several times. 

This is what collaboration means---coordination and shared work. Collaboration doesn't mean "everyone does whatever they want, however they see fit". If that were to happen, projects (personal and otherwise) would become unworkable.

I think what we're all trying to say is, "do what you like, as long as you don't make a mess other people will have to clean up."

A category is just the set of (member, category) pairs.  If it's empty, it disappears.  It can't have a parent or anything.  But it can exist (ie have members) without a page.

So. I agree that this is the formal, logical, definition of a category (in formal, logical, mathematical contexts). I'll point out that no one else is talking about categories in this way, here, or in previous discussions. But, I'll try to parse the point you're making. Please, tell me if I misunderstand or misrepresent you...

If category A is a member (ie a so-called "subcategory") of category B, the members of A aren't members of B at all.

I think what you're trying to say is that, everyone who is in the category of "Richmond, Virginia" automatically belongs to the category of "Virginia". And because Bill is also in the category of "Dunromin Cemetery, Richmond, Virginia" he's automatically in the category "Richmond, Virginia", as well as "Virginia".

Yes. Yes, this totally makes sense. 

And (again correct me if I'm mischaracterizing your argument), you take issue with the fact that we're saying that "we should put Bill in both the cemetery and the city category" because that's inconsistent, and unintelligible, from the strict formal definition of categories (after all, he's already in the city category). You, I think, are effectively saying that "because you're using categories in an arbitrary and illogical way, any structure you claim to have is gibberish."

So, yes, I agree. I'd say that sometimes the official project guidelines about how profiles should be displayed on category pages departs from the way they'd be displayed if we were using the word 'category' in the formal logical sense. 

But, for me at least, it seems transparent that most people are using the word 'category' in a different way here. By this, they mean "a collection of category pages, organised roughly around a pragmatic sense of formal categories", and the rules about which profiles should be displayed on which category pages are pragmatically defined as "whatever avoids cluttering too many upper level categories, and also makes it easy to find a profile by descending, in a hierarchical order, down a category tree".

I'm trying to understand how there is anything wrong with this usage. And why the formal usage of category you're insisting on has anything to do with personal vs project vs private categories.

Thank you, Brad. that was well-stated.

A personal project isn't necessarily a private project. Agreed. But when a personal project includes categories that are specific to one family, with a description such as "my grandmother" "my fourth great-grandfather and his siblings" etc, it seems to me that the creator is walling off this category for their own use.  Perhaps that is my mistaken interpretation. If it were to be for the use of all of WT, it would not be so specific. 

A quote from one of the WT team: "Categories are not owned. It is true, that someone initially created it, but it can be used by anyone. So we can't say, it is someone's category, unless it is personal category starting with WikiTreeID." This also means that project members can correct category errors. We try to do so politely, in conjunction with projects or WT users, and ask why something is named a certain way, but in cases like spelling errors or simple punctuation mistakes, it's not necessary. In the case of the informal name study that started the entire conversation, yes, I could have asked the person before I made the change. That won't happen again. 

There's a whole other discussion about when and how people should ask before making sensible and necessary changes. And I don't think, Natalie, that the person whose categories you renamed was upset, after you explained yourself.

I think the current discussion is maybe broader in scope than that other one. I think it's been pretty civil (and, to me, interesting). And everyone has explained their very different points of view in thoughtful, mostly clear, and useful ways. I've learned a lot.

So. I agree that this is the formal, logical, definition of a category

No, I was describing how the system works.  There's a list of categories (Special:Categories) and a list of category pages.  They're different, and neither is a subset of the other.

Categories would work without category pages.  I expect it's a configuration option.

Changes to category pages are logged and appear in the feed, though they're omitted from most views.  Changes to category memberships aren't logged at the category end and can't be monitored.  Of course they're logged at the member end, as text edits.

RJ, I'm not trying to be obtuse, but I'm struggling really hard to understand the chain of logic you're trying to make. It feels like there are three points you've made. And they seem (at least to me) more or less unrelated.

One: how to name category pages. Whether personal projects need to be labeled private projects, if they don't follow standard Wikitree categorisation rules.

Two: a complaint about the standard system of organisation for category pages. I think you're saying that the system on Wikitree doesn't make sense.

Three: category pages aren't categories. But this seems to be a tangential point.

Is this all a fair summary of what you're saying?

As to point three, I guess I'm still confused what you mean when you try to distinguish categories and category pages. Category pages are just the interface for categories. Yes, you can see the list of categories:

https://www.wikitree.com/wiki/Special:Categories

But I don't see how this is any different from a list of category pages.

As far as I understand, you create categories by creating a category page. You define relationships among categories (parent and child categories) by editing category pages. You examine (some of) the properties of categories by looking at the category page. Every category that exists can be accessed using the category page interface.

A category is just the set of (member, category) pairs.  If it's empty, it disappears.  It can't have a parent or anything.  But it can exist (ie have members) without a page.

This is just false. If you want to do a database query of the hierarchy around a category, you can:  http://wikitree.sdms.si/default.htm Just go to the search button, look for categories, and type in the category name (with spaces and commas). Search, and it'll show you the category details (including its different elements like parents, subcategories, and profiles, links to and from), and it will give you the option of viewing the hierarchy.

Thus, if the "category page" is just the interface to the category, the category itself has several essential elements. The category name, of course. But also optional category members (profiles, space pages), parents (supercategories), and children (subcategories). These elements all have different properties, and are critical to the function of categories.

And I don't see why any of this relates to points one and two (above).

Well thank goodness they do not loop - that part DID make me feel old, thanks Brad for snapping us back to where we can use some sense!

Related questions

+2 votes
3 answers
+3 votes
2 answers
101 views asked Aug 28, 2020 in WikiTree Help by Wendy Jones G2G6 Mach 1 (11.5k points)
+3 votes
1 answer
+3 votes
1 answer
+2 votes
1 answer
+2 votes
1 answer
71 views asked Jan 17, 2019 in Policy and Style by Kirk Hess G2G6 Mach 6 (64.2k points)
+5 votes
1 answer

WikiTree  ~  About  ~  Help Help  ~  Search Person Search  ~  Surname:

disclaimer - terms - copyright

...