CZ Talk:Usability

Introduction

"Usability" may not be the ideal term here, as a very significant part of the issues here involve locating material on CZ, and, especially, navigating among potentially related topic. Well-considered navigation methods and practices may be a very distinct advantage of CZ over WP. Usability applies to authors as well as readers; we want to make it as convenient as possible to tie new contributions into the navigation system, and also to add appropriate supplemental information. Supplemental information includes definitions (not restricted to Article/Definition pages), reference information applicable to multiple articles, etc.

Very briefly, the Article/Related Pages subpage is an extremely important part of the conceptual design. I'll simply say that the intended use of Related Pages and its evolved capabilities substitutes for WP Categories. CZ Workgroups may seem superficially like WP Categories, but they are more focused on collaboration than indexing.

Before I go further, note that it is possible to define Article/Definition and Article/Related Pages without defining the main Article.

While there is not an intention to create a hard and fast workplan dictated from the top down, my personal opinion is that there is an increasingly strong consensus, among people who have thought about it a lot (and know things they don't know) that CZ is far more concerned than WP about having "unifying places". Those might be as simple as a Related Articles page with no main article existing, but I think it's fair to say that we do not see search engines as the only significant entry point to CZ. Howard C. Berkowitz 13:38, 1 November 2008 (UTC)

Howard, this is a good start. I think there are two other usability issues that you have not touched upon. 1) Subgroups, the smaller communities that can exist across or within workgroups. 2) The ability to define tabs for subpages that are not globally defined but might be very important for one specific topic.

Another issue for us to realise up front, at least from my perspective as the one that wrote most of the templates, everything is negotiable. The organisation we have now has evolved over time. Many of the issues we discuss now were not even conceived when the subpages were first written. It is possible we need to redesign and that is OK. Chris Day 20:21, 1 November 2008 (UTC)

Looking at some past Forum discussions pro and con categories, and other navigation tools, it occurs to me that some of this, or where the expert guidance can be, is in structuring the elements of various navigational concepts, including ontology (no specific article, but there is Intelligence collection ontology and it might give some ideas about defining scope of structure), taxonomy, database, controlled indexing vocabulary, etc. Sometimes, we can adapt from existing structures, such as Medical Subject Headings (MeSH), but we need not be bound if there are some awkward constructs, or needs for compatibility with old techniques, still there.

As a side issue, there may be some times where we want some software to generate appropriate lists, as, for example, from a very existive "what links to here" display. I see needs, and maybe I could write something; I am not experienced with web-oriented programming languages although I can learn something. What language generates things like the lists under Special Articles? Howard C. Berkowitz 15:32, 3 November 2008 (UTC)

I have no clue with respect to programming. My guess is that Jitse Niesen is the best one to ask. As for using existing ontologies i think that is good to tap into their hierarchy. I have noticed a few Related article pages where I have seen parental topics that appear to be more like subtopics. Likewise "Other related articles" that really should be parent or subtopics. As for naming the articles it is probably not a good idea to restrict ourselves to the ontology list (we agree on this). CZ names seem to be quite different to ontology lists, although there is clearly overlap. In short we are embarking on building a huge knowledge map. I am sure we are reinventing the wheel but possibly that is good since we do not get trapped in preconceived models. We can adjust as necessary to add the best features from knowledge maps. i was playing with drawing a 3D representation of the scheme but discovered it is easier to to do as a 2D scheme. I'll try and draw my doodle as a more formal visual. Chris Day 16:06, 3 November 2008 (UTC)

First, I'm glad I reread that; I was trying to understand why you would redraw your poodle. I do have a cat sleeping on the scanner.

Second, I absolutely agree this is a huge mapping challenge. Nevertheless, giving there is, to mix metaphors, a battle for the hearts and minds of Wiki users. When was the last time a battle was one by the side that didn't have a map?

Why limit to wheels? Tracks, hovercraft, ekranoplanes, transforming land-sea vehicles... Howard C. Berkowitz 16:30, 3 November 2008 (UTC)

Hi guys--I've scanned this page, and in the little time available to me, I have to say I'm having difficulty making heads or tails of it. It's like coming in halfway in a movie; I need more context. Is there a particular idea you're trying to flesh out? If so, can you state it more clearly? Howard, you say, "I'll simply say that the intended use of Related Pages and its evolved capabilities substitutes for WP Categories." Is that the main idea? Well, that is in fact the explicitly intended purpose of Related Articles pages, and indeed, one main purpose of the R template (and the [r] links that the template outputs, which connects up Related Articles pages), as I envisioned it when I first made it, was to enable people to create a whole web of definitions. It's just that few people understood what I was banging on about, or if they did, they didn't care much. See here for the original proposal that led to Related Articles pages, and see here for something earlier and more theoretical. This section of CZ:Related Articles discusses how to build out CZ's category scheme. Also, if you notice in the back-burnered plan for the Big Subpagination, there's a bit we never started, called "The Big Relate." Check it out. Using "what links here" as a way to seed pages, in an effort to build up the category scheme, is potentially a great idea! --Larry Sanger 17:44, 3 November 2008 (UTC)

Larry, if there is a single idea here, it is that an enormous strength of CZ is that it has the potential for knowledge navigation like that of no other public Wiki. That being said, this is definitely an experimental discussion on how best to make use of features.

Related Articles pages are a very major part of this navigation, but, as you pointed, people didn't understand or care at the time. We are trying to find ways to get those ideas across in an actionable way. In some cases, we are literally developing or refining techniques; there is much interaction about how quite a few features do or can interact, but are not obvious. As you may see, putting a definition-only of a redirect to a section of an article into a Related Articles page is quite powerful, but not immediately intuitive.

Speaking for myself, I wasn't even trying to substitute for anything at WP, especially things that aren't terribly useful. Several of us are banging at ways to use not just Related Pages but other navigational aids, some of which might need to be created, or at least documented.

Yes, I agree: we definitely want ways to exploit "what links here". At least of one of my doodle-level pictures is, to a large extent, generated from such lists. Without getting into the specifics, I created, and will create, some articles (e.g., AN/) that do provide information about nomenclature, whose major purpose is to be an anchor point for "what links here" even in its present form. We also have article-free link points of Related Articles, Definition-only, and even more. This is, however, a work in progress.

I'm suggesting it in various talk page discussions that you see here; it's a large part of the guidance I'm trying to offer to one author whose contributions, in one area, don't seem to relate to one another or to unifying concepts. Howard C. Berkowitz 18:19, 3 November 2008 (UTC)

By the way, one simple but useful feature that could be added to Related Articles subpages is, at the top, a link to the "What links here" page. Yes, that's on every page in the cluster, but a lot of people don't know the page is there, and it's especially useful to remind people of it on Related Articles subpages. Just be sure to pay attention to design, keeping it as simple and user-friendly as possible... --Larry Sanger 17:49, 3 November 2008 (UTC)

Just started implementing that via {{Related Articles footer}} but can't call this from {{Subpages}}, since the latter is protected (probably rightly so). Chris, can you finish this? Thanks! --Daniel Mietchen 18:12, 3 November 2008 (UTC)

Daniel you were correct that is the only tempalte you needed to adjust. The reason it did not show up was that you had it in noinclude tags. I have tweaked it some more. Not sure which is the best what links here option. There are two options, what links here to the related articles page is more indicative of the navigation web. Whereas what links here to the article has far more links and is indicative of the more tangled relationships that all the articles have with each other. Chris Day 23:02, 3 November 2008 (UTC)

Thanks, Chris. Next time I shall notice the noinclude tags. --Daniel Mietchen 08:56, 4 November 2008 (UTC)

Chris, I freely don't know which is the best link to show. In the best software tradition, I'd prefer some variant of "you are in a maze of twisty turny passages, all alike" to "tangled" :-)

Seriously, the challenge being addressed here is steering between the Scylla of presenting overwhelming complexity, and the Charybdis (Obama and McCain?) of the loss of the information that is inherent in that complexity. We all have to feel our way, perhaps even two steps forward and one step back, exploring how to approach what I still feel has immense value.

Larry had some great ideas with Related Articles, but, as we are finding here, the best use is not intuitively obvious (a term that can only be sneered properly by one with tenure). Sandy quite correctly brings out the potential of external informational tools. Some of those tools, such many that show which papers are cited most often, are in the "deep Web": the part that is not free. Who knows -- maybe there is support somewhere for the sorts of semantic and ontological research that underlies it. I've worked with semantic tools such as ULMS, and, at present, the more complex tools are simply not going to be usable other than to some people that work extensively with them.

Perhaps a role might evolve, either as a supplemenal function of subject matter expert editors or by having some information sciences people helping with expressing relationships, in addition to the other CZ functions. I haven't forgotten the "testimonial" page, but this is the kind of thinking, which, if it goes anywhere, would put us generations ahead of the Other Place. It also may take a discipline, at least until there are some very smart automated tools, that could be difficult with volunteers.

As an editor, one of my frustrations here is with contributors that don't tie their articles into related ones, and don't see value in doing that. Sometimes, the articles created don't cleanly fit into a knowledge network, and there's pushback on any suggestion they should. It's far too early for policies on this, but it's worth considering.Howard C. Berkowitz 23:31, 3 November 2008 (UTC)

I'll leave the detail to Chris -- very good idea; I don't know if he can do hat or if it has to go into the MediaWiki code. Howard C. Berkowitz 18:19, 3 November 2008 (UTC)

Orphans

Chris pointed out that The Other Place is doing some good work on what it calls Orphans [1], which is just the sort of thing that an enhanced CZ navigation system is intended to avoid. We, I think, are going a bit farther. I am not addressing images at this time.

Orphan status

The Other Place	Citizendium
An Orphaned article is an article with less than three incoming links. Links that do not come from the main space, disambiguation pages, redirects, or members of lists (e.g., list of foo, or event in year) do not count as incoming links.	An orphaned article has less than three incoming links, except that a redirect that provides a definition of a subsection is a valid link.
An Isolated article cannot be reached via a series of links from the Main Page.	An Isolated article cannot be reached via a series of links (including Related Articles pages without corresponding Articles) from the Main Page, a Workgroup Page, or a Core Article. There may be special cases where a valid entry point could be a naming prefix (e.g., AN/).
A walled garden is a group of articles that only link to each other. It is, in effect, an orphaned group of articles.	A walled garden is a group of articles that only link to each other. It is, in effect, an orphaned group of articles. Such a group can include Related Articles pages that do not have any references outside the garden

We want to avoid orphaned and isolated articles, as well as walled gardens.

Navigation, editors and approval

Here's another matter to consider; does it need an EC proposal or just a consensus?

As we introduce these navigation tools, while many of the results of things like well-developed RA pages and such will, I hope, be quickly usable by readers, creating some of the navigation connectivity is not intuitively obvious. Other than for typography (wink wink nudge nudge), it will be transparent, to the reader, that an entry in an RA page is actually a wikilink to a definition of a redirect to an article subheading.

Does anyone see a problem with an Editor familiar with these mechanisms to create an RA page, and, in some cases, to change text in articles so links (subheads, etc.) better fit? My concern, when we are already short of Editors that can Approve, that lower-case editing to make navigation work would exclude the upper-case Editor who did that editing from approving the eventual article.

In my thinking, if this is done with wisdom, it's simply another way for the wandering elder to point out history and background and introduce potential cooperation. The complexity of some of the actual mechanisms, however, is such that I don't think merely suggesting it on a talk page is going to help someone who hasn't become very familiar with the tools.

It may be, at some point, that we might have Toolsmiths, or at least Tool Manufacturers' Technical Representatives, that may not have subject matter expertise, but basic terminology understanding and great expertise in indexing and cross-referencing. In the paper world, such people are called Librarians, species Cataloger. Howard C. Berkowitz 17:19, 4 November 2008 (UTC)

Such people can be extremely valuable. Do we need to think about ways to encourage such contributions? What can we provide for them? Training or reference material? Recognition? Mentoring? How can we attract people with that expertise? Should we perhaps even invent a new role, perhaps Cataloger? There is overlap, but someone concentrating on linkage would have rather different priorities and responsibilities than an editor, and might need different powers. Sandy Harris 04:20, 5 November 2008 (UTC)

Using Related Articles pages

Related Articles play a critical role in the rise of the citizendium. As i see it they have a role for indexing/browsing. They also have a more organisational role for planning articles. This latter role is the reason I allowed these to exist without the metadata or the article.

When planning, it isn't strictly necessary to have more than an article title to put into the R-template. No other content is necessary.

The more one knows about the Template: R, the more that one can do on a Related Articles page, but one can use these pages as planning and organizing tools even with minimal R-template usage.

Related Articles pages are now largely dependent on the "R-template", which also appear on disambiguation pages. This template can be a place-holder, show an article name, a definition-only, or an article name and definition. One of the problems is that certain short topics need the function served by a definition, but it is not possible to wikilink usefully to a Definition out of a main Article. This is a hot area of discussion.

Interim test area

PowerPoint and I are facing one another in the dusty street, ready to reach...

Here's an ideal: The Other Place has lots of individual articles on aircraft and components, but there's no coherent way to get high-level relationships, evolution, etc. Through various structures, be they Related Pages, or something I'll call a "catalog" although it actually might be an RP, I would like to see us become a place where one can see patterns not visible there.

A "catalog", for example, might start as a list of the guess -- dozen or more just fighter combat radars now flying in U.S. aircraft. Which is on what plane? What is an descendant of another? Not just the U.S., but other countries have an annoying habit, due to internal politics and appropriations, of calling something an "enhancement" when it's essentially a redesign.

There are times, for reasons of external and external politics and perceptions, that a country goes through really strange naming conventions. Switching to naval warfare, there is a section in cruiser that I can only call the U.S. Navy's period of temporary insanity that ended about 1975. In this case, there was a "cruiser gap" that never actually existed or even had meaning.

Explaining these trends is, I believe, perfectly valid synthesis and indexing, but it just isn't done Over There. There is a limit to what I have time to do, but perhaps I can create the skeleton of a navigable structure on which others can build.

As I say, I'm working on pictures. It tells me something, for example, when the chart of radar designations, the aircraft using them, naming changes in aircraft and electronics, etc., begins to resemble a knitting diagram using spaghetti. Howard C. Berkowitz 20:34, 2 November 2008 (UTC)

CEC interfaces with a wide range of systems

Did you ever consider something more graphical with the appropriate image map links? To the right you see a half finished example using the CEC template (see Template:CEC). The grey boxes at the top left are active, as are all the KG antenna boxes. Every element of the picture could link to an appropriate articles or definition. Chris Day 20:44, 2 November 2008 (UTC)

A possible application or catalog generator

Go to AN/. Click on "what links here". If you don't see the pattern emerging, click on any of the XYZ-99 links to AN, and look at the way I do the bolded designation in the first sentence.

There are a lot of possibilities here, although, at some point, they really would need a data base. It would get too complex for manual maintenance (e.g., what are all the radio receiver types in U.S. military use?) Howard C. Berkowitz 22:13, 2 November 2008 (UTC)

Defined definition for redirect

AESA. Howard C. Berkowitz 22:54, 2 November 2008 (UTC)

Now you are getting the hang of it. So {{R|AESA}} gives:

• AESA [r]: Add brief definition or description

The one difference that alerts you to the fact it is a redirect is the black bold text as opposed to the blue plain text link of {{R|Radar}} that gives:

• Radar [r]: Acronym for "radio detection and ranging"; a system used to locate a distant object by transmission of radio waves and reception of their reflection. ^[e]

And points to the same page.Chris Day 23:21, 2 November 2008 (UTC)

Take this as either a usability argument against the italicizing, or an argument for some sort of parameter to the R-template, in definitions, etc., that says "don't call attention to this because it is doing exactly what I intended it to do."

You will also note that I broke the rule and had two sentences in a definition. Mea culpa, mea culpa, mea maxima culpa. Wink wink nudge nudge, know what I mean? Monty Python redirects...what a concept!Howard C. Berkowitz 23:29, 2 November 2008 (UTC)

I didn't know there was such a rule? Better two than run on sentences. I knew you would come back with the its doing what its supposed to do argument :) Originally it was just for redirects, the idea was to nudge authors to write the article but I guess if there is no intention of writing these articles then we should have plain text. I'll think more on this. Chris Day 23:47, 2 November 2008 (UTC)

How difficult would it to be have some parameter in the definition that communicates "I'm a definition only" to the R-template, so only in that case do you not nag? Howard C. Berkowitz 00:35, 3 November 2008 (UTC)

Some diagrammatic thought

Think of this more in the kind of relationships it shows. Now, the emphasis is on the radars, although one can read evolution of radar into aircraft versions within the model, version and block numbers.

some radar and fighter concepts

This is, however, a concept not at the other place.

• See all the radars in one place! (although not significant upgrades within a radar designation, other than the APG-63 ate the APG-70 in Version 4.
• See the associations of radars with aircraft!
• Not explicitly, but show how radars incrementally upgraded within aircraft series. This doesn't show when something was released in a new build aircraft, or was a retrofit.

I needed to look at this. Maybe it will trigger ideas about the navigational edge (never mind that some of these radars do navigation) that we could have over The Other Place. Howard C. Berkowitz 00:35, 3 November 2008 (UTC)

Science Fiction generally, Doctor Who specifically

To try some of these ideas on a completely different topic, where I am familiar with every detail, I'd like to pick up on some discussion that started in Talk: Doctor Who. Here are some bullet lists, admittedly with some awkward hypothetical article titles, and what might be on Related Articles pages. It is my hope that this will further validate a useful idea and offer more to differentiate us from the Other Place.

• Multigenerational science fiction (i.e., not just multiple seasons of one production)
  • Parent topics
    • Multigenerational science fiction series
  • Subtopics
    • Doctor Who
    • Star Trek
    • Star Wars

• Multigenerational British television comedy
  • Parent topics
    • British television
    • Television comedy
  • Subtopics
    • Monty Python
    • Grace Brothers
  • Related topics
    • Continuity in film and television series
    • Multigenerational science fiction

• Monty Python
  • Parent topics
    • Life
    • Comedy
      • British comedy
  • Subtopics
    • Television
    • Films
      • Monty Python and the Holy Grail
      • Monty Python and the Life of Brian
      • Monty Python and the Meaning of Life
    • Shows
    • Fan culture
  • Related articles
    • That was the week that was
    • Dilbert
    • Terry Gilliam
    • Eric Palin

Howard C. Berkowitz 14:48, 3 November 2008 (UTC)

What Links Here and Related Articles Pages

I understand that the "what links to the main page" is apt to be more chaotic, but, I'm not clear about the prerequisites for effective use of "what links to the related article page". It would seem that the latter is potentially useful, but requires that every article that links to the page must itself have a Related Articles page. These would be articles hierarchically above, below, and in a peer relationship with the article.

Now that I think about it, more than articles could link: if a definition-only can exist, but so can Related Articles for a subject with no article, those definition-only can fit into this navigation network. Such definition only items would, at present, get bold italics.

There may be nothing wrong with this assumption. With one exception, only those articles for which the authors or editors specifically wanted conceptual connectedness will be connected abd listed. For it to reach critical mass, however, without the existence of more tools, it requires contributors to be sure those pages exist and are maintained. Looking at "what links to the main page", while creating Related Articles, may be a reminder of what should could reasonably be in Related Articles.

A minor and slightly annoying bit of information displayed by "what links to the related articles page" will include the "main page", "approval", and possibly other pages internal to the cluster.

First, definition only page will not be a useful part of the network since we are discouraging excessive linking in them. Second, from a usability standpoint the links from the Related Articles pages are far more useful. Too many connections leads to the needle in a haystack problem; just not navigable. A smaller more relevent population is always more preferable, IMO. Knowing ALL connections is fine for context but bad for navigation from a human perspective. Chris Day 18:05, 4 November 2008 (UTC)

Implications for valued articles

If this, or conceptually similar, navigation tools, become a norm on CZ, one of the criteria for judging "good" articles will be that they are well linked. A tool, indeed, might be something that walks the navigation tree and counts the articles that are directly connected and some number of intermediate hops away.

Thinking of a current situation where I've been frustrated by a large number of essentially disconnected articles from an author, one of my arguments is that if there is similarity, there should be connectedness before many more articles are created in the same area. Also, in that same situation, I've made some suggestions that some contact didn't belong in a new article at all, but could have added to something existing. Making a practice of this kind of linking, I suspect, will occasionally reveal that editing the content into an existing and related article may be, in the long run, more user-friendly than many small and poorly connected articles. Small articles, which have many links to them, are very valuable because they are likely to be definitions, common background material, etc.

It may well be that the initial deployment of this navigation network will result in a certain amount of article consolidation, with deletion of articles unlikely ever to be much more than a poorly connected stub. Highly connected stubs are a different matter.

Anyway, these are some consequences to consider. Howard C. Berkowitz 14:52, 4 November 2008 (UTC)

Implications for naming conventions

In the Doctor Who structure discussion, User: Peter Jackson pointed out some words in CZ naming conventions:

"The title should describe the contents of an article accurately. If you have written an article about a topic, only to discover that your article concerns only one aspect of the topic (such as its history), then it is preferable to place your article on a more accurately-named page, to write a short article about the original topic, and then link from that short article to the new page."

This reminds me that good naming will be even more important. This example gives the idea of a name for a detailed stub relating to a specific hierarchically higher topic. Even for this, we need to be thinking about the mechanics of naming. Should it be major, minor? major (minor)? I don't think the latter is a good idea, since that tends to be the syntax for disambiguated titles.

How many levels of relationship should be in a title? major, history, dates of coverage? More levels?

There may well be some articles that have the principal purpose of providing a common linkage point, such as AN/. Going back to the "Midway, Battle of" versus "Battle of Midway" argument raised by Richard Jensen, it would seem that the "Battle of" should itself be a valid hierarchical node, or, otherwise, it is no more than a something already handled by the abc= field in Article/Metadata. It may well be useful to have a "Battle" page (not necessarily "battle of"); Battle can take some definition. It may also be that "Second World War, Battle of Midway" would be more meaningful. A one-level level subarticle name would then be "Battle of Midway, First attacks by U.S. carrier aircraft" (long, but it shows the point).

There actually is non-stub content for "First attacks by U.S. carrier aircraft, sacrifice of the torpedo squadrons". Creating that one-level/two-part structure, however, creates an article title that shows no obvious relationships, although its Related Articles page (perhaps with additional levels of parentage) would show those relationships. Even though the sub-sub-topic here might seem obscure, it actually has relationships to whole other hierarchies on sacrificial (with some chance of survival) and, in compare-and-contrast, suicidal attacks. Howard C. Berkowitz 15:07, 4 November 2008 (UTC)

Catalogs vs. Related Articles Pages

Do we need both? I think yes, but perhaps the catalog might be more of a pure index, possibly in tabular form, but I am not convinced.

Nevertheless, consider a subject such as Prime Minister of the United Kingdom. Is it not logical that every entry in Prime Minister of the United Kingdom/Catalogs should also be a Subtopic on Prime Minister of the United Kingdom/Related Articles? If so, what is the best way to implement this? Can the entries in a Catalog be substituted into R-templates? Howard C. Berkowitz 17:59, 4 November 2008 (UTC)

Catalogs should allow easy access to additional information about each term not just define it. This might be in the form of data or even trivia. This information will allow better comparison between the different terms in a particular catalog. I see it this way. The related articles subpage is a launching pad. The catalog subpage is for enrichment. Chris Day 18:09, 4 November 2008 (UTC)

Good distinction, although RAs do enrich with definitions and context. Nevertheless, existing catalogs appear principally to be simple lists. Perhaps they need to start being imported into RA, and kept as separate catalogs only when they add value that R-templates cannot.

Look at AN/. Is the nomenclature table more appropriate in a catalog subpage? This is probably a bad example, as the table is the only real content of the article; its real importance is as an identifier for RA.

I'm thinking that catalogs, in this context, are apt to be subject-specific tables. For example, a Prime Minister table might have names, dates of service, party affiliation, predecessor(s) and successor(s). Howard C. Berkowitz 18:17, 4 November 2008 (UTC)

Another subheading

See ALR-93/Related Articles for what might be a useful additional subheading, which I've tentatively (and somewhat ineleganty) called "co-topics".

This subhead would cover articles that are related "peer" topics, with independent roles of their own. In other words, they are not "subtopics" in the sense of being dependent on the subject of the article, but are "interdependent" or "codependent". There is a stronger relationship than "related articles" suggests to me. In this example, the related articles are about missiles against which the subject defends, but the subject can do so only as part of an interaction with the subjects of other articles. Those "interdependent" articles, however, will have independent relationships with other articles. Howard C. Berkowitz 18:46, 8 November 2008 (UTC)

R template trickery

Speaking of redirects, did you notice my recent suggestion on Talk:ULTRA? i think that is what you are trying to achieve. The working example is {{r|Supernova}}; it points to a subsection of Nova (astronomy) but uses the Supernova/Definition. The key is the redirect set up at Supernova. Chris Day 23:48, 1 November 2008 (UTC)

I took a look, and decided I wanted dinner before working out the assorted levels of indirection and redirection; I am probably going to diagram it and meditate over the resulting mandala.

Now you've got me trying to remember who said "No problem in computer science is intractable to a sufficient use of recursion." Dave Parnas? The next philosophical question is if that quote is semantically equivalent to an instructor, at the Special Warfare Center, who told me that no problem was insoluble, given a skilled application of a sufficient quantity of high explosives. Howard C. Berkowitz 23:59, 1 November 2008 (UTC)

CC Image
Flow diagram describing how to use the Definition Only and Related Articles Only type subpages in the absence of a specific article.

FYI, this is how I see it: Chris Day 00:29, 2 November 2008 (UTC)

(light dawning) The existence of the supernova definition is independent of whether or not supernova were a subhead (superhead?) of nova. Howard C. Berkowitz 00:48, 2 November 2008 (UTC) (continued below)

• What software did you use to create that diagram? (Wonder if it's freeware that runs on a platform available to me. Everything I now use is too cumbersome to get quickly into a page). Howard C. Berkowitz 13:03, 2 November 2008 (UTC)
  • I used photoshop; a bit like using a sledgehammer to squeeze an orange. I'm not sure what freeware is available. Chris Day 15:29, 2 November 2008 (UTC)

That definitely helps Related Pages. The only thing to which I could wikilink, though, if it existed, would be nova#supernova. Howard C. Berkowitz 00:48, 2 November 2008 (UTC) (continued below)

I don't understand. Why can't you link to supernova? The redirect gets you to the right place now, and it still will later even if we eventually have a separate article for Supernova. I use such links routinely, consider them standard procedure.

Triple DES, CAST-128 and Feistel cipher for example, currently point into block cipher. That eliminates red links in other articles, and if someone wants to write a real article on one of those topics later, they can trash the redirect and/or mine the existing text for anything they can use. Sandy Harris 14:53, 2 November 2008 (UTC)

Sandy, this is specifically relating to the use of the {{R}} template on the Related Articles Subpages. I agree that such redirects are common in the context of the articles. The significance here is that the R template can still produce a term specific definition even though there is no actual article (in this case Supernova). This is because the supernova subpage Supernova/Definition can exist and be ultilised by the R template despite there being no supernova article or metadata. Currently we are calling these Definition Only subpages. They are used exclusively on the Related Articles subpages and on disambiguation pages. Chris Day 15:29, 2 November 2008 (UTC)

Let me try an example, which may not be the greatest. The APG-63 radar is one of those "spiral evolution" developments, where it's constantly been upgraded in F-15 Eagle aircraft, which are pure air superiority fighters; the current version is AN/APG-63 V(3). F-15E Strike Eagle aircraft is an extremely successful variant that offended the "Fighter Mafia", who used to chant "not a pound for air to ground". Anyway, the APG-70 radar was a new design for the Strike Eagle, which adds air-to-ground to air-to-air modes. Some sanity swept in, and, when there started to be a need to upgrade the radar antenna/transmitter/receiver on the Strike Eagle, it was pointed out that the V3 was far superior to the "front-end" on the APG-70. So, the V(4) is going to add the air-to-ground radar computation function from the -70 and put it behind a V(3) front end. In principle, one radar for all the aircraft.

I haven't written up the air-to-ground modes, but let's say I decide to keep it in the APG-63 article. I need to be able to link from the F-15E article to a lower-case -d definition of the APG-70 as "the first generation radar for the F-15E". In an encyclopedia, that is all, I believe, that's necessary. But, if I need to put APG-70 in related articles to show history, I need to have both a Definition so the R-template doesn't nag, and a definition in article form so the APG-63 article text can reference it. D/definition really can be the same two sentences. Howard C. Berkowitz 00:48, 2 November 2008 (UTC)

I could be lost but it sounds like all you need is APG-70/Definition and APG-70 that should exist as a redirect to the F-15E article where it is mentioned i.e. #redirect [[F-15E_Strike_Eagle#Radar]]. Chris Day 01:01, 2 November 2008 (UTC)

Actually, seeing your current content at those pages, i'm sure I'm lost. I'll let you have a try at setting one up and then we'll go from there. Chris Day 01:03, 2 November 2008 (UTC)

I'm thinking about what I really want to accomplish with the radars, aircraft, radar technologies, and aircraft roles & missions, and sometimes I approach that better with a diagram. There are multiple objective, although my stream of consciousness was undoubtedly obscure. It did help me shake out some ideas. 13:03, 2 November 2008 (UTC)

I'll see how it materialises. Chris Day 15:17, 2 November 2008 (UTC)

Is this something you expected?

Should the R-template be blue-italic for Infrared light > Related Articles > Definition? It's not that there isn't a proper cluster definition, but the specific argument of the R-template goes through indirection.

Is this, in fact, the sort of thing you had in mind when displaying articles linked to the RA page? Howard C. Berkowitz 18:56, 4 November 2008 (UTC)

The R template is not designed for such a subsubpage. Note that the subpages template places it into Category:Unexpected subsubpage. There is also a Infrared light/Definition page which is the correct place for the defintion. Which Related Articles page is has the R template linking to Infrared light/Related Articles/Definition?

As for "displaying articles" I'm not sure I understand, could you clarify? Chris Day 19:06, 4 November 2008 (UTC)

Now I'm with you (although I still don't understand the "displaying articles" bit). Just saw your experiment on Night vision devices/Related Articles. In that case, I don't see why we need a definition for the Infrared light related articles subpage. Is it not self evident that the [ r ] link after the Infrared light term serves the same purpose? Chris Day 19:14, 4 November 2008 (UTC)

My question is that which appears at the top of the RA page, the confusing part being bolded: "See also pages that link to Infrared light or to 'this page". "This page", to me, means Infrared Light/Related Articles, rather than Infrared Light. Perhaps I don't understand the point to having that displayed. Howard C. Berkowitz 19:36, 4 November 2008 (UTC)

"This page" does refer to, in your example, Infrared Light/Related Articles. The point of that link is that it defines the Related Articles subpage network (the one that does not currently exist). For me this will be the more powerful link once all the related articles are complete. It is distinct from ALL the links to the article since many articles in that group could be almost completely irrelevant to the article. This is what i mean by the needle in a haystack, the noise to signal ratio will be too high for a "what links to the article" type link once the encyclopedia is fully developed. An example for infrared light might be Gulf War; it does link to infrared light, but would its related articles page? It might be useful for bots to harvest information about the enlarged ("complete") knowledge web but it's much less useful for human consumption. Chris Day 19:44, 4 November 2008 (UTC)

Ah. I don't disagree that the interconnected related pages show something systematic; it may not be a good example, but I have a vision of their being a set of high-speed highways, from which one jumps through interchanges and then to a local exit. Structuring the high-speed network will take some thoughts, and I'll start by posing some questions:

• B is a subtopic of A. Are there circumstances for which A should not be shown as a parent of B in A's RA page?
• C has A and B as parents. Along the lines of the multiple inheritance problem in object-oriented design/programming, are there circumstances under which it is reasonable for A and B not to list C as a subtopic? Situations do exist in software when that is true. Shall I live dangerously and try a genetic metaphor: C expresses in the presence of both?
• Are there any general guidelines for what should be in "other related articles"
• Is it reasonable to have a very broad subject listed as a subtopic? Consider WWII being listed as a subtopic of Winston Churchill. Obviously, WWII was much more than Churchill alone, but one cannot consider Churchill without considering WWII. He is not, however, only a child of WWII -- or can he? How many parents is it reasonable to have?Howard C. Berkowitz 19:58, 4 November 2008 (UTC)

All good questions. As a biologist I will tell you that exceptions to the rule are normal. That is why there are very few biological "Laws". Personally I try not to overthink the network. We will get what we get. For me there are even instances where it is not clear if A is a parent of B or if B should be a parent of A (Obama > Obama Presidential Campaign > Obama). In general, if we try and define the relationships within the RA subpages by too many rules we will lose the chance for pleasant surprises. But from a theoretical perspective I'm interested to see what rules might apply to this network. For example your synergistic relationship above is interesting (C expresses in the presence of both). Chris Day 20:08, 4 November 2008 (UTC)

Subgroups

Again, some discussion on how to use them will be important. There's also the social requirements of having critical mass for subgroups.

Incidentally, can we get enough people such that subgroups aren't a solitary vice? I pick the wrong disciplines. What, does biology breed editors?  :-) Howard C. Berkowitz 00:35, 3 November 2008 (UTC)

Biology is replication. And sex. Chris Day 03:47, 3 November 2008 (UTC)

And computer types make backups. Military either blows up things or keeps people from doing so. Now, I'm not totally ignorant about parthenogetic editor reproduction, but isn't it easier to start without a Y chromosome?

Hmmm...that gives a scary association. The U.S. aircraft prefix for "on test status" is "Y".

Now, I did hear of a mad biologist who did clone himself, but apparently had some transcription or expression errors when he forced the clone to full maturity. My informant did not tell me precisely how knowledge transfer had taken place, but, it, too, had errors.

While the professor was the most soft-spoken of men, the clone had an incredibly foul mouth, exactly duplicating his creator's voice. After the professor was humiliated when lectured for his terrible telephone call to the chancellor's wife, he and the clone had it out on the roof of the high-rise biology building. This gentle soul finally punched and kicked until the clone smashed ten stories to the ground.

He was immediately arrested by campus police, and asked how they could arrest him for murdering himself.

The chief explained that the charge was not murder. It was making an obscene clone fall. Howard C. Berkowitz 04:03, 3 November 2008 (UTC)

Article-specific subpages

This is an area that needs some usage guidance. There were some cases where Larry told me not to use them; I honestly don't remember the specifics, but I had the impression, at the time, that he did not want a proliferation of nonstandard subpage types.

In other cases, I started out with something in a subpage, and found increasing needs to link directly to it. For example, I had discussed fission and fusion as types of nuclear weapons, but found we really didn't have the weapons effect defined elsewhere. Linking to them didn't seem to work, although I may have had the wrong syntax. I could also see some software being confused on finding article-specific subpages from an external point. Howard C. Berkowitz 20:40, 1 November 2008 (UTC)

It is true that Larry is not enthusiastic of it getting out of control. But I think limited use might be fine. To read more see the following page and related discussion. CZ:Proposals/Should_we_allow_article_specific_subpages?, CZ:Article-specific_subpages Chris Day 21:26, 1 November 2008 (UTC)

Broader questions

We've got links and the various categories in "related pages". Outside here, there are links on the web as a whole, links within WP, co-author relations, e.g. Erdos numbers [2], and citations which link us to a larger world of research. What can we generate? How?

Take a citation index, for example. What can we deduce that might help an outsider get useful info on the field? A seminal paper gets cited often, over a long time period, and by workers in other fields. e.g. stuff Claude Shannon wrote in the 40s. A review of the literature has lots of citations. A paper that is never cited is likely to be uninteresting. If Bloggs cites "Bloggs, forthcoming" for a decade or so before he finally publishes it, perhaps it is his master work. Co-author relations give us some teacher/student relations. If a bunch of people are always citing each other, often in the same journals, perhaps they are a school of thought. Chomskian linguistics comes to mind. All the systemicists (another school) cite each other too. If you find people from one school citing the other, perhaps it is a debate. When a biologist does a paper that mostly cites mathematicians, or an economist is citing biologists, perhaps they are bringing new techniques into the field.

Most of those relations, or at least plausible approximations to them, could be pulled out of a citation database automatically. What could we do with our metadata? Sandy Harris 18:53, 3 November 2008 (UTC)

Readability

Most of the above page is Sanskrit to me. Therefore, before worrying about usability (what is it? I didn't get it from this page), we better start concentrating on readability. --Paul Wormer 13:55, 8 November 2008 (UTC)

True :) Howard set up this talk page so Sandy, in China, could participate in the discussion. Apparently our forums are blocked in China. Consequently above is pretty random. We need to break out the information and present it in a coherent way on the CZ:Usability main page. First we need some consensus, although i think we are getting close. Chris Day 14:22, 8 November 2008 (UTC)

Accurate description of the key reason, Chris.

There were other reasons why this is a quasi-sandbox rather than a proposal or structured discussion, and why it has a fairly vague name. I've always appreciated the quote, attributed to Werhner von Braun, that "research is what you do when you don't know what you are doing, but are aware of that fact and trying to correct it." This is not just a discussion; there are a fair number of experiments and discussion of the results in it. Even before one could do an experiment with template tricks, someone had to state a goal, and there might need to be several iterations of various ways to achieve the goal, with existing or new templates/software requests.

To me, the most important observation is that a distinguishing aspect of CZ, as opposed to WP, is that it has a basic principle of considering the organization of sets of articles comparable, in importance, as the content of individual articles. To find something in WP, more often than not, requires a search engine. We expect search engines to be used, but we want to give additional tools. Part of the "guidance" role of Editors and simply experienced Citizens is to keep encouraging those tools to be made: appropriate (as opposed to inappropriate linking, avoiding orphaned information, developing index-like constructs such as Related Articles while gaining consensus on their role and capabilities, etc.

Linking

Hyperlinks are the heart of any web-based collection of information greater than a single article. To borrow my favorite description of statistics, wikilinks are like a bikini: what they reveal is suggestive, but what they conceal is vital. Concealment, in this case, is more likely to be a lack of needed links, as in #orphans or to unifying articles of, in the Related Article sense, multiple generations of parents.

Inappropriate links

While having many red/magenta or even blue italic links in a page under active development is useful to the people actively writing it, if they haven't bee updated in some time, it may not appeal to a first-time reader. If we are lucky, that reader might start filling in those links, but, at some point, Editors need to ask the question: are there links here that are unlikely ever to be completed? Is there, perhaps, a need for either an explicit Talk Page section, or even a specific Subpage, about future directions?

For example, there have been discussions that unless day/month and year links are there for a specific reason, and there are appropriate target "what happened in September 1948" pages, the links just produce visual clutter. It may well be that simply having [[1948]] is not useful, but [[Birthdays#1948|1948]] or [[Second World War Timeline#June 1944|June 1944]] is a good reason to have a date.

A broader issue is the topic of insufficient linking, as described under #orphans. There are individual writing styles here. Often, a very knowledgeable author may "brain-dump" a subject until a reasonable structure of sub-articles starts to emerge. In other cases, it really isn't clear why an article is here at all, because it discusses a very narrow aspect of a larger topic, yet there are no articles setting the context of the topic.

Appropriate links

The WP-derived rule that it should be possible to trace an article, either through direct links or Related Articles, up to a workgroup core article, main page, or some other reasonable origin is a good one. The real test of notability is whether a given new article increases the interconnections among human knowledge.

Related Articles will have a major role here, but I think we are now reexamining Larry' original idea and seeing how to harness it. Things like definition-only pages of redirects to article sections, as strange as they may seem, to have a role. Techniques that obscure, however, aren't going to be obvious or friendly to many authors, and there may need to be either Editor (without interfering with Approval) roles, or even of a new kind of Citizen, the Cataloger, intervention to create them.

There may be times where a mouse-over graphic is valuable, although it may be complex to set up, and also has accessibility implications for text-only browsers for people with visual disabilities or very slow links.Howard C. Berkowitz 17:05, 8 November 2008 (UTC)