MeatballWiki | RecentChanges | Random Page | Indices | Categories


AccidentalLinking is a property of wikis (and probably other media, now 2006-12-07 e.g. DiiGo) that use the bumpy word CamelCase as a LinkPattern, that has been used by the original WikiWiki. It is common practice just to SmashWordsTogether when thinking about an idea in the hope that the page exists. Quite often it does. Indeed, this simple convention so dramatically reduces the name space, that the name -> concept mapping is much closer to being a one-to-one mapping. This increases the probability that concepts of different authors are linked, without knowing from each other beforehand.

When a wiki adds digits, punctuation or spaces to the LinkPattern or lets the linkpattern be non-bumpy, the number of different, still similar ways to name a concept grow dramatically, thus reducing the chance to "make both ends meet".

As a workaround we can try to reduce "8 track", "8-Track", "8-track", "Eight track", "eight track", "Eight Track" to the LinkPattern EightTrack, which the initial author of this page saw as the "only decent choice with the normal LinkPattern".

When a wiki uses free links like "I feel that arctic cod tastes really good" then the chance to connect different authors are reduced even more. Not only that: One and the same author probably cannot remember this FreeLink. When a wiki uses SubPages, the chance to connect different authors through accidental linking is reduced even more. (This is part of the reason why there are no subpages on Meatball: WhyNoSubpages).

CamelCase makes the page name look like a title. That way, each page feels like it represents a concept. Versus, say, an opinion. Much different tonality.



If you want to change the tone of your wiki to be something different, closer to MetaBaby (which could be interesting), then change your LinkPattern. Also, if the existing LinkPattern is wholly inadequate to discuss the topic at hand--say chemistry--then you have to change it--to allow molecular formulae a la H2O. -- SunirShah

FreeLinks may have a weakness in enabling more frivolous links, but these do nothing to address the glaring weaknesses of WikiWords. See Wiki:WikiWordsConsideredHarmful. WikiWords are but one way to do linking, not even the only way to perform AccidentalLinking, but are ultimately a hammer in a problem space that doesn't consist entirely of nails. --ChuckAdams

Some of these problems could be reduced by ignoring case and non-alphanumerics when comparing page names. Thus:

I think the difference between "8" and "eight" is probably worth preserving. Since we're throwing technology at the problem, we might as well drop "small" words like "a" and "the", and maybe strip off plurals too. -- DaveHarris

The CamelCase links are very easy to type... that was probably a good idea in the context of HTML, but it has encouraged people to be lazy and not think about that they are doing. Some wiki pages would be better off with fewer links. A combination of free-form links (so you can do what you want), plus a syntax which is harder work (so you have to think about what you want), might be a better balance, roughly for the reasons SunirShah says. -- DaveHarris

CamelCase links match so many things accidentally that people have resorted to the ultra-techie idiom of inserting 6 single quote marks in the middle of words to switch it off. I feel there must be a better way.

Most of the arguments on this page are fairly subtle, and will depend on the community. A system which permits long names doesn't require them, so a community could get good names whatever the link pattern. Contrariwise, even WardsWiki gets page names which are too long and over-specific sometimes.

 -- DaveHarris

CamelCase encourages smaller page names because, I suppose, we're used to creating Wiki:MeaningfulNames as programmers. Non-programmers may have harder times, however.

This reminds me that programming is one of the most disciplined arts of organizing information/content in meaningful ways. Poetry is probably more disciplined, but I think programming is second. -- SunirShah

Traditional publications have names like, "On Understanding Types, Data Abstraction, and Polymorphism". They are linked and cited explicitly. Ward's link pattern is not geared up for such titles. "OnUnderstandingTypes" would fit Ward's link pattern, but it is still in a form that needs to be cited explicitly. Compare "Link Pattern", which is a pair of words that occurs naturally in text, with "On Link Patterns", which would not naturally occur at all if not cited.

Um, OnUnderstandingTypesDataAbstractionAndPolymorphism?? (Surely longer titles will end up at BookShelved.)

The MeatballBibliography can suck up those references for easy copy&pasting onto pages where they are cited. It's not as if the copyright details of a book change ever. Often you don't actually need to create a page for a book unless you plan to cite it repeatedly or you want to give it an extra sales boost because you like it.

The "Accidental" Aspect

<Sunir> I think accidental linking is rare and random.
<kensanata> kind of like... accidental

I claim the accidental aspect of this is useless in practice, because we are not trying to write a dictionary, and we are trying to avoid ShallowPages. So accidental linking is only interesting for strong concepts, things that sound important. That usually does not happen accidentally. Either we want that page, or we don't. If you want a counterexample, consider that TwinPages do not occur very often either.

I claim that wiki masters actually check whether pages matching certain terms exist beforehand, and bend their text in order to be able to link to these pages. Thus, links are placed deliberately. As such, the LinkPattern could be anything.

The important part about the LinkPattern is that the words smashed together somehow form both a visual and a conceptual token. More about this on FreeLink.

But on EnglishWikipedia, accidental linking, facilitated by massive use of PageRedirects and disambiguation pages, makes the whole thing work. With 200,000 pages, even a WikiMaster can only remember a tiny fraction of them. You could achieve something similar with NearLinks. Further, when we say "accidental" linking, we're often really thinking about "half-remembered linking" - there was a good page on this topic, but I can't remember if it was SecurityThroughObscurity or SecurityByObscurity. Techniques that promote purely accidental linking also, more usefully, promote linking by humans with fallible memories.

The natural aspect

What if we were to design a wiki engine that would create links in the rendered HTML without any need of markup or word smashing? What if every page title in the database included spaces between the words (the URL could use '_' to avoid the %20 ugliness) where that space is natural? What if, as a person edited the wiki page, he didn't need to know which page names were already created, but if he typed something like: Free Software Foundation, and that page existed (by case insensitive comparison), then the rendered HTML would contain an href even though the author didn't use CamelCase, []'s, or '-' or '_' between words?

Of course the 8-track problem exists, but it also exists for CamelCase. The new engine might have an editable list of mappings - where each target page title would have an associated list of comma separated synonyms or regular expressions. For instance, if we decide to make 8-Track the actual page name, the list would look something like (like NTFS, this is case preserving but case insensitive):

8-Track: 8 track, eight-track, eight track

This is the idea behind PlainLink. --anon.

I like this. A mash-up of the GaGaParser and EditableTitle. It's worth experimenting with. -- SunirShah

Global Accidental Linking

Using long camel case words (in wikis and beyond) increases the probability to connect different camel-case-sensible communities via Google search, thus stimulating creativity. Check it yourself on GoogleTagWiki. -- FridemarPache


For inviting collaboration, I bilink this suggestion (made in the forum [CMapForum: WikifyingCmaps]) with <-> Wiki:WikiPrinciples . Besides that and since this forum is not (re-)editable, I deposit a copy on <->[AboutusOrg:WikifyingCmaps].

I hope that we soon get the critical mass of collaborators in the wider communities that the social software CMaps really deserves - for the benefit of all involved participants and communities.

-- FridemarPache

See also FreeLink, CamelCase, GaGaParser

DiiGo: wiki annotation accidental link linking linkpattern CamelCase FreeLink GagaParser? [1] CategoryWikiTechnology Google:wiki%20annotation%20accidental%20link%20linking%20linkpattern%20CamelCase%20FreeLink%20GagaParser

Warning: TagCombinations of DiiGo got lost. There was a time, when they worked.

On del.iceo.us, an associated bookmarking community, who got automatically the author's bookmarks by the simultaneous bookmarking feature of DiiGo: there is a [remnant]


MeatballWiki | RecentChanges | Random Page | Indices | Categories
Edit text of this page | View other revisions