(anonymous guest) (logged out)

Copyright (C) by the contributors. Some rights reserved, license BY-SA.

Sponsored by the Wiki Symposium and the Nuveon GmbH.

 
This is version . It is not the current version, and thus it cannot be edited.
[Back to current version]   [Restore this version]

It's often practiced -- especially when extending existing systems -- to model the input data based on the way that the system actually works internally, instead of chosing the format optimal for the task at hand. This is unacceptable for final product.

Always design the interface (be it commands, switches, input text files or GUI) so that it is optimal for the work that the human user has to do. Never require strict or weird formatting, specific ordering of fields, duplicating information or similar things that make the user work harder just for the sake of simplier code.

Always make the system compute the values that can be derieved, present data in the form optimal for the user's task, etc. Processor power is cheap these days.

This doesn't mean that you should never require specific format or duplicated data -- they can have their own uses. For example, when changing your password, you're prompted to enter it twice. This is has its use outside of just simplyfying the system, so it's ok. Similar, if the input files are to be heavily edited by various people, it's good to keep them readable by requiring strict syntax. Of course, it should be paired with apropriate validating and error reporting.

This is a good practice: Use your computation power to ease the use of markup, even if that complicates your parser. Users will be greatful for a forgiving parser.

The most common way that good formats make the machine work harder is by allowing multiple administrator or user-group choices or plugin configurations to deal with the same simple markup. This should be done anywhere that a decision must be collective on the part of the users as a group without every single user being necessarily able to over-ride the decision. Three examples to illustrate this:

specifying emphasis without differentiating bold, italics, underlines, fonts#

When defining titles and alternate titles, quotes or editors notes, or headings, or links, we don't usually want to specify how. What matters is that everyone presents the same type of thing (external link, internal link, alternate title, quote, citation or reference) the same way for clarity. Fonts and text styles are supposed to be a subtle cue to meaning, not an excuse for users to show off how much markup they know (why we all hate PowerPoint and the PowerPointies who use it to lie to us). CSS goes a long way to make it easy to standardize these presentations so that people can easily read our text without our having to put a lot of work into learning typography, which is an expert profession of its own.

tags, lists, categories#

Bloggers use tags, wikis have categories, everyone has lists, and it's all the same thing. We are attaching keywords to items and asking for lists to be put together from them automatically. This is one way the machine really must work much harder, and where wiki has fallen behind drastically by assuming that every list must be manually maintained. It would be nice also to see lists that are assembled from internal information put into the same namespace as the manually maintained ones, perhaps with a prefix like "all" so we know they're up to date.

Also a few conventions can save manual work and redundant pages like letting [[tag:WHATEVER]] direct by default to [[category:WHATEVER]] and possibly include external resources with that tag (not just the local items in that category). It is hard to think of a case where you'd want a tag named WHATEVER to not put items into a category with the same name exactly, where that category already exists.

In practice, there are many variants of a tag all of which would redirect to a single category. If categories were smart about knowing which tags indicated things belonged in that category, a hierarchy of bottom-up tagging and top-down categorizing would quickly crystallize, with lists handling most side cases and non-hierarchical purposes.

A lot of http://let.sysops.be/wiki_best_practice suggests that categories should be used to directly support the users' core work process while tags and lists can proliferate madly and unreasonably for many purposes. Wikipedia is a good example, with very encyclopedic categories but many strange lists and quite a few tags that deal with editorial issues within the pages.

handling votes, ranks, surveys, polls #

Contentious decisions are inevitable in wiki governance and not all of them are unanimous. Modern definitions of consensus allow for the exclusion of a small number of persistent voices with very different priorities, or for those who disagree to be differentiated into a watchdog or indemnified group that is highly motivated (and maybe rewarded) to spot problems with the majority view. Some definitions go further and claim that "consensus" only means about 75 per cent of the voting members! In other social software it's become common to have surveys, polls, ranking, votes, and other formal quantitative ways to rank options. This could even apply to trivial matters like presenting people in a list or allocating some kind of points like BarnStars. In any case, the machine can offer a far far better polling booth than the ordinary and generic wiki page. Yes, wiki pages do barely work for this, but only if everyone cooperates.

So one way to make the machine work harder is to let users specify things to vote on, like pages to delete or users to elect as administrators or policy to adopt, in the simplest possible notation, preferably some variant of list notation. Characters for this can be reserved (as they often are in language/format work) without requiring the functionality to be implemented in the wiki software itself. If it isn't, then, the functionality defaults to an ordinary list and people deal with as they do now on wiki voting pages. If there is however some support for voting, then each notation can be assigned to one type of vote, which allows for a lot of methods, or a very few methods, to be applied. When we vote we are used to preferring one method and getting another (especially for say the President of the United States) so it's no tragedy if only one method is available. Being forced to agree on voting methods is part of democracy. In practice, it's not hard to list candidates and say "I prefer preference voting but if approval voting is available use that, if neither then first past the post, and if nothing then just make a normal list marked as a vote". This could be done by using list notation variants like #*# etc. When plugin developers see a market developing that wants real support to run those votes, they'll act quickly. Reserving those character combinations meanwhile costs $0.

See Talk.ListsReasoning for a proposal to deal with votes as a type of list.

Add new attachment

Only authorized users are allowed to upload new attachments.

« This particular version was published on 05-Feb-2007 19:19 by 142.177.76.130.