WysiwygIsntLinking

I just want to briefly note that there’s a problem with the WysiwygWiki implementations I’ve been seeing --

The problem is that they make it really hard to do LinkLanguage. In fact, I’d argue that it’s not link language, because linking is not easy / transparent enough.

In these wiki, to hyperlink WikiWords, you have to:

1. select the text with the mouse,
2. click “link,”
3. specify that you want to link to a wiki page, rather than an absolute URL,
4. select the page that you want to link to (even if by the same name)

This is a source of daily frustration for me, on the PbWiki?’s visual mode. This is the mode that wiki activists are encouraging new-to-wiki people to use.

How can this be resolved?

WYSIWYG is too valuable to be tossed, I think.

Here are two possibilities I can see:

• PlainLink
• CamelCase

Are there any other ideas?

In this post, I have linked to 7 different pages. If I were using WYSIWYG, I would have had to, for each one, recognized that I was linking to a page, and then gone through the whole rig-a-marole. This is basically prohibative, in the vein of SmallChangesYieldGreatChanges.

So: “No LinkLanguage.”

Some possibly useful resources about problems with WYSIWYG:

• Word Processors: Stupid and Inefficient
• What you see is Wiki (pdf)

This reminds me of another problem. Or perhaps it’s another part of a bigger problem. But I am fuzzy on exactly what the bigger problem is, naming it, and I’m having difficulty verbalizing what little I do know.

I know a part of the problem. Few people understand what a partially finished project is supposed to look like.

They are familiar with finished projects (a movie, an automobile, a printed book, a college graduate). So when they see an unfinished project, they immediately know something is wrong, and they don’t hesitate to tell you it’s wrong. But often their perception on what specifically is wrong, and what the proper next step is, is mistaken.

To most people, a couple of pages typed up in a word processor and printed on a laser printer looks more like a finished book than a pile of index cards with hand-scrawled notes – but the reality is exactly the opposite.

To most people, a few seconds of ray-traced animation looks more like a finished movie than a bunch of actors in front of a blue screen reading what appears to be a poorly-drawn comic book – but the reality is exactly the opposite.

Thrashing around in the water feels like something has gone horribly wrong – but it’s a natural part of learning to swim.

Sometimes I know something is going to be a big project. I suppose I could decide to drop everything and work on it full time. Or I suppose I could decide the benefits are not worth the cost, and not do it at all. But there are other options: Some things can be converted into a RampTask so that I can push it forward with a couple of FiveMinuteImprovement each week, and eventually finish it, even if I have to do it alone and so it takes years. And I think that maybe if I post it online, I can attract others to help me – or at least help the next guy who tries it, help him avoid some of the dead-ends paths I went down. But I feel discouraged when people who see my partially finished project, and (correctly) see that it’s not done yet, tell me “It doesn’t look right. You should just erase everything and start over”. How exactly does that help the next guy? How exactly does an empty web page help me attract people who can help me accomplish this project?

Another part of the problem is that too often, people feel they want to do something, or someone asks them to do something, and they have no idea what tools are available – much less, which of the multitude of tools available are really the best tools. So if I tell someone to do something, even if I specifically tell them “I don’t care what tool you use to do it” – if I see some shiny new tool that looks like it would help, and give it too them – too often they incorrectly assume I want them to “do that thing with this particular tool, and don’t you dare use any other tool”. Or the first tool that comes to hand that looks like it would be useful – they try to use that one tool for the entire project, rather that using one tool for one part, and a different tool for another part. And so they refuse to use simple tools that obviously can’t do the entire project, and gravitate towards complicated tools that at least appear like they could be used to do the entire project.

Any one tool makes some things easier, but other things harder (WhatIsAffordance). I suspect some tools trick people into accomplishing less than what they could have accomplished with other tools. Peter Norvig and David Morgan-Mar seem to feel the same way.

Wow, I think you are right about these “unfinished products” and how it is hard to recognize them. It didn’t occur to me before, but it all fits and makes a lot more sense now.

As for the tool selection, I think that the ParadoxOfChoice? adds to the confusion: when people are not familiar with the task before them, or not sure about the future requirements, they will “play it safe” and select the most comlicated and feature-rich tool – just in case they might need any of the features it offers, even if they don’t know what they really are. Professionals, or people who feel confident and happy to experiment will on the other hand pick a number of simple tools – and will happily change them and try different combinations.

Another factor that may play a role here is that people generally don’t like responsibilities – at least when in the “I need to have it done and forget about it” mode. This means they will try and make as few choices as possible. This means, they will pick the universal, all-whistling all-dancing tool, and preferably one that has screenshots on its download pages – screenshots that show exactly what it does (this brings us to the fact that it’s hard to show what an application does and how it works on screenshots, when the application is not WYSIWYG).

Let me just add that this is why the Oddmuse + FCKeditor implementation supports traditional links – both WikiWords and FreeLinks using double square brackets are supported. I think that a long time ago relevant issues were discussed on AccidentalLinking.

Additional reference: ENG Wikisym2008: WYSIWYG & Wiki Editors – particularly slide 14 - on slides 16-17, it suggests autocomplete as a way of promoting linking within WYSIWYG. Combined with a LocalNames directory, this could be very cool. Contacts are: Nikolay Yaremko (makiwara, humanemagica.com,) and Konstantin Kolomeetz (kolomeetz, yandex-team.ru)