[Design] Big problem to solve: good WYSIWYG on WMF wikis

[Derk-Jan Hartman]

https://lists.wikimedia.org/pipermail/foundation-l/2010-December/063225.html

I just found back this post by David Gerard from 2010 and was struck by how
dead-on the discussion and analysis was and how far we have actually come
with VE 5 years later, even though we still did not pass the finish line
just yet.

Also interesting is some of the follow up to it, which points out that the
usability of Templates is also a real problem in itself, not easily
solvable with WYSIWYG, but probably just as important.

I think VE is really close now to being usable in production, but I think
that we are FAR from done on this front. Like was stated, templates are a
real problem. A UI problem, and one that VE doesn't really solve. Citoid
sort of does, but just for one small subset of templates.

I think it is important to remember that VE is a framework. The piece that
will open up other possibilities, but that we will need to still do a lot
of work to find what those possibilities are, how they can make page and
article authoring more usable etc...

The post starts with a quote of Fred Bauder: "There has to be a vision
though"

So I'm asking: What is the vision for this next step ?
- What ideas do people have with regard to usability and templates.
- What examples of good editors can we find that also deal with
templates/objects etc.
- What are our unique challenges ?
- What kind of research and development would be needed to deliver this ?

I would love it if we could end up with a discussion and summary as we had
back then. A guide for us towards solving this next problem within another
5 years. We and the WMF can probably not start on this tomorrow, but we can
start thinking about it.

DJ

[Amir E. Aharoni]

The first problem I always think about is figuring out which template to
insert - https://phabricator.wikimedia.org/T55590

Template parameter insertion works mostly fine if TemplateData is defined.
This requires some effort from the community, but it's reasonable. But the
editor always has to type the template name, and this means that the editor
needs to know all the available templates. Needless to say, there are
thousands of them in the active wikis. Some kind of a template picker would
be a Very Big Nice-to-Have.

A wider problem is that templates do A LOT of very different things, which
could be (very roughly) grouped into several types:
* talk page templates (different for user talk and article talk)
* article tags - {{unreferenced}}, {{Expert-subject}}, etc.
* inline tags - {{citation needed}}, {{who}}
* infoboxes
* inline content - IPA, unit conversion
* block content - {{Quote}}

Talk page templates should be largely irrelevant for VisualEditor
development, although they should inform Flow development.

Article-level tags should become page metadata and taken out of the page's
code. (Done well, this could be a boost to editor engagement by helping get
subject experts to take care of problematic content in a way that is better
organized than the current backlog categories... but I digress.) Something
similar should be done for inline tags like {{citation needed}}, although I
don't see a way to separate them from the page's code.

Infoboxes should be migrated as much as possible to Wikidata; ideally, they
shouldn't be in the page's code either.

This leaves us with inline content and block content. This is a mix between
visual formatting and semantic markup. Largely, the editing communities
have done a pretty good job at maintaining it the way they need it for
their projects, but unfortunately the implementation is different in every
language. A coordinated effort to find common templates that all projects
need and integrating their insertion in a smoother way would be great. This
was more or less with citations, and it can and should be done for other
areas.

As James said in the MediaWiki summit 2015, "global everything" :)


--
Amir Elisha Aharoni · א־מ֮י׹ אֱל֎ישׁ֞ע אַהֲךוֹנ֎י
http://aharoni.wordpress.com
‪“We're living in pieces,
I want to live in peace.” – T. Moore‬

Post by Derk-Jan Hartman
https://lists.wikimedia.org/pipermail/foundation-l/2010-December/063225.html
I just found back this post by David Gerard from 2010 and was struck by
how dead-on the discussion and analysis was and how far we have actually
come with VE 5 years later, even though we still did not pass the finish
line just yet.
Also interesting is some of the follow up to it, which points out that the
usability of Templates is also a real problem in itself, not easily
solvable with WYSIWYG, but probably just as important.
I think VE is really close now to being usable in production, but I think
that we are FAR from done on this front. Like was stated, templates are a
real problem. A UI problem, and one that VE doesn't really solve. Citoid
sort of does, but just for one small subset of templates.
I think it is important to remember that VE is a framework. The piece that
will open up other possibilities, but that we will need to still do a lot
of work to find what those possibilities are, how they can make page and
article authoring more usable etc...
The post starts with a quote of Fred Bauder: "There has to be a vision
though"
So I'm asking: What is the vision for this next step ?
- What ideas do people have with regard to usability and templates.
- What examples of good editors can we find that also deal with
templates/objects etc.
- What are our unique challenges ?
- What kind of research and development would be needed to deliver this ?
I would love it if we could end up with a discussion and summary as we had
back then. A guide for us towards solving this next problem within another
5 years. We and the WMF can probably not start on this tomorrow, but we can
start thinking about it.
DJ
_______________________________________________
Design mailing list
https://lists.wikimedia.org/mailman/listinfo/design

[Brian Wolff]

On the non-VE front, this sounds a lot like the use case of method
name completion in an IDE.

Maybe the usability of non-VE editing would be improved if we
basically made an IDE, with syntax highlighting (I guess people have
already done that with WikiEd), and auto-completion, where the moment
you type "{{a", there's a tool tip immediately above suggesting the
most common template names starting with the letter a, and getting
more specific as you type, and similar for parameters.

Of course when we start looking at programming tools for usability
suggestions, this strongly suggests our tools are not open to the
"average" individual. :s

On the more general front. Often I see people bashing templates for
how convoluted they are, and how horrible they are. Which fair enough,
they do scare people. But I think its important to remember how great
they are too, and that there is a reason they are popular. Instead of
designing systems meant to be take they crapiness out of templates, we
should have the mental model of designing systems that have all the
good things that made templates so popular, but also happens to not
suck. On the other hand, that type philosophic sentiment doesn't
actually solve the problem, and I don't have any great ideas about it.

--
bawolff

Post by Amir E. Aharoni
The first problem I always think about is figuring out which template to
insert - https://phabricator.wikimedia.org/T55590
Template parameter insertion works mostly fine if TemplateData is defined.
This requires some effort from the community, but it's reasonable. But the
editor always has to type the template name, and this means that the editor
needs to know all the available templates. Needless to say, there are
thousands of them in the active wikis. Some kind of a template picker would
be a Very Big Nice-to-Have.
A wider problem is that templates do A LOT of very different things, which
* talk page templates (different for user talk and article talk)
* article tags - {{unreferenced}}, {{Expert-subject}}, etc.
* inline tags - {{citation needed}}, {{who}}
* infoboxes
* inline content - IPA, unit conversion
* block content - {{Quote}}
Talk page templates should be largely irrelevant for VisualEditor
development, although they should inform Flow development.
Article-level tags should become page metadata and taken out of the page's
code. (Done well, this could be a boost to editor engagement by helping get
subject experts to take care of problematic content in a way that is better
organized than the current backlog categories... but I digress.) Something
similar should be done for inline tags like {{citation needed}}, although I
don't see a way to separate them from the page's code.
Infoboxes should be migrated as much as possible to Wikidata; ideally, they
shouldn't be in the page's code either.
This leaves us with inline content and block content. This is a mix between
visual formatting and semantic markup. Largely, the editing communities
have done a pretty good job at maintaining it the way they need it for
their projects, but unfortunately the implementation is different in every
language. A coordinated effort to find common templates that all projects
need and integrating their insertion in a smoother way would be great. This
was more or less with citations, and it can and should be done for other
areas.
As James said in the MediaWiki summit 2015, "global everything" :)
--
Amir Elisha Aharoni · אָמִיר אֱלִישָׁע אַהֲרוֹנִי
http://aharoni.wordpress.com
‪“We're living in pieces,
I want to live in peace.” – T. Moore‬

Post by Derk-Jan Hartman
https://lists.wikimedia.org/pipermail/foundation-l/2010-December/063225.html
I just found back this post by David Gerard from 2010 and was struck by
how dead-on the discussion and analysis was and how far we have actually
come with VE 5 years later, even though we still did not pass the finish
line just yet.
Also interesting is some of the follow up to it, which points out that the
usability of Templates is also a real problem in itself, not easily
solvable with WYSIWYG, but probably just as important.
I think VE is really close now to being usable in production, but I think
that we are FAR from done on this front. Like was stated, templates are a
real problem. A UI problem, and one that VE doesn't really solve. Citoid
sort of does, but just for one small subset of templates.
I think it is important to remember that VE is a framework. The piece that
will open up other possibilities, but that we will need to still do a lot
of work to find what those possibilities are, how they can make page and
article authoring more usable etc...
The post starts with a quote of Fred Bauder: "There has to be a vision
though"
So I'm asking: What is the vision for this next step ?
- What ideas do people have with regard to usability and templates.
- What examples of good editors can we find that also deal with
templates/objects etc.
- What are our unique challenges ?
- What kind of research and development would be needed to deliver this ?
I would love it if we could end up with a discussion and summary as we had
back then. A guide for us towards solving this next problem within another
5 years. We and the WMF can probably not start on this tomorrow, but we can
start thinking about it.
DJ
_______________________________________________
Design mailing list
https://lists.wikimedia.org/mailman/listinfo/design

[Jon Robson]

On Tue, Jun 16, 2015 at 2:53 AM, Derk-Jan Hartman <

Post by Derk-Jan Hartman
https://lists.wikimedia.org/pipermail/foundation-l/2010-December/063225.html
I just found back this post by David Gerard from 2010 and was struck by
how dead-on the discussion and analysis was and how far we have actually
come with VE 5 years later, even though we still did not pass the finish
line just yet.
Also interesting is some of the follow up to it, which points out that the
usability of Templates is also a real problem in itself, not easily
solvable with WYSIWYG, but probably just as important.
I think VE is really close now to being usable in production, but I think
that we are FAR from done on this front. Like was stated, templates are a
real problem. A UI problem, and one that VE doesn't really solve. Citoid
sort of does, but just for one small subset of templates.
I think it is important to remember that VE is a framework. The piece that
will open up other possibilities, but that we will need to still do a lot
of work to find what those possibilities are, how they can make page and
article authoring more usable etc...
The post starts with a quote of Fred Bauder: "There has to be a vision
though"
So I'm asking: What is the vision for this next step ?

Personally... I think it should be eradicate the need for templates.

If you've got a template that makes a graph, railway map, a gallery, an
infobox... what's really interesting is the data that powers and we can and
should provide better tools to store that data, and create presentation
layers on top of it.

I'd like to see "templates" evolve into something more akin to a namespace
that can be mashed up into all sorts of views e.g. [[Pie Chart:GraphData
1]], [[Line Graph:Graph Data 1]] - same data - different presentation.

I think the existence and reliance on templates is a testament to our
failure to keeping up with editors demands on the software. They should be
the exception not the norm.


- What ideas do people have with regard to usability and templates.

Post by Derk-Jan Hartman
- What examples of good editors can we find that also deal with
templates/objects etc.
- What are our unique challenges ?
- What kind of research and development would be needed to deliver this ?
I would love it if we could end up with a discussion and summary as we had
back then. A guide for us towards solving this next problem within another
5 years. We and the WMF can probably not start on this tomorrow, but we can
start thinking about it.
DJ
_______________________________________________
Design mailing list
https://lists.wikimedia.org/mailman/listinfo/design