[Dev-flock] next phase

[Lucy Mendel]

Yo,

What is the next phase for the prototype? Navigation. Eventually 
there will be many behaviors, many theses (arguments attached to each 
behavior), and many discussion threads (arguments attached to arguments).

  Usability principle: Users shouldn't have to do the work to find what
  they're looking for. The old model is that as much information is listed
  on the screen as possible for the user to look through. The new model is
  that we do the filtering.

Right now behaviors are shown vertically with arguments inlined. I think 
we should show behaviors horizontally right below the node's information 
block. The behaviors would be in a film strip, like the comic strip slide 
view 
(http://www.thoughtandmemory.org/blog/wp-content/uploads/DataHunt/index.html). 
This filmstrip just adds dynamic fun. The main thing is that a limited 
number of behaviors are shown at a time *in a designated space*, while 
arguments are shown below in their own designated space.

Behaviors can be sorted by radio buttons that sort based on
   * nothing ("show all")
   * last modified ("most recent")
   * evaluations ("most popular")
   * ratings ("pro" "con")
   * tags--this is actually a *filter* operation. Selecting tags opens a 
div that shows check boxes labeled with tags. For our demo data set, the 
tags on a set of behaviors might be "Additives"  and "Hoax".

Eventually we may want to discover tags or sorts dynamically depending 
on the data set. We shouldn't show sorts or tags that return the empty set 
(or should at least grey them out). Etc.

The vertical column now contains only arguments, and only arguments for 
the selected behavior (but perhaps there should still be an 'all' button 
to see everything. At first it shows only theses, or arguments attached 
directly to the behavior. Eventually we'll want to sort these, too! 
Arguments can be opened into the normal looking discussion thread.

  Usability principle cont: The user shouldn't be able to do work that
  doesn't help them find what they're looking for (eg, a filter option that
  returns the empty set, or a filter option that returns the universe). We
  should do as much work as we can to show the user what they're looking
  for (eg, automatically showing the argument thread if a selected behavior only
  has one argument thesis-- we should show information when the set size is
  small, and intelligently hide information when the set size is large.

Ok, this is fairly general and intuitive, but hopefully headed in a good 
direction. I expect to be up and running tonight on my local machine or 
personal work space on the server and can start implementing (or designing 
the implementation for) some of this.

Mike, if you have any feedback or inspiration let me know. I won't be able 
to complete this alone, so this will have to be a team effort.

We'd like to 'go live' with the prototype Monday night, but it's not a 
killer deadline. We've reached the functionality necessary for user tests, 
and now we just need to steadily increase the functionality and usability.

Todo
  * users should be able to add nodes and add behaviors via the ui
  * users should be able to navigate bheaviors and arguments using 
different sorts and filters
  * are votes being tallied correctly? they should be displayed
  * user status, feathers should also be tallied. and displayed on a user 
centric page that also shows a user's impact.

Cool. Also, if you're interested in doing a work retreat on the cape some 
weekend soon let me know (if and when). The goal is to get together for 
efficient and fun work sessions (hopefully more productive than most 
weekends), and also to have fun in general in a friends beach house.