data as interface: flow

I believe that there are two kinds of ideas in the world: those that divide things into two types, and those that don’t… and then there’s a third, which tries to wriggle out of either. This is one of those.

The basic idea: a better interface to data would be to turn the data itself into the interface, as a flow between an overview and actual experience.

What is an ‘overview’?

  • visual language of overview: parallel
  • movie trailer, menu, signage
  • graphs, piles, sorts, maps, matrices
  • optimizes attention
  • optimizes action
  • browsing, analysis, editing, outlining, listing
  • powerpoint, excel

What is ‘experience’?

  • visual language of experience: serial
  • long, slow changes
  • reading content
  • still photos
  • based on important details
  • narrative
  • based on sustained attention
  • email & twitter

This is false duality of course; the actual value of either references the other, and both are necessary and interesting and have been around since the dawn of time. The reason this is an interesting thing to revisit is that there is a huge advantage in new media and social contexts: you can jump between them quickly and continuously, so that they start to merge into a ‘flow.’ This flow is an interface, perhaps a good one.
In basic terms, the flow interface resembles a lot of existing interfaces:

  • a list of emails > reading an email
  • table of contents > page
  • map > walking

But the differences that would be possible include:

  • use content itself in the overview, not a label/symbol/sign
  • browse and refine both the overview and the experience
  • interact with one through the other
  • interaction with an overview element shows an experience
  • interaction with an experience element references an overview

Now speed things up, so that the interface can offer:

  • “you can sort this stuff into piles three ways, which one works the best?”
  • “we have these three prototypes that show how the product could work”
  • “have a taste of 12 dishes before you pick your meal, then change your mind halfway through”

And of course, this can benefit from the language of data visualization, but we need a framework for overviews and a framework for experiences, not just one, and the ability to pick the framework itself should be content-based and usable through familiar conventions:

  • show proportional distances instead of vertices or measurements
  • make maps
  • show size relation, piles
  • show a tree graph
  • show nodes and edges

And here is where I sketch some actual designs for the very vague concepts I’m throwing around, but of course I haven’t got there yet. But I think it would be easy to show new interfaces for twitter, netflix, and digg that work this way.

Post a Comment

Your email is never shared. Required fields are marked *

*
*