[ckan-discuss] Article about DGU

[Sean Hammond]

> Here is a really interesting article about data.gov.uk and the value
> of the data it contains:
> 
> http://www.bbc.co.uk/news/uk-politics-19065913
> 
> "The government estimates four out of five people who visit its
> data.gov.uk website do so without accessing any links to data."
> 
> This is of course caused by several factors, but I wonder what can be
> improved on the CKAN front (UI, UX, visualizations, etc.)

Hmm. I think CKAN has a lot of these kind of features already, but
probably a lot of work can be done to emphasise them more, let people
know that they exist, present them better, make them easier to use. So I
think the main thing that CKAN can do is probably to improve its user
experience.

Seems to me that some of the important things for someone interested in
data on something like data.gov.uk (rather than someone who is using
CKAN to publish data) are:

- Finding interesting data: good categorising, tagging, searching, clear
  UI presentation, being able to find related datasets and other
  interesting stuff once you've found one dataset that you're interested
  in, of course getting interesting and high quality data published
  first is a prerequisite

- Being able to link to datasets and resources, and specific revisions,
  visualisations, etc. thereof

- Being a able to keep track of data as it changes over time, see a
  changelog with diffs, get notifications from activity streams

- Some way to give feedback by commenting on data, flagging up issues,
  etc.

Of course CKAN's user interface and presentation are going to improve a
lot once the work on the new templates comes out.

I think it'd be really cool to do some proper UX testing of CKAN. I've
seen recently that Ubuntu report on lots of UX tests that they do of new
Ubuntu features. They get in a small number of users with varying levels
of experience with computers and with Ubuntu, give them tasks to
complete with Ubuntu, and watch how they do. This seems like a really
good thing to do and seems to have worked well for them. I've done this
sort of thing to test apps that I've developed in the past, it's amazing
the things you can learn that you never would have thought of, but that
seem the most basic and obvious thing in the world as soon as you see
them.

I don't think this necessarily needs to be expensive to do either, all
you need is a few willing volunteers, a room with some computers, and a
plan. I don't think you need to be a user experience or testing expert
to be able to pull this off and learn a lot from it. But it does take
time, especially to pursue the results that you get and follow through
by changing and then re-testing the UI.