[ckan-dev] Socrata

[Seb Bacon]

Hi,

Just been looking at "competitor" software to CKAN (though of course
we don't compete, we collaborate...!)

The number 1 product out there is probably Socrata.  It's not only a
catalogue, but a place where the data is stored.  It is a very
relevant comparison, and I highly recommend CKAN developers take some
time to explore it.  I wish I'd looked at it sooner.

My overall personal verdict is: we can learn a lot from their
presentation and a bit from their implementation.  What we are aiming
for is a more technical, data-savvy audience.  I think their data
tools are impressive on one level, but pitched wrongly on another
level -- you can do a little bit with them, but not enough to actually
do anything useful, and that's because they've tried to hide
complexity (I think).

Anyway, here's a tour:

Their open data catalog is at http://opendata.socrata.com/.  Click
through to a random data package from the home page.

The default view of a package is a tabular format (plus two other
layouts in increasing detail).  In fact, they don't have a concept of
a package, having what we call resources as their top-level entity.

There are then various things you can do with the data, which are
relevant for our Data Workbench type discussions.  The things you can
do are sort, roll-up, show/hide columns, reorder columns, make various
graphs, plot on a map, etc.  The graph features of course rely on the
correct data types, so a datetime column must be marked as such in the
system, and there's no way I can see for an editor to do that.

Without allowing you to improve the data or link it to other datasets,
this toolset is limited; however, I *really* like the way it
encourages the casual visitor immediately to start playing with the
data, and then gets you to save it as a "view".

Thus you can have this chart:
http://opendata.socrata.com/Government/Timeline-of-Governors-by-Political-Affiliation-197/hmak-vpvn

And you can see at the top the dataset from which it was derived.

I also like the way the "about" is actually made much less prominent
that the data.  Effectively their "about" tab is CKAN's entire package
screen.  I think they go too far for our own purposes, but I
personally feel CKAN is a bit too technical at the moment.

There are user profile pages where you can see all the things a user
has done, e.g. http://opendata.socrata.com/profile/Saf/eudm-snef

For backend users, they have the usual analytics, branding, and
authorization stuff.  The publishing part includes hosting the
dataset.  I presume that part of the dataset upload process tries to
help the uploader mark columns as date, location, etc.

They have a widget for embedding a data view in your own website.

There's a "collaboration" feature which presumably allows an
organisation to keep records inhouse and then publish a subset to a
federated satellite system -- harvesting, as far as I can tell.

Finally, they have a nice "developers" website with a fancy
explore-our-api-interactively tool: http://dev.socrata.com/

That's all for now!

Seb