| 3 minute read |

** Rant warning **

Every week I try to find a few new interesting spatial datasets to use for Mango testing. Needless to say I spend more time than most doing battle with the various “Open Data” portals that the governments of this world—in their infinite wisdom—have gifted upon us.

They generally fall somewhere between rubbish and unusable in terms of the overall experience, and most don’t seem to understand what the word open in open data even means. They all stink of something that has been designed by committee and crippled by bureaucracy.

Is it really too much to ask to have a site where the data I’m looking for can be discovered within a few clicks and then just downloaded?

If you are creating an government open-data portal, here are some things you need to know:

We don’t care about your internal bureaucratic structure!

We don’t want to be sent on a wild goose chase following link after link and being bounced from one crappy government department website to the next. We want everything in one place, if I want to know who collected or maintains the data I will look in the meta data (if it has any).

If we can’t access the raw data, it’s not open!

Open data is free as in speech not free as in beer (Gratis versus Libre). If you are peddling your data in PDF, WMS or any other format which is a representation of the data and not the raw data itself then it is not open.

Truly open data should be available without barriers or obstructions!

If you place unnecessary barriers between users and the data then you are not embracing open data you are merely paying it lip service. If you email the download link to the users, that’s a barrier. If you make the user sign-up to access the download, that’s a barrier. If you only make the data available via an API, that’s a barrier. If you only make the data available in a proprietary format, that’s a barrier. If you mix your open data in with your paid offerings, that’s a barrier.

Open data is useless if no one can find it!

We don’t know your strange internal naming conventions, we don’t understand the bizarre internal tree structure used to categorise your data. If finding data in your portal is any more complicated than search → click → download, then you’re doing it wrong.

Here’s a great example of open data done right (hat tip to @berniejconnors), it’s the data portal for the Province of New Brunswick.

To get data you simply click on ‘Data Catalogue’, look at the sensible names and preview thumbnails to choose the data you’re interested in, and then download in your format of choice.

Job done.

Getting this stuff right isn’t rocket surgery.

** Rant over **