At Govhack a small team of folks started working on the idea of visualising all the activities of the (Australian) federal government, so that as citizens we are better positioned to identify where we’d like to intervene in government processes. This post documents the progress we made.
Our work focussed on a single data file from this National Archives set – agencies.xml, which lists all the Commonwealth Agencies ever to have existed (around 7000, with around 2000 extant), along with 212 functions that they perform.
Rob Manson did some invaluable work in converting the data into more usable forms – taking out the sizeable – but sometimes empty – NOTES field, and producing various json (and plain text) files that included different combinations of data.
Here is my tagcloud of functions of the Australian government made using ManyEyes:
It’s an interesting array but its meaningfulness is questionable. Text is sized according to its occurrence in the data, which could mean that multiple agencies are responsible for that function, or that it has been passed around a lot from agency to agency, or simply that it has been around since Federation or before (eg, telecommunications, taxation). A smaller font could also indicate a relatively new function (eg, criminology), or one subject to terminological shifts (eg, foreign policy has become international relations). Then there are perplexing aspects of the data, such as that quarantine started in 1832 but ended in 1993 even though it remains a preferred level-two term in AGIFT – under primary industries. Hm, there are plenty of issues here.
Officially (ie, according to the Functions Thesaurus) there are only 25 top-level functions – the rest are either sub-functions or non-preferred terms. But clearly there are discrepancies between the official terminology and the actual descriptive data. Ultimately, for this project to succeed as a public map, we will need to also enable people to use other, unofficial, vernacular, folksonomic terminology as well. Has anyone ever tried tagging terms? Tag tagging – hm, there’s a concept.
Another aspect of this project is that people might identify functions that the government should be performing but is apparently not.
Rob also produced some great images, focusing on a single function and showing the complex array of agencies that perform that function. For example, here are some of the current agencies doing SCIENCE – click for the full graphic:
Finally, Brenda Moon made significant progress on her path to use Processing to visualise the agencies and functions in a spectacular array based on NYTimes 365/360 by Jer Thorp – again, click to download the full PDF:
These are all first steps, but we had fun exploring, and we’ve raised many questions and issues. As well as how to show – and enable interaction with – the dynamic relationships between agencies and their functions, incorporating date ranges, locations, and vernacular terminology, there is no doubt much else to consider.
I’m thinking, for example, about how functions themselves could be structured so that the array is more visually readable. Would it work to plot or colour them according to two axes – whether the object of the function is individual or collective, tangible or intangible? (Eg, some more individual-oriented functions include health, employment and immigration, where communications, civic infrastructure and environment are more collective. Environment and civic infrastructure are also more tangible, along with primary industries, in comparison to governance, business support and regulation, and justice administration, each of which has fairly intangible processes and products.
I’m also wondering whether the agencies.xml file is more problematic than useful as a source of functions data. Other sources of similar information might work better. AGLS is a contender, although it has not been consistently implemented across government. Maybe the AAOs that I used to make this other tagcloud are a more reliable source. At least there the frequency/size of the term corresponds to the amount of legislation passed to manage it. Or perhaps we should look at this issue another way. For example, in talking to Gordon Grace, from AGIMO, it might be possible to obtain some bulk data on what people are searching for at australia.gov.au. Possibly, whatever data forms the foundation of the array of government activities, the rest will need to be crowdsourced.
One more point to note: once the initial interface was populated and making sense, it would be good to direct users to several destinations:
- current online material about each activity
- contact details for the Minister currently responsible
- historical records
So, this project seems to be growing some legs ! and I’m looking forward to continuing the discussion with all interested parties.