Working from a brief from a campaigning NGO at the Open Data Campaigning Camp in Oxford, 24th March 2011, a group explored how data could play a part in a global campaign about the impact of high food prices.
The focus was on working out how data could support campaigning on this issue.
The first thing the group did was to map out different potential influences on food prices, from commodity speculators, to government policies and oil price rises; and to look at how has influence on control of food prices - from governments and international organisations, through to pension funds and individual investors (through impacts on currency speculators).
The second thing the group did was sketch out the 'audiences' for the campaign - including southern partners and individuals; UK based citizens; and researchers, journalists and experts interested in the subject area. Each group would need a different sort of information product to help them understand the campaign - delivered across different platforms, from mobile to web to print.
We discussed the need to visualise the different parts of a causal chain between diverse actors and harmfully high food prices. (are you a good butterfly; or a bad butterfly in chaos theory impacts).
We used the sourcing data section of the open data cook book to identify different data sources that might be available to us. Using data from the Food Security Portal (developers section) and Google Fusion Tables we created a number of visualisations of commodity price change over time. These quick widgets (created in 30 minutes or so) help to show commodity price volatility (and, by turning oil on and off on the line-graph display), the impact of oil on food prices.
We discussed how these visualisations could be enhanced by showing how much the food price change would affect different people according to their income or where in the world they are (and thus how much of their diet a particular commodity may be predicted to provide). We noted that this was global food prices, not local - but we couldn't find good datasets that provided local food price indexes.
Using the idea of Food Price 'Hot Spots' we also explored what data we could link to food prices to show the impact of high food costs. As a proof of concept we took a list of riots from Wikipedia and picked out a number that could have a food connection. We first used Google Fusion Tables to geocode all the riots and to see if any geographical patterns could be seen, using the Geocoder feature to extract place names from descriptions of the riot. This gave a few results, but not many successfully coded - so we manually added country names (though we didn't use them in the end).
As a proof-of-concept to explore the 'storyline' visualisation in Google Fusion Tables, we added formatted the dates against riot stories in the same format as the dates used for commodity price reports. This allowed us to use the merge feature of Google Fusion tables to combine our datesets so that stories were attached to dates. The storyline visualisation can then be used to show how different events may be related to trends in food prices.
We discussed how linked data tools like dbpedia.org could help with crowdsourcing data and translation. For example, the dbpedia page for Rice gives translations and descriptions of rice in a range of languages in machine readable forms.
This could help allow crowd sourcing in multiple languages and support data integration.
We identified a range of approaches to using open data in campaigning:
NGOs may be able to source, prepare or provide as open datasets that were not already available. Providing these could support a wide range of visualisations and mash-ups. Creating a 'data' page for the campaign where datasets are listed, and notes on how well different datasets combine, would be useful to help others use data to support the campaign.
The missing 'data' might be a formula for mapping two datasets together: If researchers have worked out how to make global commodity prices into more useful data for understanding the impact of food prices on individuals (e.g. a formula for effectively combining Purchasing Power Parity, Commodity Prices and Average Income), clear documentation of the process and some datasets prepared for this would be really useful.
By creating your own visualisations and widgets that others can embed and work with you can give people content to share on their own blogs and websites which tells the story of the campaign. Linking statistics and stories (including photos, video and audio) can be powerful here - showing how distant actions impact people across the world.
Visualising chains of consequences was also seen as potentially useful here.
As Lisa Evans notes in her observations on data journalism, getting news coverage out of data often involves being able to respond fast. Compiling and standardising a library of data as part of preparing for a campaign means that:
-
A) Campaigners can support journalists in getting to stories faster;
-
B) Campaigners can support the key data-journalistic pattern of showing
“patterns where a poltical or buisness system is not working or is corrupt and then digging down to find individuals responsible” (
REF)
-
C) Campaigners can respond to news stories by creating their own infographics (quickly; within hours of the story gaining traction) to benefit from google traffic and general interest in the story.
We discussed the idea of allowing individuals to tell the story of their own food price hot spots, using freely available data. By providing the data used to tell this story, some pre-moderation of what can be a hot spot is provided; but not too much emphasis should be placed on moderation. Let people do lots of different things with the data: just feature and give backing and profile to a small selection that support the campaign.
One of the big possibilities of data is bringing together different datasets to look for relationships and trends. Encourage researchers and others to bring their own data along and mash it up with yours and to highlight potential trends.