Goodwill Is Not Enough to Make Open Data Work
Government open data portals are important, but they’re little more than facades for transparency unless they’re updated, maintained, and user-centred.
2011 was a big year for the Kenya open data community. The government, with help from the World Bank, launched a national open data portal dubbed the Kenya Open Data Initiative or KODI. We are the first country south of the Sahara to launch such an initiative.
At its inception, it had over 160 datasets sectioned under 6 categories: education, energy, health, population, poverty, and water and sanitation. Data was made available not only on a country level but also on a county and sub-county level. Kenya has 47 counties.
The vision was for a domino effect to be triggered. By providing data freely to the public on key issues including budget and expenditure, along with information on the provision of health services and educational facilities, developers could create apps and tools to assist citizenry in making better decisions and entrepreneurs could create new products and improve service delivery.
“It now falls to Kenya’s dynamic and entrepreneurial citizens to create user-friendly and relevant applications that will benefit Kenyans by identifying development solutions and improving development outcomes,” said former Kenya World Bank country director Johannes Zutt during the launch event.
Johannes Zutt, former Kenya World Bank country director. Source
Nearly a decade later, the site has 689 resources across 11 categories. However, most of them were last updated or uploaded in 2017, and I could only find 10 datasets from 2019.
Why is this a problem though? Isn’t it groundbreaking enough that Kenya has a portal where the public can quickly and easily access non-sensitive information about how the country works and runs?
Well, Kenya ranks at number 78 out of 94 countries with a score of 15% on the latest Open Data Index of 2017 run by the Open Knowledge Foundation, a position it shares with Saint Lucia and Namibia. This is quite a dismal ranking for a country that has a national open data portal. Kenya scored especially low because of a lack of proper data on administrative boundaries, air quality, national maps, weather forecast, company register, election results, locations, water quality, government spending and land ownership. That’s quite a number of very important areas.
With this in mind, what does it really take to run an open data portal? What are the most important factors to focus on?
In a nutshell, consistency and communication.
Consistency
Data never sleeps. Approximately 2.5 quintillion bytes of data are produced every day. That’s 25 followed by 29 zeros. It’s roughly equivalent to over half a billion HD movie downloads. And this number is only set to increase.
Maintaining an open data portal means keeping up with this demand. In this case, it means uploading government records as they are released.
The frequency of release of this data varies from indicator to indicator. For example, the Consumer Price Index in Kenya is released monthly but data that requires a longer time to research such as the infant mortality rate in each county may take a longer time to be updated. Regardless, there should be a system in place to keep track of these releases as they happen and make them available as soon as possible.
The Kenya open data portal provides a way to create your own visualizations such as graphs and charts. Additionally, technologists can download raw versions of the data through an API in order to interpret it on a deeper level and use it to build apps.
These are all helpful tools to assist in understanding the data, however, it means little when the resources are not readily available or regularly updated. All the bells and whistles are not as necessary as simply being consistent and providing the information as it comes out.
As a data journalist, my key concern is getting the latest statistics. If I’m writing a story on Kenya’s primary to secondary school transition rate, then I need the number of students completing their primary education and the number who have enrolled in a high school of the most recent year.
This is even more crucial in fact-checking. If a public official makes a claim of any statistical kind, how can I ensure and inform the public that they are correct?
I can create my own tables and graphs if necessary, but I cannot create my own statistics.
During the launch of the Kenya Open Data Initiative, Paul Kububo, the then CEO of the Kenya ICT Board said moving forward, he saw a Kenya with a more informed citizenry: “For the first time ever, people in our communities will be empowered to choose the best schools for their children, locate the nearest health facility that meets their needs, and use regional statistics to lobby their constituency representative for better infrastructure and services in their county.”
But how can this happen with two-year-old data?
Communication
At the time of writing, the portal’s blog leads to an error page.
This is worrying as having a means to update your audience on behind-the-scenes functionality, website updates and upcoming datasets is extremely important.
Kenya recently went through a census exercise in August 2019. This would have been a good opportunity to explain the history and purpose of the census and the different indicators that the government is looking at, especially any new ones that were added.
However, the last post on their Twitter page was in 2017.
Open data does not only pertain to the resources, but also to the methodology and process of collecting these resources.
I also requested data from them on 25 January 2020 using their requests page. At the time of publishing, I have not yet received any communique as to whether it is available or not.
At the end of the day, more consistent websites such as that of the Kenya National Bureau of Statistics is much more useful to data journalists in Kenya, because it constantly updates the data on its website.
The KODI is perhaps a cautionary tale for the open data community. Just a little over a year after the launch of the portal, it was reported that the project hit a snag because ministries and agencies of the same government that created the portal were refusing to release data to be uploaded to the public portal.
How could the tide have been turned?
The model used to set up an open data portal can determine how sustainable it is in the long run. Other than the government-funded model used by the KODI, there are three other options: securing funding for the open data portal; finding a way to generate revenue from the portal; or gathering that data using a few volunteers and input from the public.
In my next blog post, I will be talking about the latter option and how you can create a large, consistently updated database through crowdsourcing.