Responsible data revolution

The agriculture sector is creating increasing amounts of data, from many different sources. From tractors equipped with GPS tracking, to open data released by government ministries, data is becoming ever more valuable, as agricultural business development and global food policy decisions are being made based upon data. But the sector is also home to severe resource inequality. The largest agricultural companies make billions of dollars per year, in comparison with subsistence farmers growing just enough to feed themselves, or smallholder farmers who grow enough to sell on a year-by-year basis.

When it comes to data and technology, these differences in resources translate to stark power imbalances in data access and use. The most well-resourced actors are able to delve into new technologies and make the most of those insights, whereas others are unable to take any such risks or divert any of their limited resources. Access to and use of data has radically changed the business models and behaviour of some of those well-resourced actors, but in contrast, those with fewer resources are receiving the same, limited access to information that they always have.

Therefore, it is important also to look at open data trends from a responsible data perspective.

One example of a new trend within the sector which brings these issues to the fore is that of precision agriculture, a farm management concept based on observing and measuring crops, environment variables and management operations with sensors and satellites. At the moment, most precision farming applications are employed in highly capital-intensive farming systems and most of the access to technologies and data remains in the hands of a few, large-scale farmers and service providers.

Responsible data challenges

Data breaches are not uncommon and they are growing in number; the loss of control associated with open data can lead to breaches that may have personal consequences for farmers. For example, SMS services used to reach populations with high mobile penetration bring with them a risk of a personal data breach. Especially with shared mobile phones, there is no way to know who is reading any given SMS message. As a result, it is hard to control who has access to information on a farmer’s financials, crops or land if they are sent by SMS.

There is also the need to anonymise or restrict access to sensitive data on human subjects. Sensitive data is more than simply personally identifiable information, and what is deemed sensitive is often contextually specific. For example, great care should be taken with data on community held land, resources and agriculture, especially when it comes to data on water resources and forest rights. Communities can be pushed from their lands or water resources in case official data is inaccurate to identify who depends on lands and other natural resources.

Accessibility and reliability is another challenge that needs to be addressed as more agricultural data is opened up. Large agricultural enterprises have skills and resources in-house to analyse and use data, thus it is very feasible for them to benefit from agricultural open data. However, small-holder farmers often do not have access to the Internet or the skills to use this data. Some might not even know that there is data that they could be using. In order to maximise use by the least powerful actors, open data needs to be communicated in a way that is useful for all actors. This means targeting offline and online communication channels to disseminate information building capacity for analysis and sharing insights rather than raw information.

Data ownership

With increasing amounts of data being created about farming and by farmers, one key issue is around ownership of data. When it comes to ownership of data created by farmers once it is aggregated with other farmers’ data – in many cases, this is then considered to be in the ownership of the company responsible. In the current set up, it seems clear that actors with access to more resources are more able to gather data and to understand the legal environment surrounding that data. Farmers are not legally equipped to ensure that they are benefitting from the data and legal mechanisms in place to ensure rights are weak.

This fear of sharing information sits in almost direct contrast to the push for more open data on the sector: in the case of indigenous peoples, knowledge that could then be used against them by malicious actors looking to profit from that knowledge is particularly sensitive information. Traditional seeds and recipes held by indigenous populations need ownership mechanisms to prevent knowledge from falling into commercial hands that may license them and require traditional communities to pay royalties. However, this type of ownership stands in contrast to open data, in which all data should be reusable by anyone for any purpose.

To avoid data to fall in the wrong hands, as part of CGIAR’s Open Data and Open Access Initiative, the Systems Office team supports Centers in helping them identify whether the data they are working with falls into exception categories (sensitive information, or information that can identify individuals when combined with other datasets) in order to put data management structures in place at the project planning stage. The Systems Office also works with Centers to ensure that informed consent is an important part of their project framework.

Shared decision-making

Given the huge power disparities present within actors in the agricultural sector, it comes as no surprise that this can increase tensions when it comes to the use, creation and analysis of data.

In order to find solutions for challenges and reduce future tensions, decisions on data should be made in a responsible way. Sharing the decision-making responsibility with people from the communities themselves seems to be the best way of ensuring no harm or negative unintended consequences. Co-design methods and collaboration early on in the data sharing process is also recommended as a way of getting solid buy-in from relevant communities.

Furthermore, there is the need to build capacity among smallholder farmers and less well resourced actors in the sector on how to deal with the growing amounts of data that are becoming available. Simply making data available is not enough to address these differences, and more needs to be done, potentially through providing low-cost advisory services on data use, or more accessible capacity building options. The responsibility for addressing this does not lie solely with the smaller players in the sector, though. Practising responsible data approaches should be a key concern and policy of the larger actors, from Ministries of Agriculture to companies gathering and dealing with large amounts of data on the sector. Developing policies to proactively identify and address these issues will be an important step to making sure data-driven insights can benefit everyone in the sector.

This article is a short version of the GODAN research publication “Responsible Data in Agriculture” (September 2016). You can download the full publication by using this url https://goo.gl/2KVeb6 or via the GODAN website (www.godan.info).