Session: Data and Culture: Tuesday 3:40-4:50



Facilitator and Convener: Mars
Documenter: Juliet
Participants: Dennis, Ian, Ravi, Jane, Frank, Dave, others

Photo of mapped exploration.

How to make sure data is useful to a stakeholder:
What is a stakeholder? How do you know what the stakeholder is? It depends (on the data, who owns it) Who decides who the stakeholders are?
Data needs to be of appropriate scale
Data needs to be able to tell about itself - quality, completeness, provenance (where it came from, how it was manipulated)
Focus on the relationships and transactions (interactions between stakeholders that can generate a flow of information) between stakeholders

  • Transactional Metadata
  • “Transactions are Metadata about the world”
    Ask them* However, the customer is not always right
    Data needs to be actionable, context of what it means, not divorced from he meta data
  • Explaining what it isn’t so it is not extrapolated too far.
  • Organizable, searchable, and visualizable needs for same dataset differ per stakeholder group
    Usefully by being timely and appropriately shared or not.
    Stakeholders access to the data: connectivity to data, level of education required to interpret data, user flexibility in generating indices, end use

Communication between collaborations what and how farmer-researcher activist
Farmers, researchers, government, researchers are the people that need to share and communicate data.

Privacy: What and How
Could there be an ethics/code of conduct guide (on data sharing)?
Rubric - if it meets these criteria.
ISO/IEC 29100 Information System Privacy standards - how can these standards inform best practices, where does it fall short?
Community driven standards - A memorandum of understanding

  • - grew out of American farm bureau farm federation
    Privacy may vary among data elements
  • Each attribute in the dataset needs to have its own privacy setting
    Lop of digits of a GPS coordinate to get a general area rather than a pin point location.
    Spatial component can’t be anonymized. You have to utilize the spatial component behind a wall and give them back their result data. You’d have to encrypt, obfuscate the code.
    Anonymizaion needs to be well vetted and verified process. Otherwise it is a false sense of security.
    Fear of government or insurance companies telling you what to do - we need a trusted anonymization service.
    IRB in university research

Governance and Permissions of Databases
Can we apply GPL3 to Datasets/Databases? Is there a creative commons equivalent?
Research data alliance - legal interoperability principles and implementation guidelines
Do universities have to publish their data?
Association of Computing Machinery and IEEE - do they have services to offer to developers outside of universities

Opt-in Data Sharing Structure
Who is going to host it? Google? Amazon?
If you share, then you get the data for free, if you don’t then you have to pay.
A raspberry pie with 1TB of storage is enough to host some data…

We need a Open Ag Tech Code of Ethics and Conduct in efforts to do our best. However….
We need to keep in mind that security is impossible to ensure, no matter how hard we try
We need to remember everything we build will incomplete in some way
We need to remember that open ag tech doesn’t have the financial backing to adhere to hard to develop standards.