Open data policies in food security: presentations from IASSIST 2017

IASSIST 2017, held at the University of Kansas in Lawrence from 23 to 26 May, placed “Data in the Middle: The Common Language of Research” at the centre of discussions about how research data travels between disciplines and institutions. Its archived program, schedule, plenary information and presentation materials preserve an early snapshot of debates on big data, deep learning, digital agriculture and global food security.

Food security depends on evidence from many settings: grain and livestock farms, ports and border agencies, wholesale markets, public health laboratories, food banks and household budgets. Open data policies determine whether those records can be discovered, compared and reused responsibly. Without common metadata, accessible licences and safeguards for sensitive information, valuable datasets remain isolated.

Australian readers can apply the archive’s lessons to a country with strong public statistics, a large export sector and pronounced regional differences. The questions raised in Lawrence resemble choices made by the Australian Bureau of Statistics, data.gov.au, state agencies and local councils. They also matter for producers from the Northern Territory to Tasmania and for buyers tracing food through Melbourne, Adelaide and Sydney. The archive is historical evidence rather than a current policy statement, so its ideas should be tested against present Australian standards.

Why the 2017 archive still matters

An archived conference site records the sequence of questions surrounding a presentation, not simply its final claims. IASSIST 2017 gathered researchers working with environmental, agricultural, social and computational data. Its central theme recognised that shared evidence depends on common concepts, methods and ways of describing information. Food security was a particularly clear example because a national food balance sheet may conceal shortages experienced by particular communities.

The archived materials on big data and digital agriculture show how rapidly the policy environment was changing. Researchers were exploring satellite imagery, machine learning, mobile information and large administrative datasets as tools for understanding production. Each source carried different rules about access, consent, confidentiality and ownership. A dataset described as “open” might still be unusable if its licence was unclear, its coordinates were exact or its fields lacked definitions.

The site also retains practical information about check-in, transportation and local recommendations from the Kansas event. That operational detail is a reminder that conferences create knowledge through formal sessions and through the practical systems that connect participants. For contemporary policy, the most useful legacy lies in the questions that accompany the data: Who collected it, who benefits, who bears the risk of disclosure, and can another researcher reproduce the result?

Connecting global research with Australian realities

Australia has several public data foundations that can support an open food-security information system. The Australian Bureau of Statistics produces agricultural, household and economic statistics, while data.gov.au provides a national route to government datasets. CSIRO and the Bureau of Meteorology contribute research, environmental and climate information. Their usefulness increases when agencies publish machine-readable formats, persistent identifiers, clear licensing statements and information about revisions.

Food markets in Australian cities offer a practical bridge between public data and everyday food access. Aggregate price and volume information from Melbourne’s Queen Victoria Market, Adelaide Central Market and Sydney Fish Market could support analysis of affordability and procurement without exposing individual traders. Customs and border records from the Australian Customs Service and Australian Border Force add another dimension. Open summaries of commodity volumes, origins and processing delays can reveal supply-chain risks, while shipment-level commercial information remains protected.

Regional realities make national averages less informative. Producers around Wagga Wagga and Moree may use satellite imagery, soil sensors and precision-agriculture tools, but machinery records, farm coordinates and contract details can be commercially sensitive. In remote northern Australia and places such as Alice Springs, food-access data also requires careful governance. Community-controlled information about transport, affordable food and health services may be more useful than a national dataset that ignores local knowledge, language or consent.

Policy choices that make data useful

An open policy should make data accessible while preserving legitimate controls over confidential material. The governing principle is often expressed as being as open as possible and as closed as necessary. Agencies need to distinguish between data that can be released directly, data that can be aggregated or spatialised, and records that should remain under restricted access. The Australian Privacy Act, the Office of the Australian Information Commissioner and relevant state regimes provide essential boundaries for that process.

Metadata is part of the policy, not an optional extra. A dataset without units, geographic coverage, update frequency and quality notes may create false confidence even when the underlying figures are accurate. Persistent identifiers and version histories allow researchers to cite the exact release they used. A comparison of common food-security data domains illustrates the choices involved.

Data domain	Open release	Controlled alternative	Food-security value
Farm production	Regional yields, crop areas and pest summaries	Farm coordinates, contracts and machinery records	Production forecasts and targeted extension support
Customs and imports	Commodity totals, origins and clearance-time aggregates	Named importers and shipment-level records	Biosecurity planning and supply-disruption detection
Food markets	Price indices and category-level sales volumes	Individual vendor transactions	Affordability analysis and institutional procurement
Climate and environment	Gridded weather, rainfall warnings and fire information	Precise locations of vulnerable households	Early warning and local adaptation planning
Household wellbeing	Aggregated nutrition and food-access indicators	Identifiable health, income and benefit records	Targeting support while protecting personal information

Openness also depends on interoperability. A crop dataset, rainfall record and market series become more useful when they use shared place identifiers, time periods and definitions. Policymakers should fund documentation and data engineering alongside publication. They should also require agencies to explain why a dataset was collected, whether it represents the intended population and how omissions or sampling bias affect its interpretation.

What the presentations reveal about data gaps

A linked presentation on the impact on smallholder farming illustrates the importance of context. Large datasets can reveal planting patterns, yield variation and responses to weather, but information produced elsewhere may not match local soils, crop varieties, farm sizes or market arrangements. Smallholder farmers also need a route to correct records, contribute local knowledge and retain control over commercially valuable information.

The archived deep-learning material raises a related issue: more data does not automatically produce more accurate or fair decisions. Models depend on labelled examples, and the people and landscapes represented in those examples determine where the systems work well. Agricultural image tools may perform differently across districts, seasons and farming practices. Open training-data documentation, evaluation results and model cards can expose these gaps, while confidential farmer information can sometimes be protected through aggregation or trusted research environments.

Presentations on global food security demonstrate the need to combine evidence across scales. International trade data, national production estimates, household surveys and community observations may all describe the same system while using different boundaries and dates. A common language must therefore include units, denominators and definitions of access. The IASSIST theme remains relevant because food-security questions cross organisational borders at the same time as technical barriers separate datasets from the people who could use them.

Turning the archive into Australian action

Australian agencies can translate the archive into a practical release programme. The programme should connect open-data policy with the decisions involved in collecting, funding, publishing and correcting food-security information.

For agencies and public data stewards:

• Publish machine-readable metadata with every release, including coverage, units, quality limits and update frequency.
• Apply enduring open licences while documenting exceptions for personal, commercial or culturally restricted information.
• Use aggregation, geographic displacement and controlled-access pathways when open release would create harm.
• Maintain version histories, correction procedures and named contacts so users can identify authoritative data.

For universities, farms, community groups and market operators:

• Agree on data ownership, consent, attribution and reuse rules before collection begins.
• Test datasets for representativeness across states, farming systems, cities, remote communities and vulnerable groups.
• Release code, dependencies and processing steps when security and commercial conditions permit.
• Support community governance and Indigenous data sovereignty rather than treating all information as a common public resource.

The lasting contribution of IASSIST 2017 is its insistence that data has a social and institutional life. Open-data policies for food security must connect technical access with public value. Australian systems can build on that foundation by making public information discoverable, protecting sensitive records and ensuring communities can participate in the governance of evidence about their own food systems.