The Catchment Data Deal
A small NZ technology company wants to commercialise a water-quality app for councils. It has negotiated access to 40 years of handwritten rainfall, turbidity and E. coli records kept by a hapū-led catchment group, plus 12 years of sensor data. The records include the locations of culturally sensitive spawning sites. The company's draft agreement offers a one-off payment of NZ$15,000 and an open-data licence that would let anyone reuse the dataset. The company also plans to train a predictive model on the data and patent the calibration method. The catchment group wants the data to remain under its control, wants any council using the app to report results back, and wants the spawning-site locations withheld. The company has a six-month runway and a council pilot worth NZ$120,000.
1. Identify the main ethical and legal issues in the draft agreement and recommend how the parties should change it.
Read the worked answer
The draft agreement treats the catchment group's records as a commodity, but the main issues are data sovereignty, collective ownership, informed consent, and benefit sharing.
- Ownership and control: the group's records are intergenerational and collectively held. A one-off payment with an open licence transfers control. The agreement should state that the group retains ownership and grants the company a limited, purpose-specific licence.
- Cultural IP and sensitive sites: the spawning-site locations must be withheld or aggregated. An open licence is incompatible with that requirement.
- Privacy: if any records identify individuals, the Privacy Act 2020 applies; consent and storage rules must be explicit.
- Benefit sharing: $15,000 against a $120,000 pilot is not a fair share. The group should receive a royalty, co-ownership of the model, or a paid data-steward role, and results should be reported back.
- Patent: patenting a calibration method trained on the group's data may appropriate mātauranga Māori and may fail novelty. Any patent should be joint and limited to genuinely new algorithm work.
The parties should replace the draft with a data governance agreement: tiered data access, a community licence, a review panel, and a clear exit clause. The company's six-month runway is a commercial pressure, not a reason to skip consent.
2. Evaluate the company's claim that an open-data licence is the fairest option, using the catchment group's stated concerns.
Read the worked answer
Open data can be fair, but not automatically. The company's claim rests on the idea that openness maximises reuse and transparency. That is true for some public datasets, but it does not answer the catchment group's concerns.
The group wants the data to remain under its control, wants reporting back, and wants the spawning sites withheld. An open-data licence would allow anyone to reuse the dataset, including the sensitive locations, and would remove the group's ability to decide who benefits. That is not fair to the group, even if it is open. Openness and equity are different things.
A better test is whether the licence protects the group's stated interests. It does not. The company could instead use a controlled licence: the data is available to the company and approved researchers, with attribution, a benefit-sharing clause, and a requirement to report results back. Sensitive sites could be aggregated or removed. If the company wants to publish, it can publish the model or aggregate results, not the raw records.
The company should also consider that its own commercial position depends on exclusive access. If it releases the data openly, it loses leverage and the group loses the ability to negotiate. The fair option is not 'open vs closed' but 'who controls access, and who benefits'. On the group's concerns, the draft open licence fails.
3. Explain how technological change (cheap sensors and machine learning) has shifted the power relationship between the company and the catchment group, and what that means for the project.
Read the worked answer
Cheap sensors and machine learning have changed the economics of this data. Forty years of handwritten records used to be a local archive with little commercial value; now they can be digitised, combined with sensor data, and used to train a model that councils will pay for. That shift gives the company power it did not have before: it has the capital, the algorithms, and the council market. The catchment group has the long record and the place-based knowledge, but less bargaining power.
Technology is not neutral here. The app's design choices - what data is collected, how it is aggregated, who can see the spawning sites, whether the model is retrained on community data - shape the relationship. If the company uses the data without control or benefit sharing, it turns mātauranga Māori into a raw material. If it designs for data sovereignty, it can use federated learning, on-site processing, or a data trust so the group keeps control.
The law lags behind the technology, so the parties cannot rely on copyright alone. The company's six-month runway is real, but a project built on disputed consent is a commercial risk as well as an ethical one. The judgement is that technological change has made a governance agreement necessary, not optional. The company should treat the group as a partner with a veto over sensitive data, not as a data supplier.