Agricultural Technology IP Toolkit: Farm Data, Machinery, Agronomic Models, and Grower Terms

By Casey Scott McKay · 2023-10-22

A modern combine harvester generates more data per hectare than most software products generate per customer, and the question of who owns it has never been settled by anything except contract. This toolkit assembles the working material for practitioners advising equipment manufacturers, agronomy platforms, input suppliers, cooperatives, and growers. It covers farm data rights and the grower agreements that allocate them, the repair and software access fight that has reshaped machinery contracting, the agronomic models built on aggregated field data, and the trial data provenance that underpins every agronomic claim. It sets out the interaction with plant variety rights and seed contracts, the cooperative structures that complicate consent, and the clause language that makes the arrangements workable.

IP and Technology > Information Technology | Toolkit | Published 22 October 2023 - Updated 12 September 2024 | Casey Scott McKay - marksy.us

Summary. A combine generates more data per hectare than most software products generate per customer, and ownership of it is settled by contract or not at all. This toolkit covers farm data rights and the grower agreements that allocate them, the repair and software access dispute that reshaped machinery contracting, the agronomic models built on aggregated field data, the trial provenance behind every agronomic claim, the interaction with plant variety rights and seed contracts, and the cooperative structures that complicate consent.

Keywords: agtech IP · farm data ownership · precision agriculture · yield data · prescription maps · equipment telematics · right to repair · section 1201 exemptions · agronomic models · grower agreements · cooperative data · trial provenance · seed traits · variety protection · aggregated benchmarking · machinery interfaces

Start Here

Agricultural technology combines four legal traditions that were never designed to meet, and every problem in the practice arises where two of them touch.

Machinery is a durable good with software in it. A tractor lasts thirty years, is repaired locally, and now contains embedded control software with access restrictions. That collision produced the sector's most public intellectual property fight.

The field generates data about somebody else's business. Yield maps, soil measurements, application records, moisture, and machine telemetry describe the grower's operation in detail. The equipment manufacturer collects it; the agronomy platform analyses it; the input supplier wants it; and the grower, who generated it, frequently has the weakest contractual position of the four.

The biology is separately protected. Plant variety rights, plant patents, utility patents on traits, and seed contracts form a distinct regime with its own exhaustion rules and its own enforcement culture.

And the agronomic advice is a claim about the world, requiring trial evidence, which requires provenance, which requires knowing where the data came from and under what terms.

Four questions organise the work.

Who holds the farm data, who may use it, and for what?

What access does the equipment owner have to the software in the machine?

What may be built from aggregated field data, and what is being represented about it?

And how do the seed, trait, and variety rights interact with the data layer?

See The Data in the Dirt for the doctrinal treatment, Advising an Agricultural Technology Business for the sequence, and the Agricultural Technology IP Checklist for the working list.

Part one: farm data and who holds it

There is no property right in the data as such. Facts about a field are facts, and Feist Publications, Inc. v. Rural Telephone Service Co. forecloses ownership of them. A compilation may attract thin protection under 17 U.S.C. § 103 in its selection and arrangement. Everything else is contract, access control, and trade secret. See Selling Something You Cannot Own.

So stop asking who owns it and allocate uses instead.

Operational use — running the machine, servicing it, supporting the grower — is uncontroversial.

Grower-facing analytics, returning insight to the grower whose field it is, is the product the grower thinks they bought.

Aggregated benchmarking across growers is where the platform's value sits and where the grower's discomfort begins.

Sale or licensing to third parties — input suppliers, commodity traders, land valuers, lenders, insurers — is a separate question requiring separate consent and separate pricing, and it is the use growers most object to when it is discovered rather than disclosed.

The categories of data differ in sensitivity. Yield data reveals productivity and, by extension, land value and the grower's financial position. Application records reveal input purchasing. Machine telemetry reveals utilisation and maintenance. Field boundaries and operations reveal tenancy arrangements. Each has a different set of interested outsiders.

Consent quality is the recurring weakness. A grower who accepted a dealer's terms at the point of purchase, on a tablet, has consented to something. Whether that something includes contribution to an industry dataset, sale to an input supplier, or retention after the machine is sold depends on a document nobody read.

Landlord and tenant complicates it. A substantial share of farmland is rented. Data generated by a tenant on a landlord's field is claimed by both, and the lease says nothing.

Custom operators complicate it further. A contractor harvesting a hundred farms generates data about all of them on equipment they own, under terms nobody has read.

And industry principles exist but are voluntary. Codes of conduct on agricultural data have been published by industry bodies; they are useful drafting references and they are not law.

Part two: machinery, repair, and software access

The right to repair dispute is the sector's defining intellectual property fight and it is worth understanding precisely.

The grower's position. A machine is capital equipment with a thirty-year life, repaired seasonally, often urgently, frequently by the owner or a local mechanic. Software locks that require an authorised dealer to enable a repair convert an afternoon into a week during harvest.

The manufacturer's position. The software controls emissions compliance, safety systems, and machine calibration. Unauthorised modification creates safety, regulatory, and liability exposure, and diagnostic tooling is a legitimate service business.

The legal terrain.

Copyright in the embedded software is ordinary, and 17 U.S.C. § 117 provides limited rights to owners of copies of computer programs to make adaptations essential to use.

Section 1201 is the operative provision. 17 U.S.C. § 1201 prohibits circumvention of access controls, with statutory exceptions and a triennial rulemaking that has granted exemptions covering diagnosis, repair, and lawful modification of motorised land vehicles including agricultural equipment. The exemptions are narrow, time-limited, and do not authorise trafficking in circumvention tools. See The DMCA's Other Half and Navigating Section 1201.

Patent exhaustion limits what a manufacturer can control post-sale. Impression Products, Inc. v. Lexmark International, Inc. held that an authorised sale exhausts patent rights regardless of post-sale restrictions, and the repair-reconstruction line from Aro Manufacturing Co. v. Convertible Top Replacement Co. distinguishes permissible repair from impermissible reconstruction. See The Part That Broke.

Contract does the remaining work, through licence terms, warranty conditions, and dealer agreements — and state repair legislation increasingly constrains what those terms may say.

Practice notes for manufacturers. Provide a documented repair pathway with tools and information available on fair terms; separate genuinely safety- and emissions-critical functions from ordinary maintenance; avoid warranty terms conditioned on dealer-only service where they are unenforceable; and treat the memorandum of understanding route with grower organisations as a commercial decision rather than a concession. See the Aftermarket and Repair IP Checklist.

Practice notes for growers and independent repairers. Understand the exemption's scope and its limits; do not traffic in circumvention tools; document what was done and why; and preserve the machine's compliance-critical functions.

Part three: agronomic models and aggregated data

The model is the product. A prescription engine that tells a grower how much nitrogen to apply, at what rate, in which zone, is built from aggregated field data, trial results, weather, soil surveys, and agronomic literature. Its value is the aggregate.

Patent protection is weak. A method of recommending an input rate is an abstract idea implemented on a computer, disposed of by Alice Corp. v. CLS Bank International and Mayo Collaborative Services v. Prometheus Laboratories, Inc.. The surviving claims are to sensors, hardware, and specific technical implementations, following Enfish, LLC v. Microsoft Corp..

So trade secrecy carries it, under 18 U.S.C. § 1836 with the reasonable measures analysis of Rockwell Graphic Systems, Inc. v. DEV Industries, Inc..

Aggregation design determines whether the arrangement is defensible. Minimum contributor counts, no attribution to any individual grower or field, sufficient aggregation and lag, a documented methodology, and an audit right for contributors. These controls also address the competition question that arises when a platform aggregates data across growers who compete for land and for buyers.

Re-identification is a genuine risk, not a theoretical one. A field boundary is a fingerprint. Yield data tied to a boundary identifies a farm. "De-identified" agronomic data can frequently be re-identified from public parcel records, and platforms should test rather than assert.

Derived data terms are the recurring dispute. A grower who leaves a platform wants their data back and wants it removed from the models. The platform's position is that a trained model cannot be untrained. Address it at signature: what is exported, what is deleted, what remains in aggregate form, and for how long models trained on the contribution may remain in production.

Input supplier arrangements raise a specific problem, since a platform owned or funded by a supplier of seed, fertiliser, or chemicals has an interest in the recommendation. Disclose it, and expect scrutiny of recommendations that favour the owner's products.

And model outputs are advice. A prescription that produces a crop failure is a professional liability question before it is an intellectual property one, and the terms should address the standard, the disclaimer, and the insurance. See Buying a Model and the AI Procurement Checklist.

Part four: trial data, agronomic claims, and provenance

Every agronomic recommendation and every product claim rests on trial data, and the provenance of that data is both a regulatory and an intellectual property question.

Field trials are expensive, multi-season, and site-specific. Ownership of the resulting data should be settled before the trial, particularly where the trial is run on a cooperating grower's land, by a university, or by a contract research organisation.

University-run trials bring institutional policies that assign to the university and publication expectations that may conflict with confidentiality. Address both before the season starts. See From Laboratory to Licence.

Cooperating growers should be under written terms addressing confidentiality, data ownership, publication, and compensation.

Product performance claims are advertising claims requiring substantiation, and the substantiation is the trial data. 15 U.S.C. § 45 supplies the deception standard and 15 U.S.C. § 1125 supplies the competitor claim. An unsupported yield claim invites both. See Selling Green for the parallel analysis on environmental claims and the Environmental Claims and Cleantech IP Checklist.

Regulatory claims for crop protection products and biologicals carry their own data requirements and data protection periods, and the registration data has a value and a legal status distinct from copyright or trade secret.

Carbon and sustainability programmes now sit on the same trial and field data, with verification requirements, additionality analysis, and a rights question about who owns the sequestration credit — the grower, the landowner, or the programme operator. Address it in the grower agreement rather than in the programme rules.

And publication timing matters for patents. A trial result presented at a field day or an extension meeting is a public disclosure, starting the 35 U.S.C. § 102 grace period in the United States and foreclosing foreign rights. See Prior Art in a First-Inventor-to-File World.

Part five: seeds, traits, and varieties

The biological layer has its own regime and it interacts with the data layer more than practitioners expect.

Three protection routes. Plant patents under 35 U.S.C. § 161 for asexually reproduced plants; plant variety protection certificates for sexually reproduced varieties, with research and farmer-saved-seed exemptions; and utility patents on traits, plants, and seeds, confirmed in J.E.M. Ag Supply, Inc. v. Pioneer Hi-Bred International, Inc..

Utility patents are the strongest because they carry no farmer-saved-seed exemption, which was the point of the litigation that established their availability.

Exhaustion does not permit replanting. Bowman v. Monsanto Co. held that planting patented seed to produce a new generation is making a new infringing article, not a permitted use of a purchased one. The rule is specific to self-replicating technologies and is the foundation of the seed contract system.

Seed contracts do the operational work: single-season use, no saving, no research use, licensed trait stacks, stewardship obligations, and audit rights.

Enforcement is evidentiary. Field sampling, genetic testing, purchase records, and planting records. The culture of enforcement in this sector is unusually vigorous and unusually visible.

Variety naming interacts with trademark: the denomination of a protected variety is generic for the variety and cannot function as a mark, so a brand must be separate from the denomination. This trips up breeders regularly.

Genetic resource provenance matters for material sourced internationally, with access and benefit-sharing obligations that reach research use.

And the data layer intersects here: a platform that records which varieties were planted where, at what population, with what result, holds competitively valuable information about a seed company's performance, and the seed company will want terms about it.

See Owning a Living Thing, Protecting and Licensing Plant and Agricultural Innovation, and the Plant and Agricultural IP Toolkit.

Part six: cooperatives, tenancy, and the consent problem

Agricultural structures make consent harder than in almost any other data practice.

Cooperatives aggregate members' data by design, and the member agreement is the consent instrument. Address what the cooperative may do with member data, whether it may licence it onward, how a departing member's data is handled, and whether members may see aggregate outputs derived from their own contributions.

Tenancy splits the interest. Data generated on rented land describes the tenant's operation and the landlord's asset. Leases are silent, and the question arises when the tenancy ends or the land is sold. The workable answer is a lease provision allocating data rights, which almost no lease contains.

Custom operators and contractors generate data about many farms on their own equipment, under terms with the equipment manufacturer and none with the farms.

Lenders and insurers increasingly want field-level data as a condition of finance or cover, which raises the question of whether a grower can meaningfully refuse.

Buyers and processors impose traceability and sustainability data requirements down the supply chain, with contractual audit rights reaching the field.

Aggregators and brokers sit between all of them, assembling datasets from multiple sources whose consent chains differ.

Practice note. Map the consent chain for every dataset before building a product on it. The recurring failure is a platform that obtained consent from the equipment owner, built a product on data about the landowner's field, and licensed the output to a buyer with an interest in the grower's margin — three parties, one consent.

Clause bank

Grower data grant. Grower grants Platform a non-exclusive licence to use Farm Data solely to: (a) provide the Services to Grower; (b) support, maintain, and improve the Services; and (c) create Aggregated Outputs in accordance with the Aggregation Standard in Schedule [X]. Platform shall not: (i) sell, licence, or disclose Farm Data identifiable to Grower, a Field, or a Parcel to any third party; (ii) use Farm Data to inform any input pricing, land valuation, or lending decision affecting Grower; or (iii) retain Farm Data more than [period] after termination, other than in Aggregated Outputs already created.

Aggregation standard. An Aggregated Output must: (a) incorporate data from no fewer than [N] contributing operations, no one of which contributes more than [X] per cent of the underlying records; (b) exclude field boundaries, parcel identifiers, and any geospatial precision finer than [resolution]; (c) be lagged by no less than [period]; (d) be produced by a documented methodology retained by Platform and available to any contributor on request; and (e) be tested for re-identification against publicly available parcel records at least annually, with the test results retained.

Repair access. Manufacturer shall make available to Owner and to any repair provider nominated by Owner, on fair and non-discriminatory terms: diagnostic software sufficient to identify fault codes and to complete a repair; service documentation; calibration procedures; and any part-pairing authorisation required to return the Equipment to operation. Nothing in this Agreement conditions the Warranty on the use of an authorised dealer for maintenance or repair, save where the work affects an Emissions-Critical Function or a Safety-Critical Function listed in Schedule [Y].

Trial data. (a) Trial Data generated under this Protocol is owned by [Sponsor], and Cooperator hereby assigns all right, title, and interest in it. (b) Cooperator may use Trial Data relating to Cooperator's own operation for its own agronomic purposes. (c) Cooperator shall not publish or disclose Trial Data before [date / Sponsor's written consent], such consent not to be unreasonably withheld after the Publication Window opens. (d) Sponsor shall acknowledge Cooperator in any publication unless Cooperator requests otherwise. (e) Where the Trial supports a regulatory submission or a public claim, Sponsor shall retain the underlying records for no less than [period].

Carbon and sustainability rights. Rights in any Environmental Attribute generated by practices carried out on the Land during the Term, including carbon sequestration credits, belong to [Grower / Landowner] as between the parties. Programme Operator's rights are limited to those expressly granted in Schedule [Z] and terminate on expiry of the Programme. Programme Operator shall not register or claim any Environmental Attribute in its own name.

Exit and model treatment. On termination, Platform shall: (a) export all Farm Data in [format] within [30] days; (b) delete all Farm Data identifiable to Grower within [60] days and certify deletion; (c) exclude Grower's contributions from any Aggregated Output created after termination; and (d) with respect to models trained wholly or partly on Grower's Farm Data before termination, retire or retrain them within [period], during which such models may remain in production.

Worked scenarios

A platform discovered selling to an input supplier. An agronomy platform licenses aggregated yield and application data to a fertiliser company, which uses it for territory-level demand forecasting. A grower organisation obtains the arrangement and publicises it. The platform's terms permitted "aggregated and anonymised" use, and the aggregation was at county level with a low contributor count in sparsely farmed counties, from which individual operations were identifiable. Nothing in the contract was breached on a literal reading. The commercial consequence was severe. The fix — a documented aggregation standard with minimum contributor counts, a lag, a re-identification test, and disclosure of the categories of recipient — was available at the outset and would have cost nothing.

A repair dispute at harvest. A combine faults during a narrow harvest window. The fault code requires an authorised dealer to clear it and the nearest dealer is four days out. The grower's own mechanic can diagnose the mechanical fault but cannot return the machine to operation without the authorisation. The legal position is genuinely contested — the triennial exemption may cover the grower's own diagnosis, it does not authorise a tool vendor supplying a circumvention device, and the warranty condition purporting to require dealer service is unenforceable in several states. The practical answer for the manufacturer is a documented repair pathway that separates emissions- and safety-critical functions from everything else, because the alternative is legislation drafted by people who have just lost a harvest.

A tenancy that ends. A tenant farms a parcel for six years, generating detailed yield and soil data on a platform under their own account. The tenancy ends and the landlord asks for the data, arguing it describes their land. The tenant refuses, arguing it describes their operation. Both are right. The lease is silent. The platform holds the data and has no contractual basis to give it to either party alone. A one-paragraph lease provision would have resolved it; instead the parties argue while the incoming tenant starts from nothing.

A trial claim without provenance. A biological product is marketed on a yield uplift claim supported by trials run over three seasons on cooperating farms. A competitor challenges the claim. The sponsor cannot produce complete protocols, plot-level records, or the cooperating grower agreements for the earliest season, because the trials were run informally by a field agronomist who has since left. The claim is withdrawn. The underlying product may well work; the evidence was never assembled in a form that could be defended.

Failures that recur

Consent obtained from the equipment owner for data about a landowner's field.

Aggregation at a resolution that permits re-identification from public parcel records.

A grower agreement silent on what happens to models trained on the grower's contribution.

Warranty terms conditioning service on an authorised dealer where such terms are unenforceable.

Trial data with no protocol, no plot records, and no cooperating grower agreement.

A field day presentation that starts the grace period and forecloses foreign patent rights.

A variety denomination used as a brand, which cannot function as a mark.

Carbon credit rights unallocated between grower, landowner, and programme operator.

A cooperative licensing member data onward under a member agreement that never contemplated it.

Derived data terms drafted for a software product and applied to a thirty-year machine.

Platform terms accepted on a dealer's tablet at the point of purchase, never read, and relied on for a decade.

And a prescription engine treated as software when its output is agronomic advice with a liability profile.

Part seven: equipment interfaces and interoperability

Beyond repair sits a quieter and commercially larger question: whether machinery from different manufacturers can work together, and whether third-party implements and software can connect.

The physical and electronic interface between a tractor and an implement is standardised in principle and proprietary in practice, with extensions, profiles, and certification schemes that determine what actually works together.

Certification gates market access for implement makers in the same way that interconnection listing does in energy: an uncertified implement may function and will not be supported, and support is what a dealer sells.

Data export from the machine is the interoperability question growers care about. A yield monitor that writes a proprietary format that only one platform can read locks the grower to that platform regardless of the ownership language.

Reverse engineering for interoperability is available within limits. Sega Enterprises Ltd. v. Accolade, Inc. and the reasoning in Google LLC v. Oracle America, Inc. support reimplementation of interface elements in defined circumstances, and 17 U.S.C. § 1201(f) contains a narrow interoperability exception to the circumvention prohibition. Both require care and both are worth taking advice on before starting rather than after shipping. See Running a Reverse Engineering or Interoperability Program and the Interoperability and Reverse Engineering Checklist.

Standards participation carries the ordinary licensing commitments where a participant holds patents reading on the standard, with the injunction constraint of eBay Inc. v. MercExchange, L.L.C.. Link participation to portfolio management so the commitment is a decision.

Drafting response for growers and dealers. Require a documented, non-proprietary export format; require the data to be exportable without the manufacturer's cloud service; and require the export right to survive the end of any subscription. A data ownership clause without an export format is a sentiment.

Drafting response for manufacturers. Distinguish interface openness from software openness. Publishing a data format costs little and removes most of the lock-in complaint; opening the control software raises genuine safety and compliance questions. Concede the first early and the second becomes a narrower argument.

Part eight: diligence in agricultural technology

Whether the matter is an acquisition, an investment, or a platform partnership, the diligence has a predictable shape.

Consent chains. For every dataset, identify who consented, in what capacity, under which version of which terms, and whether the consenting party had authority over the land or the operation the data describes. Expect gaps between equipment owner, operator, tenant, and landowner.

Aggregation practice. Obtain the aggregation methodology, the contributor counts, and any re-identification testing. A platform that has never tested re-identification against public parcel records has an unquantified exposure and, frequently, a reputational one.

Onward licensing. Identify every third-party recipient of data or derived outputs, and test each against the consent that permitted it. Input suppliers, lenders, insurers, land valuers, and commodity traders are the recurring recipients.

Model provenance. Every production model, its training data, the rights under which each contribution was obtained, and what happens to models when a contributor leaves.

Trial records. Protocols, plot-level data, cooperating grower agreements, and the substantiation file for every public performance claim. This is routinely the weakest area and the one most likely to produce a competitor challenge under 15 U.S.C. § 1125.

Seed and trait licences where the business touches genetics, including field-of-use limits, stewardship obligations, and audit exposure.

Repair and warranty terms against the current state legislative position, since terms enforceable when drafted may not be now.

Open source in embedded software, with attention to copyleft obligations that a distributed device triggers. See Copyleft and Consequences.

Chain of title on the ordinary invention assignment analysis, which in this sector is complicated by heavy use of agronomists, contractors, and university collaborations. See the Employee Invention Checklist.

And carbon and sustainability programme rights, which are increasingly material and almost never allocated cleanly among grower, landowner, and operator.

Part nine: the grower's side

Most writing in this area is addressed to platforms and manufacturers. The grower's position deserves its own treatment, because it is negotiable more often than growers believe.

Read what was accepted at purchase. The platform terms agreed on a dealer's tablet are the operative document, and most growers have never seen them. Obtain the current version, in writing, and identify the data grant.

Ask for the export. A right to obtain one's own data in a documented, non-proprietary format, without a subscription, is the single most valuable term available and is frequently granted on request because refusing it is indefensible.

Ask who receives it. A list of the categories of third party receiving data or derived outputs, and the ability to opt out of onward licensing while continuing to use the service.

Ask about aggregation. Minimum contributor counts and geospatial resolution limits, so that county-level "aggregate" outputs do not identify a farm in a sparsely farmed county.

Ask about exit. Deletion on termination, exclusion from future aggregate outputs, and a stated position on models already trained.

Address the tenancy. Where land is rented, a lease provision allocating data rights costs a paragraph and prevents a dispute that will otherwise arise at the worst moment.

Address the custom operator. Where a contractor harvests the crop, the contract should say what happens to data generated on the grower's fields by the contractor's machine.

Keep the trial agreements. A grower cooperating in a trial should have written terms covering confidentiality, ownership, publication, acknowledgement, and compensation — and should keep the plot records, because they may be the only complete copy.

And negotiate collectively where possible. Cooperatives and grower organisations obtain terms that individual growers do not, for the same reason that collective bargaining produces stronger data provisions in other sectors: the counterparty can refuse one grower and cannot refuse all of them.

A ninety-day programme

Weeks 1–2. Build the data map. Every stream, every source, every system, every recipient. Record which party consented to each and in what capacity.

Weeks 3–4. Test the consent chain against the map and flag every dataset where the consenting party did not control the land or the operation the data describes.

Weeks 5–6. Document the aggregation standard and run a re-identification test against public parcel records. Adjust contributor counts and resolution until the test passes, and retain the results.

Weeks 7–8. Inventory the models: inputs, provenance, contributors, and exit treatment. Draft the exit clause and apply it to new agreements immediately.

Weeks 9–10. Review the repair and warranty position against current state legislation, and design the documented repair pathway if one does not exist.

Weeks 11–12. Assemble the substantiation files for every public performance claim, and stop making any claim whose file cannot be produced.

Throughout. Add the export-format obligation to every new agreement from week one, because it is the term that determines whether the relationship is a partnership or a lock-in — and because conceding it early is cheaper than conceding it under regulatory pressure later.

Part ten: the wider frame

Three developments will shape this practice over the next several years, and each changes the drafting.

Autonomy. Machines that operate without an operator in the cab shift the liability analysis, multiply the sensor data, and raise the question of who is responsible when a decision made by a model damages a crop or a person. The robotics analysis applies directly — component and stack mapping, training data provenance, safety documentation, and integrator terms. See The Machine That Decides and the Robotics and Autonomy IP Checklist.

Traceability down the food chain. Buyers, processors, and retailers are imposing provenance and sustainability requirements that reach the field, with audit rights and documentary obligations resembling those in the trade compliance regimes. A grower's data becomes a compliance artefact for somebody else's supply chain. See the Trade Compliance Checklist.

Environmental attributes as a distinct asset class. Carbon sequestration, water quality credits, and biodiversity units are being created from farming practices, verified by third parties, and traded. The rights question — grower, landowner, tenant, programme operator, or buyer — is genuinely unsettled and is being allocated by contracts drafted before anybody thought about it. Allocate it expressly in the grower agreement, the lease, and the programme terms, and expect the allocation to be tested.

And underneath all three sits the same structural fact with which this toolkit began: the field generates data about somebody else's business, nobody owns it, and everything therefore depends on what the contract says and on whether the party who signed it understood what they were agreeing to. A platform that treats that as a design constraint rather than as a nuisance will build arrangements that survive; one that treats consent as a formality will build arrangements that survive only until somebody publishes them.

Two documents worth keeping current

The consent map. One row per dataset: source, collecting device or system, the party who consented, that party's capacity — owner, operator, tenant, landowner, contractor — the agreement and version relied on, the permitted uses, the recipients, the retention period, and the exit treatment. This single document answers the question that every grower dispute, every regulatory enquiry, and every acquisition diligence exercise begins with, and it is almost never maintained.

The claims substantiation file. One folder per public performance claim: the claim as made, the trials relied on, the protocols, the plot-level records, the cooperating grower agreements, the statistical analysis, and the date of the last review. A claim without a complete file is a claim that will be withdrawn the first time a competitor writes a letter, and the withdrawal is worse than never having made it.

One paragraph to remember

In agricultural technology nobody owns the data, so everything turns on the contract and on whether the person who signed it had authority over the land and the operation the data describes. Publish an export format and most of the lock-in complaint disappears; document an aggregation standard and test it against public parcel records; allocate carbon and environmental attributes expressly before somebody else does; keep the trial records that support every claim; and build a documented repair pathway, because the alternative is legislation written by people who have just lost a harvest.

The five-question test

The quickest diagnostic on an agricultural technology position takes an afternoon. Pick one dataset the business relies on and ask five questions.

Who consented to its collection, and did that party control the land and the operation it describes? What does the agreement they accepted actually permit, in the version then in force? Who has received it or an output derived from it, and was each recipient within the permitted scope? If a contributor left tomorrow, what would be exported, what deleted, and what would remain in the models? And can the aggregation applied to it survive a re-identification test against public parcel records?

A business that answers all five in an afternoon has a defensible position. One that answers two has the ordinary position and a year of quiet work ahead. One that answers none will learn the answers from a grower organisation, a journalist, or an acquirer's counsel — and in each case the answers will be published before they are negotiated.

A note on tone

Growers are a sceptical audience with long memories and effective organisations, and the sector's data arrangements have already produced one public dispute that reshaped machinery contracting. Advisers who treat grower consent as a formality tend to be advising the businesses that end up in the trade press. Advisers who treat it as a design constraint tend to be advising the ones that keep their customers.

Key Authorities at a Glance

Data. Feist Publications, Inc. v. Rural Telephone Service Co.; 17 U.S.C. § 103; 18 U.S.C. § 1836 and 18 U.S.C. § 1839 with Rockwell Graphic Systems, Inc. v. DEV Industries, Inc.; 18 U.S.C. § 1030 with Van Buren v. United States.

Software and repair. 17 U.S.C. § 102; 17 U.S.C. § 117; 17 U.S.C. § 1201 with its statutory exceptions and triennial exemptions; Sega Enterprises Ltd. v. Accolade, Inc. and Google LLC v. Oracle America, Inc. on interoperability.

Patent and exhaustion. 35 U.S.C. § 101 with Alice Corp. v. CLS Bank International, Mayo Collaborative Services v. Prometheus Laboratories, Inc., and Enfish, LLC v. Microsoft Corp.; 35 U.S.C. § 102; 35 U.S.C. § 271; Impression Products, Inc. v. Lexmark International, Inc.; Aro Manufacturing Co. v. Convertible Top Replacement Co..

Plants and seeds. 35 U.S.C. § 161; J.E.M. Ag Supply, Inc. v. Pioneer Hi-Bred International, Inc.; Bowman v. Monsanto Co.; Diamond v. Chakrabarty. Variety protection under 7 U.S.C. § 2402.

Claims and advertising. 15 U.S.C. § 45; 15 U.S.C. § 1125; trademark registration under 15 U.S.C. § 1051 and 15 U.S.C. § 1052.

| Authority | Governs | Practical consequence | | --- | --- | --- | | Feist | Facts | Farm data is contractual, not owned | | 17 U.S.C. § 1201 | Access controls | The repair exemption and its limits | | 17 U.S.C. § 117 | Owner adaptations | Narrow but real | | Impression Products | Exhaustion | Post-sale restrictions are contractual only | | Aro | Repair v. reconstruction | Where permitted repair ends | | Bowman v. Monsanto | Self-replicating technology | Replanting is making, not using | | J.E.M. Ag Supply | Utility patents on plants | No farmer-saved-seed exemption | | 35 U.S.C. § 161 | Plant patents | Asexual reproduction route | | Alice | Eligibility | Agronomic methods are not patentable | | 18 U.S.C. § 1836 | Trade secrets | Where the models are protected | | 15 U.S.C. § 45 | Deception | Yield claims need trial substantiation | | 35 U.S.C. § 102 | Grace period | A field day is a public disclosure |