When an Unreviewed Paper Takes the Stand: Irrigation Science in a Western Water Court

Water rights disputes in the American West have always depended on documents: prior-appropriation records, diversion histories, engineering reports, and the testimony of people who know how water moves across a field. In spring 2026, a state water court in the Mountain West added a newer kind of document to that stack. A preprint on irrigation efficiency, posted openly in 2024 and never published in a peer-reviewed journal, became a central exhibit in a contest over how much water a large agricultural district could reasonably be expected to conserve.

The case did not turn on whether the preprint was elegant. It turned on whether an unreviewed working paper could bear the weight that adversarial litigation places on technical evidence. The answer, according to the public briefs and hearing transcripts, was neither a clean yes nor a clean no. The court admitted the paper for a limited purpose, allowed vigorous cross-examination of its methods, and ultimately treated it as one input among many. That outcome may be the most useful precedent for agricultural researchers who post early.

A paper enters the record

The preprint, titled Irrigation Efficiency Under Deficit Conditions in Forage Systems, combined field measurements from three farms with a model of consumptive use under reduced-water scenarios. Its authors argued that certain scheduling practices could stretch a given allotment further than the district's existing conservation plan assumed. The paper was plainly relevant to the dispute. It was also plainly incomplete: the authors described the work as a first-stage analysis, noted that one site had only a single season of data, and invited replication.

Opposing counsel moved to exclude the preprint, arguing that it lacked peer review, that its authors were not testifying as expert witnesses in the usual sense, and that its model assumptions had not been subjected to the scrutiny a journal would have provided. The district's lawyers countered that the paper was not offered as a final verdict on efficiency but as evidence that the district's plan had overlooked a plausible alternative. The distinction mattered. Courts routinely admit technical reports that are not journal articles, including agency memoranda, consultant studies, and field surveys. The question was not whether the preprint was perfect. The question was whether it was reliable enough to inform a factual finding.

What the judge allowed

The court declined to treat peer review as a gatekeeping requirement. In a written order, the judge noted that admissibility standards ask whether evidence rests on sufficient facts or data, whether it reflects reliable principles and methods, and whether those principles have been applied reliably to the case. Peer review can inform that inquiry, the order said, but it is not a precondition. The judge allowed the preprint's authors to testify about their methods and allowed opposing experts to challenge the model.

The ruling was narrower than it first appeared. The court limited the preprint's use to the question of whether the district had considered reasonable conservation alternatives. It did not adopt the paper's efficiency estimates as findings of fact. It also required the authors to disclose their data-processing scripts, which had been posted alongside the preprint, and to explain why one site's measurements had been excluded from the headline figure. That disclosure became the most consequential part of the proceeding. Because the preprint had been posted openly, opposing counsel could examine the same materials the authors had used. The litigation did not have to rely on a summary produced for court.

Cross-examination by PDF

In hearings, lawyers on both sides treated the preprint less as a published authority than as a contested dataset. Opposing counsel walked the lead author through the difference between applied water and consumptive use, pressed on whether the model had accounted for soil salinity, and asked why the paper's sensitivity analysis had not included a lower-bound scenario. The author conceded that the model was not designed to answer every question the court was asking. That concession did not destroy the paper's value. It clarified its limits.

The district's lawyers, meanwhile, used the preprint to argue that the existing conservation plan had been built on assumptions rather than measurements. They did not claim the paper proved the district could conserve a specific volume of water. They used it to show that a reasonable alternative had been available and documented. The court's final decree reflected that narrower use: it ordered additional monitoring and a revised conservation schedule, but it did not adopt the preprint's quantitative estimates as the governing standard.

Weight, not gatekeeping

For agricultural researchers, the case offers a practical lesson. Preprints do not enter legal proceedings as neutral objects. They enter as evidence, and evidence is tested by the parties who need it or fear it. The absence of peer review affected the weight the court gave the paper, not whether the court would hear it. What mattered more was whether the authors had documented their methods, whether the underlying data could be inspected, and whether the paper's claims matched the questions the court was actually deciding.

That is a familiar standard in policy analysis. A working paper can inform a decision without settling it. A preprint can be cited in a brief without being treated as a final scientific conclusion. The danger is not that courts will mistake a preprint for a journal article. The danger is that researchers will post work without the provenance that makes adversarial scrutiny possible. In this case, the public data and code did more for the paper's credibility than any claim of authority could have.

What it means for AgriXiv readers

The broader significance is institutional. Water rights, conservation compliance, and agricultural resource disputes are increasingly decided with technical evidence that moves faster than journal publication. If preprints are going to appear in those proceedings, they need to be built for inspection: clear version histories, accessible data, explicit limitations, and authors willing to defend their methods outside the seminar room. The court did not ask the preprint to behave like a journal article. It asked it to behave like evidence. That is a standard agricultural researchers can meet, and one that open servers are well placed to support.