Golden Rice Cleared Every Safety Review by 2021. What Has Kept It From Farmers Since Is Litigation, Not Science.

On 21 July 2021 the Philippine Department of Agriculture's Bureau of Plant Industry issued the world's first biosafety permit for the commercial propagation of Golden Rice. It was the fifth regulatory decision on the GR2E event in four years, and the fifth to reach the same conclusion: food from this rice is as safe as food from any other rice, with one difference, which is that it contains beta-carotene. Over the next two wet seasons the Philippine Rice Research Institute grew it on 202 hectares in 24 provinces, seventeen of which had asked to be included, and delivered 77 tonnes of milled grain to households with young children and lactating mothers in the areas of the country where vitamin A deficiency is worst. Then, on 17 April 2024, the Court of Appeals in Manila revoked the permit, ordered PhilRice to cease and desist, and enjoined the programme until the government could furnish "proof of safety." It did so at the request of two advocacy organisations and their co-petitioners, on a reading of the precautionary principle, and without finding that the rice had harmed anyone or anything. The government appealed to the Supreme Court in October 2024. As this article goes to press, two and a half years after the writ was first issued, the court has not ruled and the programme is on hold.

This article makes a narrower claim than the one Golden Rice's defenders usually make, and a sharper one. The usual claim is that anti-GMO advocacy has cost the world a quarter-century. That claim is not defensible, and the first section below concedes why: from the 2000 proof-of-concept until roughly 2013, Golden Rice was late because breeding it was hard, and the scholars who say so are right. The claim this article defends is that from 2013 onward every material delay to Golden Rice reaching the Filipino farmers and children it was built for is traceable to the actions of named organisations, in the field, in the press, and in court, and not to any new scientific finding, any regulator's own safety judgement, or any failure of the product in the ground. Between December 2017 and July 2021 four regulators on three continents examined the same event and cleared it. Between 2022 and 2023 it was grown at scale and eaten. What ended that was a Writ of Kalikasan, an environmental remedy that the petitioners sought, the Supreme Court issued, and the Court of Appeals enforced. The science did not change in 2024. The forum did.

The argument runs in six steps. First, the concession: what the breeding record actually shows and why admitting it strengthens rather than weakens the case. Second, 2013 as the hinge year, when the technical explanation ran out and the obstruction began. Third, the four regulatory reviews, what each tested, and why "no consensus on safety" is a factual error rather than a point of view. Fourth, the pilot, which closed the field question. Fifth, the court decision read closely: what the judges actually held, what they did not, and why a precautionary-principle injunction against a product with four completed safety assessments is an advocacy outcome wearing judicial robes. Sixth, the two honest objections, Bangladesh and the Tufts trial, taken at full strength. The conclusion is not that courts should never hear environmental cases, nor that advocacy organisations should be silent. It is that a legal system which lets a completed risk assessment be reopened indefinitely, on no new evidence, by any party that disagrees with the result, has stopped regulating risk and started regulating permission.

The Concession: 2000 to 2013 Was a Breeding Problem

The paper that made Golden Rice famous was published in Science in January 2000. Ingo Potrykus, Peter Beyer and colleagues had put two genes, a daffodil phytoene synthase and a bacterial carotene desaturase, into the rice endosperm, the starchy part of the grain that humans eat, and the endosperm turned yellow. It was a landmark in plant metabolic engineering. It was not a nutritional intervention. The best lines accumulated about 1.6 micrograms of carotenoids per gram of dry grain. Against the daily requirement of a small child, and allowing for the fraction of beta-carotene a body actually converts to retinol, a child would have needed to eat an implausible quantity of that rice to make a difference. Critics who said so in 2000 and 2001 were not obstructing anything. They were correct.

The fix took five years. In 2005 a Syngenta group led by Jacqueline Paine reported in Nature Biotechnology that replacing the daffodil synthase with a maize one raised carotenoid accumulation more than twenty-fold, to around 37 micrograms per gram. This was "Golden Rice 2," and it is the generation from which every line now in regulatory dossiers descends. But a transformation event is not a variety. The gene construct has to be introgressed, by repeated backcrossing, into the high-yielding varieties farmers already grow, and the specific site at which the construct inserted into the rice genome has to be one that does not disrupt anything the plant needs. The lead event through most of the following decade was GR2-R1. In multi-location trials in the Philippines in 2012 and 2013 it yielded below its conventional comparators, and the International Rice Research Institute moved the programme onto a different event, GR2E, which did not. The reason for GR2-R1's failure was later established in a paper from the Indian Agricultural Research Institute: in the mega-variety Swarna, backcross lines carrying GR2-R1 were dwarf, pale, and drastically lower-yielding, because the construct had inserted inside OsAux1, an auxin transporter gene. The trait was fine. The address was wrong.

This is the record that Glenn Davis Stone and Dominic Glover drew on in their 2016 paper in Agriculture and Human Values, which argued that the popular story blaming activists for Golden Rice's failure to arrive did not survive contact with the breeding data: the product had simply proved difficult to breed into varieties that grow well in the Philippines. For the period they were examining, which ended before any regulator had ruled on GR2E, that conclusion is sound and this article adopts it. Thirteen years of delay, from the 2000 paper to the selection of a usable event, belong to the ordinary difficulty of plant breeding, to an early generation that under-delivered on dose, and to an event-selection decision that took longer to get right than anyone hoped. Advocacy organisations did not cause that. Anyone who wants to blame them for it is giving them more credit than they are due.

Making this concession is not a tactical retreat. It is what turns an accusation into an argument. The question that matters for policy is not "has Golden Rice been slow?" but "what, specifically, has been slowing it, and when?" A thirty-year grievance cannot answer that question, because it bundles the breeders' delay with everyone else's. A dated account can. From 2013 the breeding problem was solved, and from that point the record of what stood between the rice and the field consists of a different kind of entry: a trampled trial plot, a press statement, a petition, a writ, a judgment. Those are the entries this article is about.

2013: The Year the Explanation Changed

On 8 August 2013 a group of several hundred people entered the Department of Agriculture's regional field office compound at Pili, in Camarines Sur, one of five sites where GR2 lines were being evaluated under a biosafety permit, and destroyed the trial. The Department's regional office said that members of militant organisations had "deliberately uprooted and trampled the rice plants in the experimental plots." Who exactly pulled the plants was disputed at the time, and this article does not need to settle it, because the movement that organised the day has since settled it on its own behalf. In August 2026 the Manila-based outlet Bulatlat, reporting from the farmers' side, described the Pili action as planned and successful, quoted a participating farmer saying the farmers "rushed in like flood," and noted that 8 August has since been declared an International Day of Action Against Golden Rice. An act that its participants commemorate annually is not an act whose attribution is in serious doubt.

The immediate scientific cost of Pili was limited: the other four sites completed, and the trial data that went into every subsequent dossier came from them. The cost that mattered was of a different kind. The destruction of a permitted field trial, in a country whose biosafety rules require exactly such trials before any further step, announced that the opposition to this product would not confine itself to the regulatory process. It would act outside it, and it would do so after the breeding problem had been solved rather than before. That is the significance of the date. The same twelve months that saw IRRI abandon GR2-R1 for an event that worked saw the first physical attack on the evidence-gathering that the new event required. The two tracks on the hero graphic above diverge here.

Three years later the same pattern appeared in print. On 29 June 2016 a letter signed by more than a hundred Nobel laureates called on Greenpeace to end its opposition to Golden Rice and asked, in its most quoted line, how many people must die before the campaign counts as a "crime against humanity." The letter's rhetoric was stronger than its argument needed to be, and this article does not endorse the phrase. What matters is the reply. Greenpeace Southeast Asia's campaigner responded the next day that the organisation was not blocking Golden Rice because the product "has failed as a solution and isn't currently available for sale." Read that against the timeline. In June 2016 the GR2E dossier was already with Food Standards Australia New Zealand, which would approve it eighteen months later; Health Canada and the US FDA would follow within six months of that. The product was not available for sale because its approvals were pending. The organisation that had spent fifteen years opposing those approvals was citing their absence as proof that the product did not work. That is not an argument from the science. It is an argument from the obstruction.

Four Regulators, Four Clean Reviews

A food-safety assessment of a genetically engineered crop is not a vote of confidence. It is a specified set of tests. The assessor asks what new proteins the crop makes, whether any resembles a known allergen or toxin, whether the crop's composition, its proteins, fats, minerals, vitamins and anti-nutrients, differs from its conventional counterpart beyond the range of ordinary varietal variation, and whether the inserted DNA is stable across generations and inserted where the developer says it is. The GR2E dossier that IRRI assembled from the Philippine field trials answered those questions, and between 2017 and 2019 four regulators marked it.

Food Standards Australia New Zealand went first. Its approval report on Application A1138, dated December 2017, concluded that food derived from GR2E rice is as safe for human consumption as food from conventional rice varieties. FSANZ was explicit that the approval was not about putting Golden Rice on Australian shelves; it was to prevent trade disruption if GR2E were ever found in imported milled rice. That makes it the most disinterested of the four reviews. An agency with no stake in the product's success, assessing it only so that its presence would not trigger a border rejection, found nothing to object to. Health Canada reached the same conclusion on 16 March 2018, stating it had no objection to the food use of GR2E, and the US FDA concluded its consultation on 24 May 2018 concurring with IRRI's assessment of the rice's safety and nutrition.

The Philippine review was the one that counted, because the Philippines was where the rice was going to be eaten. The Bureau of Plant Industry issued the biosafety permit for direct use as food, feed and processing on 18 December 2019, after a review under the national biosafety framework that drew on the completed multi-location field trials at Muñoz and San Mateo. The commercial propagation permit of 21 July 2021 followed a separate environmental assessment, because growing a crop raises questions, about gene flow, about weediness, about effects on non-target organisms, that eating it does not. Both permits were issued by the agencies the law designates, under the procedures the law prescribes, on the evidence the law requires. That is what "regulatory approval" means, and it is the fact that any account of the subsequent litigation has to start from.

It is also the fact that disposes of the phrase "no scientific consensus on safety," which appears in nearly every statement the petitioners have made. Consensus is a description of the state of expert assessment. Four agencies with statutory mandates, independent expert committees, and no common interest reviewed the same data and reached the same answer; no agency has reached the opposite one. Whatever else is unsettled about Golden Rice, including its storage stability, its uptake, its long-term agronomic performance across environments and its place in a wider nutrition strategy, the food-safety question is not unsettled. To say that there is no consensus is to say that the opinion of an advocacy organisation counts against the completed assessments of four regulators. It does not, and no court should have treated it as if it did.

It Worked in the Field

The strongest evidence against the claim that Golden Rice "has failed as a solution" is that for two years it was a solution, in operation, with data. The account of the Philippine pilot deployment was published in Scientific Reports in May 2026 by Ronan Zagado and seventeen co-authors at PhilRice and IRRI, and it is the first peer-reviewed description of Golden Rice being grown and eaten at scale anywhere. Its figures deserve to be stated plainly because they are the figures the court order interrupted.

From the 2022 wet season to the 2023 wet season, Malusog Rice, the name under which the GR2E variety was registered in the Philippines, was grown on 202 hectares across 24 provinces. Seven provinces were selected on nutritional need, using provincial stunting data as a proxy; the other seventeen joined because their local governments asked to, after briefings, which is not the profile of a product being forced on anyone. Average yield across all on-farm plots was 3.9 tonnes per hectare, with a maximum of 8.3, against a national average palay yield that the Philippine Statistics Authority put at roughly four tonnes in the same period. The carotenoid content of the harvested grain was within the range needed for a cup of cooked rice to supply 30 to 50 percent of a young child's estimated average requirement for vitamin A. More than 77 tonnes of milled rice went to households identified by provincial health offices as at risk, specifically those with lactating women and children aged six months to five years; eight tonnes went to market testing. Of 694 beneficiary households surveyed in the seven priority provinces, 97 percent said they wanted to keep eating it when it became widely available, 97 percent were willing to pay for it, and 95 percent were satisfied with how it stored, cooked and ate.

None of this proves that Golden Rice would, on its own, eliminate vitamin A deficiency in the Philippines, where the Department of Science and Technology's nutrition institute puts the prevalence among children under five at 15.5 percent, about two million children. The authors do not claim it would; they describe it as a complement to supplementation and fortification, not a replacement. What the pilot proves is narrower and sufficient: the variety grows in farmers' fields at yields comparable to what they grow now, accumulates the nutrient it was built to accumulate, and is accepted by the households it was built for. The field question, which was a legitimate open question in 2013, was closed by 2023. Everything that has happened since has happened to a product that works.

The Writ of Kalikasan: What the Court Actually Held

In October 2022, fifteen months after the propagation permit and in the middle of the first pilot season, a petition for a Writ of Kalikasan reached the Supreme Court of the Philippines. The petitioners were the farmer-scientist network MASIPAG, Greenpeace Southeast Asia–Philippines, the Southeast Asia Regional Initiatives for Community Empowerment, the peasant movement Kilusang Magbubukid ng Pilipinas, two smaller organisations, and eight named individuals. The respondents were the Secretaries of Agriculture, Environment and Health, the Bureau of Plant Industry, PhilRice, and the University of the Philippines Los Baños, which held the permit for Bt eggplant. The petition asked the court to stop the commercial propagation of both crops. On 18 April 2023 the Supreme Court, sitting en banc, issued the writ and ordered the respondents to file a return within ten days; in June it referred the case to the Court of Appeals to hear and decide. The writ is not a judgment. It is the court's determination that the petition states a case serious enough to be heard. But it was enough to put the programme under a cloud eighteen months before any judge had ruled on the evidence.

The Fourth Division's decision in CA-G.R. SP No. 00038, promulgated on 17 April 2024, runs to 143 pages. Its dispositive portion is eight numbered orders. It grants the privilege of the writ; issues a writ of continuing mandamus against the government agencies; directs UPLB to cease and desist from propagating Bt eggplant; revokes PhilRice's biosafety permit for the commercial propagation of Golden Rice; orders PhilRice to cease and desist from propagating it; enjoins both crops "until such time that the concerned respondent government agencies submit proof of safety and compliance with all legal requirements"; orders the agencies to report to the court on their monitoring mechanisms and on measures to strengthen risk assessment under the 2021 Joint Department Circular; and, in its eighth order, enjoins any application for contained use, field testing, food or feed use, propagation or importation of any genetically modified organism until that reporting is done. On 15 August 2024, on the government's motion for reconsideration, the court deleted the eighth order, accepting that a nationwide freeze on crops that were not before it violated due process. Every order concerning Golden Rice stood. The Office of the Solicitor General petitioned the Supreme Court for review on 17 October 2024.

What did the court find? Two things, and it is important to be exact about both. First, it applied the precautionary principle as the Rules of Procedure for Environmental Cases instruct: where, in the court's judgement, there is "a lack of full scientific certainty in establishing a causal link between human activity and environmental effects," the burden of evidence shifts to the party carrying out the activity, and the activity may be stopped until that burden is met. The court heard expert witnesses from both sides, including testimony from the petitioners' experts on unintended effects and on the regulatory review process, and concluded that the required certainty was lacking. Second, it examined the procedural history of the earlier biosafety permit for the Golden Rice field trial, found that permit to have been "improperly issued," and reasoned that because the propagation permit "has as its prerequisite the satisfactory and valid conduct of the field trial," it fell with it. The expiry of the field-trial permit, the court said, "is not a reason to allow its improper issuance to go unnoticed."

What the court did not find is just as important. It did not find that GR2E rice is unsafe to eat. It did not find that it harms the environment. It did not find that any of the four regulatory assessments was wrong on the science, or identify any hazard those assessments had missed. It did not find that anyone had been harmed by the 77 tonnes already eaten. It found that the state of scientific certainty, as the court assessed it from the testimony before it, fell short of a standard that the precautionary principle, as the Philippine rules frame it, allows a court to demand. That is a finding about a legal threshold, not about rice. And the "proof of safety" the court now requires before the injunction lifts is a proof that four regulators have already accepted, assessed against a standard the court did not define.

Why call this an advocacy outcome rather than a legal one? Because of what the precautionary principle does when it is applied to a product that has completed its risk assessment. The principle is sound where it belongs: at the front of a process, when a new activity is proposed, little is known, and the cost of waiting for evidence is low relative to the cost of being wrong. It was invoked at the front of this process, and the answer was the field trials, the compositional analyses, the allergenicity screens, and the environmental assessment that four regulators then reviewed. To invoke it again at the end, after those assessments have been done, on the ground that "full scientific certainty" has still not been reached, is to set a bar that no agricultural product, conventional or engineered, has ever cleared or could. Full certainty is not available for any variety of rice. A rule that demands it for one variety and not others is not a safety rule. It is a rule about who gets to decide, and in this case the answer was: whichever party is prepared to file. That is the sense in which the outcome belongs to the petitioners rather than to the science. The petitioners did not bring the court a hazard. They brought it a standard, and the standard did the work.

As of this writing the Supreme Court has not decided the government's petition. A peer-reviewed survey of agricultural biotechnology litigation published in August 2025 by Drew Kershen in Frontiers in Bioengineering and Biotechnology describes the Court of Appeals ruling as "apparently stayed (in abeyance)" while the appeal proceeds. The pilot paper, written by the people running the programme, says further rollout waits until Malusog Rice "has been cleared from any legal cases." Reporting from the petitioners' side in August 2026 says the high court "is yet to decide on the appeal" and that the programme is on hold. Three sources with opposite interests agree on the fact. Two and a half years after the writ, the rice that four regulators cleared and 24 provinces grew is in storage, and the children it was grown for are two and a half years older.

The Bangladesh Objection

The honest objection to this article's thesis is a country, not an argument. Bangladesh has no Writ of Kalikasan, no comparable litigation, and an anti-GMO movement far weaker than the Philippines'. In 2013 it became the first country to approve a genetically engineered food crop in South Asia when it released Bt brinjal. The Bangladesh Rice Research Institute bred GR2E into BRRI dhan29, the country's most widely grown dry-season variety, and tested it in confined multi-location trials at five sites. The Golden Rice line yielded 7.0 tonnes per hectare against 7.0 for the conventional parent, with no significant difference in any agronomic or grain-quality trait measured. BRRI submitted its dossier to the Ministry of Agriculture on 26 November 2017 and to the Ministry of Environment a week later. The 2025 USDA Foreign Agricultural Service report on Bangladesh lists Golden Rice among several events still "waiting for regulatory approval," describes the national biosafety committees as still reviewing it, and characterises biotechnology policy in the country as "at a standstill." Nine years without a decision, and nobody sued anybody.

So regulatory inaction also delays Golden Rice, and an article whose title says "litigation, not science" owes the reader an account of why Bangladesh does not refute it. The answer is that the title is a claim about the Philippines, and Bangladesh shows the limit of the claim rather than its falsity. The thesis is not that advocacy is the only thing that can stall a biofortified crop. It is that in the one country where the regulatory system did its job, said yes, and let the crop into the field, what reversed that decision was litigation brought by advocacy organisations on no new evidence. Bangladesh is the control case for a different proposition: that a regulator which never decides can be as costly as a court that reverses a decision. That proposition is true, and it deserves its own article. It is also worth noticing what the Bangladesh committees have not done. They have not found the rice unsafe. They have not rejected the dossier. They have let a file sit, through a change of government and the death of a committee member, for longer than it took IRRI to breed the variety. Inaction is a regulatory failure; it is not a scientific finding, and it does not supply the petitioners in Manila with one.

The Tufts Trial

The second objection is an episode that the Golden Rice programme's supporters would rather not discuss and that its opponents discuss constantly. In 2008 a team led by Guangwen Tang of Tufts University, with Chinese collaborators, fed Golden Rice to schoolchildren in Hunan province to measure how efficiently its beta-carotene is converted to retinol. The paper, published in the American Journal of Clinical Nutrition in 2012, reported that the conversion was about as efficient as that of beta-carotene delivered in oil, which was good news for the product. The problem was not the result. A Tufts review found that the consent process had not met the university's ethical requirements: parents had not been clearly told that the rice was genetically engineered, and approvals had not been handled as the protocol required. In 2015 the journal retracted the paper on those grounds. This was a genuine failure of research ethics, it was handled badly for years, and it did lasting damage to the programme's standing in China and beyond. Nothing in this article excuses it.

What the episode does not do is bear on whether Golden Rice is safe, whether it works, or whether the 2024 injunction was justified. The retraction was for consent, not for data; the measurement it reported has not been contradicted. The children were not harmed by the rice. The regulatory dossiers that four agencies reviewed did not rest on the Tufts study. Yet in the petitioners' public case the Tufts affair functions as evidence against the product itself, cited alongside speculative hazards as if an ethics violation in a 2008 feeding study were a finding about the rice in a 2022 field. That transposition, from "the people who tested this behaved badly" to "this is dangerous," is the characteristic move of the campaign, and it works precisely because the first half is true. An article that indicts advocacy has to say so plainly: the movement's best evidence is real, and it is evidence of something other than what it is used to prove.

Counting the Cost Honestly

Vitamin A deficiency impairs immune function, raises mortality from measles and diarrhoeal disease, and in its severe form causes xerophthalmia and irreversible blindness. The Philippine nutrition surveys put its prevalence among children under five at 15.5 percent. The question any argument about delay has to answer is how much of that burden Golden Rice would have relieved had it not been delayed, and here the honest answer is that nobody knows precisely, because the counterfactual depends on adoption rates, dietary patterns, carotenoid retention in storage, and the reach of the supplementation and fortification programmes the rice was meant to complement. Readers will encounter figures, some of them in the defenders' own literature, of hundreds of thousands of children blinded each year by the delay. Those figures are not traceable to a primary source and this article does not use them.

What exists instead is a model. In 2014 Justus Wesseler and David Zilberman published in Environment and Development Economics a real-options analysis of Golden Rice for India, treating the decision to approve as an investment under uncertainty and estimating what the opposition would need to believe about the product's risks for delay to be rational. Their central estimate was that a decade's delay in India had cost on the order of 1.4 million life-years, and that opponents were implicitly valuing the perceived costs of the product at around US$199 million a year. The estimate depends on assumptions the authors state: a particular adoption trajectory, a particular efficacy against deficiency, a particular valuation of a disability-adjusted life-year. Stone and Glover, among others, have disputed the adoption assumptions, and the point is well taken; a product that was not yet bred into Indian varieties could not have been adopted on the schedule the model assumes. But the dispute goes to the size of the number, not its sign. On any adoption assumption that is not zero, the cost of keeping a working micronutrient intervention out of the diets of deficient children is counted in healthy years of life, and it accumulates for every year the Supreme Court does not rule.

The honest version of the number is therefore this: large, uncertain, and growing. It is not the headline figure, and it should not be. It is the figure that survives the kind of scrutiny the petitioners' claims have never been subjected to in court.

The Counterargument, Acknowledged

The fairest version of the opposing case has four parts, and each deserves to be stated at full strength. First, vitamin A deficiency has fallen substantially since the 1990s through supplementation capsules, fortified oil and sugar, and dietary diversification, and these interventions are cheaper, proven, and already running. A biofortified staple is one tool among several and was never going to be the decisive one; Greenpeace's argument that the money would do more good elsewhere is an argument about priorities that a reasonable person can hold. Second, beta-carotene degrades in stored grain, and the fraction that survives months in a village store and a long boil is lower than the fraction in a freshly harvested sample. The pilot paper acknowledges this, citing the Zambian maize experience, and relies on the finding that meaningful amounts survive. The margin is real and so is the concern. Third, the Tufts affair was a real ethics failure in a programme that asked poor families to trust it, and trust is not restored by a retraction. Fourth, the petitioners' procedural complaints about how the field-trial permit was issued, which the Court of Appeals accepted, concern public consultation and documentation requirements that exist for good reasons; a regulator that cuts corners on process invites exactly the litigation it then loses. This article's claim survives all four. None of them is a finding that the rice is unsafe or that it does not work, and none of them was the basis of the court's injunction, which rested on the precautionary principle. A product can be imperfect, secondary, and badly handled by its champions, and still be a product that four regulators cleared and that a court stopped on no evidence of harm.

What Proportionate Governance Would Look Like

The first article in this series argued that regulation of crop biotechnology should be product-based and proportionate to risk, not process-based and indifferent to it. The Golden Rice case shows what happens downstream of a regulatory system that gets that right and a legal system that does not. Five features would have changed the outcome without weakening any genuine protection.

Precaution with an expiry date. The precautionary principle should bind at the start of a risk assessment and be discharged by its completion. A court invoking it against a product with a completed assessment should be required to identify what the assessment missed. "Full scientific certainty" is not a standard; it is the absence of one.

Deference to completed risk assessment absent new evidence. A litigant seeking to reopen a regulatory approval should have to bring the court something the regulator did not have: a new study, a new hazard, a documented harm. The Golden Rice petitioners brought testimony about hypothetical unintended effects that the regulators had already considered. That should not have been enough to revoke a permit, and in most jurisdictions it would not have been.

Procedural defects cured by procedure, not by injunction. If a field-trial permit was issued without a required consultation, the remedy is to hold the consultation. Revoking a downstream propagation permit, and with it a running nutrition programme, for a documentation defect in an expired upstream permit is a disproportion that no risk-based system would produce.

A clock on the court. The Rules of Procedure for Environmental Cases require the Court of Appeals to decide a Writ of Kalikasan case within sixty days of submission. No comparable clock runs on the Supreme Court, and the appeal has now been pending for two years. Where an injunction stops a public-health programme, delay is not neutral. It is a decision, and it should have a deadline.

Regulators that keep their own process clean. The petitioners won on procedure because there was procedure to win on. An agency administering a product this contested should treat every consultation, posting and endorsement requirement as a condition it cannot afford to miss. Bangladesh's lesson belongs here too: a regulator that will not decide hands the product's opponents a victory without a hearing.

Conclusion: A Pending Case Is the Live Test

The story this article has told can be checked at every date. Until 2013 Golden Rice was late because its developers had not yet produced an event that delivered the nutrient without costing the yield, and the scholars who said so were right. From 2013 the developers had such an event, and from that year the record of what kept it from the field contains no scientific setback. It contains a destroyed trial that its organisers now commemorate, a press statement citing the product's unavailability as proof of its failure while opposing the approvals that would make it available, a petition, a writ, and a judgment that revoked a completed regulatory approval on a precautionary standard the court did not define and no product could meet. Two and a half years on, the Supreme Court of the Philippines holds the only decision that matters, and it has not made it. When it does, it will either restore a regulator's completed risk assessment to the status that regulators' assessments are supposed to have, or it will confirm that in the Philippines any approved crop can be stopped indefinitely by any party willing to file, on no new evidence, for as long as the courts take. Fifteen and a half percent of Filipino children under five are deficient in vitamin A. Seventy-seven tonnes of rice that would have helped a few thousand of them has already been eaten, with no ill effect anyone has documented. The rest is in storage, waiting on a ruling. That is not a scientific state of affairs. It is a legal one, and it was chosen.