A negotiated environmental settlement in Washington worked exactly as designed. That is the problem.
In 1994, after a fight over a proposed gravel mine and barge terminal near the Nisqually National Wildlife Refuge, the Nisqually Delta Association signed a settlement with Lone Star Northwest. The dock moved a mile and a half farther from the delta. A million dollars went to the City of DuPont to offset project impacts. The city's shoreline was redesignated from urban to conservancy.
The agreement also contained a dispute resolution clause, and this is where the story gets interesting, because the clause worked.
The association invoked it, and negotiated until CalPortland, by then the operator, agreed in 2007 to come to the table and pay the association's legal fees. More than two years of face-to-face negotiation followed, and roughly $100,000 of the association's own attorney costs. A draft supplemental agreement was released on June 24, 2011, and the DuPont City Council approved the final version on January 26, 2012, five votes to one.
Look at who signed. CalPortland. The City of DuPont. The Washington State Department of Ecology. The Nisqually Delta Association, the Washington Environmental Council, People for Puget Sound, three Audubon chapters, the Anderson Island Quality of Life Committee. The agreement established a process for restoring flows to Sequalitchew Creek using CalPortland-provided funds, aimed at permanently protecting the creek and its ravine, and it permitted CalPortland to apply for mining permits in areas already designated for extraction. It stated plainly that it "does not approve or authorize any mining," and that the restoration plan and any mining proposal would each go through normal regulatory review.
By every standard the literature recommends, this is a good outcome. A covenant with a built-in re-examination mechanism. The mechanism used. A renegotiated framework with funding attached, a restoration plan, a wider set of parties, and the state environmental agency as a signatory rather than a bystander.
On June 11, 2026, a hearing examiner for the City of DuPont approved permits to expand the Pioneer Aggregates mine by roughly 313 acres: about 188 acres of surface expansion plus 125 acres of deeper re-mining, extending mine life by about 14 years. Reaching the gravel below the water table requires dewatering, which means wells pumping groundwater away so the deposit can be dug dry. The examiner named dewatering of the Vashon Aquifer the proposal's most significant impact. The environmental impact statement projects groundwater discharge to the Sequalitchew Creek ravine falling by an annual average of as much as 83 percent at peak dewatering, and baseflow running about 79 percent lower permanently, after mining ends.
Nobody defected. The 1994 plaintiff wrote to the hearing examiner in July 2025 supporting the application, so long as it stayed consistent with the 2012 agreement. The Nisqually Indian Tribe, which had challenged the adequacy of the environmental statement, settled and stipulated to dismissal on May 26, 2026, three weeks before the decision.
This is what it looks like when the process works and the substrate is spent anyway.
The creek was already down to one channel
The permit is worse than its own numbers suggest, and to see why you have to know what Sequalitchew Creek currently is.
It is not a healthy creek losing 83 percent of its flow. Its upper watershed was severed decades ago. A diversion at the outlet of Sequalitchew Lake removed roughly 50 cubic feet per second, leaving low summer flows averaging 1.4 cfs, and a rail embankment built around 1915 closed the estuary mouth to an undersized box culvert that still functions as a partial fish passage barrier. The draft EIS describes the consequence without editorializing: above river mile 1.0 and below Edmond Marsh the creek is typically dry, surface connection between the marsh and the lower creek is rare, and the water in the lower mile is groundwater discharged from the Vashon Aquifer as springs in the ravine.
So the aquifer is not one support among several. It is the only one left.
Here the framework vocabulary earns its place. Substrate is the layer other things are drawn on; expressions are what it supports. Mining has its own word for whatever sits between the surface and the commodity: overburden. Overburden is a category rather than a substance, and its defining property is that it is in the way. At DuPont the material in the way is the aquifer, which is simultaneously the creek's water, the marsh's water, the salmon's habitat conditions and the forest's water. A system that reclassifies its own substrate as overburden has inverted its layering, and that inversion is diagnosable in advance. You do not have to wait for the springs to fail. You only have to notice that the permit is written about gravel and the water appears in it as a recovery cost.
The severance history turns this into what the framework would call contracting slack. The first cut, the lake diversion, took the creek's surface supply and left one redundant channel: groundwater. The current proposal spends most of that. There is no third channel behind it.
It also puts the restoration plan in an awkward light. The recovery path identified in the state's own salmon recovery record is to replace the rail culvert and remove the diversion structures so water flows back into the creek. That is a 50 cfs lever. The mine's contribution is a restoration plan it funds, on a creek whose present water it is removing. The two ledgers are never required to be netted against each other, because the 2012 agreement sends the restoration plan and the mining proposals through separate normal reviews. Each is adequate on its own page. Nobody owns the sum.
What the framework traded
The 1994 agreement, at least as its critics quote it, contained a substantive constraint: a commitment to seek no permits to mine in a way that would significantly impact the flow of Sequalitchew Creek or destroy conditions for native salmon. That clause is contested and I have not verified it against the original. But the structural difference between the two agreements does not depend on the wording.
The 1994 instrument, whatever it said, was a prohibition. The 2012 instrument is a process. It establishes review, funding, a restoration plan and a right to apply. It explicitly does not authorize mining, which is the sentence its defenders reach for and which is exactly the point: a process cannot say no on its own. It can only route the question to a forum that will answer it later, under whatever standards that forum applies.
Converting a substantive constraint into a procedural one is the most common way a hard-won environmental settlement dissolves without anyone violating it. The prohibition was a rule about outcomes; the process is a rule about steps. Every step can be honored perfectly and still arrive where the prohibition existed to prevent.
And the dispute resolution clause, the thing that made the 1994 agreement look durable, turns out to be a one-shot. It was invoked once, and spending it produced a successor. Whether that successor carried its own re-examination mechanism is the question nobody was required to ask in 2011, because the mechanism had just worked and the parties were looking at a win. Institutional memory does not only decay by being forgotten. It decays by being spent.
The enforcement layer became a party
The deeper problem is who signed.
Ecology is the state agency that would ordinarily sit outside an agreement like this and test its outcomes against statutory duties. Instead it is a signatory to the framework that permits the application. This is not corruption; agency participation was almost certainly what made the 2012 deal credible enough for the environmental caucus to accept. But it is a role collapse. The referee joined the agreement, and an agreement whose referee is a party has no external check left in it.
Compare what would have held. An obligation that attaches to the resource rather than the parties, so it survives changes of ownership, operator and membership. A funded obligation whose term is set by the duration of the impact rather than the operation, which is what Montana built after the Pegasus Gold bankruptcy left the public holding a perpetual water treatment bill: a legislatively locked trust whose principal cannot be touched without a two-thirds vote of both chambers and whose earnings fund treatment with no end date. And scheduled, mandatory re-examination points, so that "is this still doing what it was written to do" gets asked on a calendar rather than only when someone can afford $100,000 in legal fees to ask it.
Mitigation with an expiry date
The mitigation concept intercepts groundwater and conveys it back toward the creek and marsh through surface channels, with infiltration in ponds on the existing mine floor. It is contingent, adaptive and subject to a feasibility determination during operations. What it is not is permanent, and the draft EIS says so in a sentence that ought to be the most quoted line in this fight: many of the restoration measures "are likely, at best, a temporary action because pumping in perpetuity is not feasible."
That is an honest sentence and a confession of the structure. Extraction is scheduled in decades, the aquifer's response in centuries, mitigation to the end of the profitable window, roughly 14 years. Three clocks, none synchronized, and the shortest is the only one anybody is required to fund.
Washington's statute contains the mismatch in miniature. Chapter 78.44 RCW defines reclamation as reestablishing, "on a perpetual basis," the vegetative cover, soil stability and water conditions appropriate to the approved subsequent use. The performance security backing that obligation, under RCW 78.44.087, is sized to the estimated cost of reclamation work "for the next thirty-six months." The standard says perpetual. The instrument funds three years. Washington already knows how to do better: RCW 78.56.110, the metals mining statute, requires security covering post-closure monitoring and cleanup of problems revealed after closure, held until Ecology releases it in writing. That regime exists and does not apply to sand and gravel.
One further wrinkle. Since 1945, RCW 90.44.040 has declared all natural groundwater in Washington public, belonging to the public and subject to appropriation for beneficial use. Weyerhaeuser owns the land; nobody owns the water. But the mitigation design intercepts groundwater and returns it to the ground on site rather than consuming it, and Washington water law is built around appropriation for beneficial use. Whatever the intent, a design that consumes nothing sits most comfortably outside the frame in which the public's ownership would bite. The doctrine exists; the agency that would apply it signed the framework; and neither fact reaches the hearing examiner, who was applying a municipal critical areas ordinance to a resource he has no jurisdiction over.
What building without this actually requires
The region does need to build. Washington pulled 54.8 million tons of aggregate worth $686 million out of the ground in 2023, its most valuable mineral commodity. Here is the honest ledger, including the part that cuts against the argument I would prefer to make.
Closing the demolition loop is mostly already done, and smaller than it sounds. The intuitive move is that a metro demolishing itself continuously should feed its own aggregate demand rather than open new pits. The numbers do not support it at scale. EPA puts concrete debris recovery at about 83 percent nationally and asphalt at about 93 percent; even total recovery of all US concrete demolition debris caps near 16 percent of aggregate demand, and USGS puts recycled material's actual share of the priced market at about 5 percent. Nor is the specification the barrier: WSDOT already permits recycled concrete aggregate at 100 percent as coarse aggregate for pavement and Class 3000 concrete, and Washington has 153 permitted recycling facilities. It permits zero percent as fine aggregate, which matters, because crushing produces a large fine fraction and sand is the scarcer half of the market.
The barrier is price, and it is thin. A Washington yard this year listed recycled concrete at $12.75 a ton against virgin crushed base at $13.75, a gap consumed by two miles of trucking. RCW 70A.205.700 already requires WSDOT to use 25 percent recycled aggregate, subject to it being "readily available and cost-effective." In 2019 and 2020 WSDOT used 157 tons out of 361,479 eligible, and every contractor that missed cited cost. The Alaskan Way Viaduct demolition reused roughly 103,000 tons, and it worked because demolition and reuse were the same project on the same site. Haul distance zero. That condition does not generalize.
Material substitution is decarbonization wearing a dematerialization costume. Aggregate is 60 to 75 percent of concrete by volume; cement is 7 to 15 percent. LC3, calcined clay blends and geopolymers all operate inside that small fraction, and remove exactly zero tons of sand and gravel from the pit. LC3 arguably adds quarrying, since it substitutes limestone and clay into the binder. Hempcrete's barrier is physics: it was never structural. Mass timber is the real one and it is partial, because foundations, below-grade work, podium transfer slabs and the acoustic topping over every residential floor plate stay concrete, and in high-seismic Puget Sound the lateral system is still overwhelmingly a concrete core.
Demand reduction is the largest genuine lever and has the ugliest economics. A building you do not demolish requires no new aggregate at all. Seattle removed the zoning barriers to office conversion in July 2024 and added a tax deferral in February 2025, and has four projects, roughly 400 units. The barrier is price: city analysis put typical conversions above $400,000 per unit, about three times new construction. No zoning exemption closes a 3x gap.
Institutional design is where the unoccupied space actually is. Every extraction levy that exists, including the UK's £2.16 per tonne, is a flat tonnage charge wearing an environmental label; Sweden's gravel tax is the only one whose stated rationale is aquifer protection, and it too is flat-rate. Nobody has built a levy priced to the substrate a specific pit consumes. England requires every mineral planning authority to put recycled, secondary, marine-won and land-won supply on one annual balance sheet, so demonstrated recycled supply reduces the reserve the authority must permit. Washington's mineral resource lands rule, WAC 365-190-070, is a pure geology exercise that does not mention recycled supply at all.
And the haul-distance premise is weaker here than in the general case. Roughly 80 percent of this mine's output leaves by barge. Aggregate economics collapse at about 25 miles by truck and stretch to hundreds of miles by water. If the product ships by water anyway, what the deposit needs is proximity to tidewater, not to the metro.
The fork
The environmental impact statement analyzed exactly two options: no action, and the proposal. A no-dewatering alternative, using a groundwater interceptor channel instead of extraction wells, was raised during scoping in 2021 and listed in the city's scoping summary as warranting further evaluation. It never appeared in the draft or final statement, and no explanation is given. The question "can we take the gravel without spending the aquifer" was asked, written down, and dropped out of the document. Not refused. Dropped.
CalPortland is not the villain here. The company is doing what its incentive geometry directs, on a deposit its predecessors have worked since the 1890s, under a framework the environmental parties negotiated and the state signed. Set malice aside and the point gets stronger: this outcome is produced by the arrangement of the institutions, and any competent firm in the same position would produce it again.
What the region could have instead is not a moratorium. It is constraints that are structural rather than procedural. A substrate floor, meaning extraction above the water table treated as a line rather than a variable. Bonds sized to the duration of the impact rather than the operation. Restoration credits and extraction debits required to appear on the same page before either is approved. Reserve planning that counts the demolition stream and the barge radius alongside the geological deposit, so that "we need this pit" becomes a claim somebody has to prove. Water-right review in the same room as land-use review, before an agency that is not also a signatory.
The Clover Creek Restoration Alliance filed an appeal on June 26, 2026. Whatever it produces, the lesson of 1994 is already available: winning a negotiated framework is not the same as winning a constraint, and the mechanism that gets you the framework is usually good for exactly one use.
Figures are drawn from the City of DuPont's June 2025 staff report, the project's draft EIS as quoted in the public record, the June 11, 2026 hearing examiner decision as reported, Washington's Recreation and Conservation Office salmon recovery project record for the Sequalitchew Creek estuary, and state and federal statutes and agency data cited inline. The 1994 settlement clause referenced above has not yet been verified against the original document. Can anyone find a unredacted copy?
