Real problems worth solving

At the International Conference on Learning Representations (ICLR) 2026, an analysis by Pangram Labs of all 75,800 peer reviews found that 15,899 reviews (21%) were fully AI-generated. More than half of all peer reviews showed signs of AI involvement. The discovery only happened because dozens of academics noticed red flags -- hallucinated citations, unusually verbose feedback, non-standard requests -- and raised alarms on social media. Carnegie Mellon professor Graham Neubig offered a bounty for analysis, and Pangram's team parsed every review within 12 hours. The conference's own systems caught none of it. The consequence is not just embarrassment for the AI research community. Each AI-generated review represents a decision point: a paper accepted or rejected based on feedback that no human actually read or evaluated. Researchers who spent months on a paper had their careers shaped by a language model's output. Junior researchers denied acceptance may lose conference presentation slots that are critical for job market visibility, tenure cases, and grant competitiveness. Senior researchers whose mediocre work was rubber-stamped by an AI reviewer gain unearned credibility. The peer review system's entire legitimacy rests on the assumption that expert humans are evaluating work -- when that assumption fails at 21%, the signal-to-noise ratio of the entire venue collapses. The structural cause is a volume crisis. ICLR submissions nearly tripled in three years: 7,304 papers in 2024, 11,672 in 2025, and 19,814 in 2026. Each reviewer was assigned multiple papers on tight deadlines with no compensation. AI-generated reviews are a rational individual response to an irrational collective workload. Conferences have no reliable detection mechanism, no penalties for AI-assisted reviews, and no way to reduce submission volume without gatekeeping that contradicts open science principles. The reviewer pool cannot scale with submissions because qualified reviewers are a finite resource, but submission volume is effectively unbounded.

At PLoS ONE, the world's largest megajournal, a 2025 analysis published in PNAS found that just 45 editors -- representing 0.25% of the journal's editorial pool -- accepted papers that account for more than 30% of the journal's 702 retractions issued by early 2024. These 45 individuals handled only 1.3% of all articles published between 2006 and 2023, yet the papers they shepherded through peer review were retracted at wildly disproportionate rates. Twenty-five of these editors also authored papers in PLoS ONE that were themselves later retracted. This matters because those retracted papers do not simply vanish. They are cited by hundreds of subsequent studies, forming the foundation for clinical decisions, grant applications, and entire research programs. When a paper on cancer biomarkers or drug efficacy is retracted, every downstream study that relied on its findings is weakened -- but almost none of those citing papers are flagged or corrected. Clinicians who read systematic reviews built on this contaminated evidence base may make treatment decisions informed by fabricated data. Patients are the final victims of a chain that starts with a compromised editor clicking 'accept.' This problem persists because megajournals like PLoS ONE operate at a scale where centralized editorial oversight is structurally impossible. The journal publishes tens of thousands of papers per year using thousands of volunteer academic editors, each acting with near-total autonomy over their assigned submissions. There is no systematic audit of editorial acceptance patterns, no real-time dashboard flagging editors whose accepted papers are being retracted at abnormal rates, and no financial incentive for the publisher to build one -- since each published paper generates revenue via APCs regardless of whether it is later retracted. The editorial model treats each submission as an independent event, when in reality, a small number of compromised or negligent editors can poison the entire corpus.

Formosan subterranean termites (Coptotermes formosanus) cause an estimated $300 million per year in property damage in New Orleans alone. Nationally, the damage from this single invasive species runs to $1 billion annually. Several homes in New Orleans have been demolished because repeated treatment attempts failed and the structural damage was irreparable. The core issue is that when chlordane was banned in 1988, the termite control industry lost a product that provided 25-30+ years of soil barrier protection. Modern replacements -- fipronil (Termidor), imidacloprid (Premise), and various pyrethroids -- provide 5-10 years of protection at best. Homeowners who assumed their termite treatment was permanent now face recurring $1,500-3,000 re-treatment costs every 5-8 years, plus the risk of undetected damage between treatments. Formosan termites are categorically more destructive than native subterranean termites. A single Formosan colony can contain 5-10 million individuals (versus 300,000 for native species), consume up to 1,000 pounds of wood per year, and forage across a territory of 300+ feet. They build above-ground carton nests that retain moisture, allowing them to infest structures without maintaining a connection to the soil -- which means soil barrier treatments can be bypassed entirely. In New Orleans and coastal Louisiana, the termites infest live oak trees, utility poles, railroad ties, and historic buildings. The French Quarter -- with its 18th and 19th century wood-frame construction -- has been particularly devastated. The federal government and Louisiana state agencies ran 'Operation Full Stop,' an area-wide management program, from the 1990s through roughly 2010, which reduced infestation rates by 50%. But when the program's funding ended, populations began recovering. This problem persists because of a fundamental gap between the durability of older organochlorine termiticides and their modern replacements. Chlordane worked for decades because it was extraordinarily persistent in soil -- the same property that led to its ban due to environmental contamination and cancer risk. Current termiticides are designed to degrade faster, which is better for the environment but worse for long-term termite exclusion. Bait systems (like Sentricon and Trelona) offer an alternative approach by eliminating colonies rather than creating soil barriers, but they require continuous monitoring and bait replenishment, adding ongoing annual costs of $300-500. Homeowners, especially in Louisiana's lower-income parishes, cannot afford either the recurring chemical treatments or the annual monitoring contracts. Insurance companies typically exclude termite damage from homeowner policies, so the full financial burden falls on the property owner. The Formosan termite continues to expand its range northward as winters warm, now established as far north as Tennessee and North Carolina.

In December 2024, the EPA proposed a rule banning most food uses of chlorpyrifos, an organophosphate insecticide that was for decades the most widely used insecticide in U.S. agriculture. But the ban includes exemptions for 11 specific crops: alfalfa, apple, asparagus, tart cherry, citrus, cotton, peach, soybean, strawberry, sugar beet, and spring/winter wheat. As of July 1, 2025, growers of every other crop -- including most vegetables, tree nuts, and specialty crops -- lost access to chlorpyrifos with no guaranteed equivalent replacement. The United Nations' Stockholm Convention subsequently listed chlorpyrifos for global elimination under Annex A, further cementing the phase-out. The two-tier system creates perverse outcomes. A peach grower can still use chlorpyrifos to control Oriental fruit moth, but a nectarine grower next door cannot, even though the crops are botanically almost identical and face the same pests. Vegetable growers who relied on chlorpyrifos for soil-dwelling pests like wireworms and root maggots must now use alternatives that are often less effective, more expensive, or have narrower pest spectra. For some pest-crop combinations, there is simply no registered alternative with comparable efficacy. University of Florida researchers have explicitly advised producers on cancelled crops that 'finding alternatives is advisable,' acknowledging that good alternatives may not exist. The transition cost -- trialing new products, adjusting application equipment, potentially accepting higher pest damage -- falls entirely on the farmer, with no government assistance for the transition. This problem persists because of how pesticide regulation handles risk-benefit analysis. The 11 exempted crops were those where EPA determined the benefits of chlorpyrifos use outweighed the health risks, considering the availability of alternatives and the economic importance of the crop. But this crop-by-crop calculus ignores the reality that pests do not respect crop boundaries. A root maggot population that thrives in an untreated vegetable field migrates to adjacent fields. More fundamentally, the exemption system was a political compromise between agricultural interests that wanted continued access and health advocates who documented chlorpyrifos's neurodevelopmental toxicity in children. The result satisfies neither side: farmers growing non-exempt crops lose a critical tool, while health advocates point out that any continued use exposes farmworkers and communities. Meanwhile, the exemption creates a competitive distortion where cotton and soybean growers retain a cost-effective pest control option that vegetable growers -- often smaller, more diversified operations -- do not.

The spotted lanternfly (Lycorma delicatula), first detected in Pennsylvania in 2014, has spread to 19 states and the District of Columbia despite quarantine orders in every affected state. The quarantine system requires businesses operating in infested zones to obtain permits, inspect vehicles before leaving quarantine areas, and certify that shipped goods are lanternfly-free. In practice, this system is collapsing under its own weight. Trucking companies must train drivers to inspect trailers, obtain permits in each quarantined state, and document compliance -- all for a pest that lays inconspicuous egg masses on any outdoor surface including trailer undersides, rail cars, stone shipments, and nursery stock. Virginia repealed its quarantine entirely in March 2025, effectively acknowledging that enforcement was futile. The economic damage is concentrated and severe. Cornell University researchers estimated that New York's grape industry alone could face losses of $1.5 million in the first year of infestation, $4 million in the second, and $8.8 million in the third. In Pennsylvania, where the infestation is most established, damage is estimated at $50.1 million per year with 484 jobs lost, with worst-case projections of $92.8 million annually. Nationally, the spotted lanternfly threatens $915 million in grape and tree fruit industries and $2.6 billion in ornamentals. The insect feeds on over 70 plant species, excreting copious honeydew that promotes sooty mold growth on everything beneath -- cars, patios, outdoor furniture, playground equipment -- creating a quality-of-life nightmare for homeowners in infested areas. This problem persists because quarantine-based containment is structurally incompatible with modern logistics. The U.S. freight system moves 50+ million truckloads per year. Inspecting even a fraction of vehicles leaving quarantine zones is impossible with current staffing. The spotted lanternfly's primary host plant, the tree of heaven (Ailanthus altissima), is itself an invasive species that grows ubiquitously along highways, railroads, and disturbed land -- exactly the corridors used for freight. Killing the tree of heaven could reduce lanternfly populations, but Ailanthus is so widespread (estimated in every U.S. state) that eradication is impractical. Chemical control of the lanternfly itself requires repeated applications of systemic insecticides like dinotefuran, which kills pollinators and beneficial insects. Biological control research is underway -- two parasitoid wasps from the lanternfly's native range in China are being evaluated -- but host-range testing and regulatory approval for releasing non-native biocontrol agents takes 5-10 years. In the meantime, the lanternfly advances 40-50 miles per year along highway corridors.

Forty-three U.S. states have enacted pesticide preemption laws that strip cities and counties of the authority to regulate pesticide use within their jurisdictions. These laws prevent local governments from enacting any restrictions on pesticide 'use, sale, notification, marketing, or distribution' that go beyond state or federal standards. Only seven states -- Alaska, Hawaii, Maine, Maryland, Nevada, Utah, and Vermont -- allow local communities to set their own pesticide rules. In 2025, legislators in Florida, Mississippi, Missouri, Oklahoma, and Wyoming introduced new bills to further limit pesticide liability by making EPA registration a legal shield against failure-to-warn cancer claims. This matters because pesticide exposure risks are intensely local. A school adjacent to a golf course, a playground next to a commercial lawn care operation, a residential neighborhood abutting a farm field -- these are hyperlocal exposure scenarios that only local governments can address with the specificity needed. Federal EPA rules set broad tolerances for agricultural use across the entire nation. State rules rarely distinguish between a rural county and a dense urban neighborhood. When a city council in a preempted state tries to ban cosmetic lawn pesticides near schools or require notification before spraying, the pesticide industry sues, citing preemption, and wins. The result is that communities with documented pesticide exposure problems -- elevated childhood cancer clusters near treated fields, pollinator die-offs in suburban areas, contaminated well water -- have zero regulatory recourse at the local level. They must lobby their state legislature, where the pesticide industry's lobbying budget dwarfs any community group's resources. This problem persists because pesticide preemption laws were the product of a deliberate industry campaign. In 1991, after the U.S. Supreme Court ruled in Wisconsin Public Intervenor v. Mortier that local communities had the right to enact pesticide restrictions stricter than federal rules, the pesticide industry formed the 'Coalition for Sensible Pesticide Policy' and systematically lobbied state legislatures to pass preemption statutes. Within a decade, most states had enacted them. These laws are now deeply entrenched -- repealing a preemption law requires the same state legislature that passed it to reverse course, against ongoing industry lobbying. The 2024 Farm Bill negotiations included a provision for federal-level preemption that would have overridden even the seven remaining non-preempted states, though it did not pass. The structural incentive is clear: a patchwork of local restrictions increases compliance costs for pesticide manufacturers and applicators, so industry has a permanent financial interest in maintaining preemption.

The EPA's Office of Pesticide Programs has over 12,000 pesticide registration reviews that are overdue relative to statutory timelines, plus 504 new active ingredient applications awaiting initial review. More than $500 million worth of crop protection products developed by agricultural technology companies are stuck in this backlog, unable to reach farmers. When a new insecticide, fungicide, or herbicide is developed, it must go through EPA registration before it can be sold. The average timeline for a new active ingredient registration is 2-3 years, but actual wait times frequently stretch to 5+ years due to the backlog. For specialty crops and minor uses, the delay can be even longer because these applications generate less revenue and receive less industry follow-up. This delay has direct consequences for farmers fighting resistance. When a pest population evolves resistance to existing chemicals -- as diamondback moth, Palmer amaranth, and bed bugs have done -- growers desperately need new modes of action. But even when a company has developed a promising new chemistry, it sits in EPA's queue for years while the pest resistance crisis worsens. A delay of 12-18 months in registration can mean missing two or more growing seasons, during which farmers lose crops and resistance spreads further. The delay also disadvantages U.S. farmers relative to international competitors, since the same product may be registered and in use in Brazil, Australia, or the EU while U.S. growers cannot access it. For biopesticide companies -- often small startups developing microbial or RNA-based solutions -- the registration delay can be existential, burning through venture capital while generating zero revenue. This problem persists because Congress has chronically underfunded the Office of Pesticide Programs. The office requested $166 million in appropriations but has received significantly less, leading to insufficient staff to process applications. The Pesticide Registration Improvement Act (PRIA) was meant to address this by allowing EPA to collect fees from registrants, but the fees have not kept pace with the growing workload. PRIA's fee structure also creates perverse incentives: it sets decision timelines that EPA routinely misses without consequence, and it charges the same fees regardless of application complexity. Since January 2025, the EPA has reduced the backlog by over 5,000 actions, but this was achieved partly by streamlining reviews in ways that environmental groups argue compromise safety assessment. The fundamental tension is that thorough safety review takes time and expertise, but the current system has neither enough reviewers nor a triage mechanism to prioritize the most urgently needed products.

Texas is home to an estimated 2.6 million feral hogs -- nearly half the U.S. population -- and their numbers increased 39% over the past three years despite aggressive control efforts. The state spends $474 million annually on hog control, requiring over 17 million labor hours. Across 13 surveyed states, feral hogs caused $1.6 billion in agricultural losses in a single year, with Texas accounting for $871 million of that total. The damage includes $193 million in pastureland destruction, $103 million in crop-related property damage, $375 million in livestock infrastructure damage, and $122 million in lost income from farmers changing what they plant to avoid hog damage. The math of feral hog reproduction makes control feel hopeless. A sow can breed at 6 months old, has two to three litters per year with up to 12 piglets per litter, and wildlife biologists estimate that 70-75% of the entire population must be removed annually just to prevent growth. No current method approaches that threshold at scale. Recreational hunting removes a trivial fraction. Aerial gunning from helicopters is effective but costs $2-5 per acre and is impractical over the 170+ million acres of Texas farmland. Trapping works on individual sounders but has a critical flaw: if any animals escape, the survivors become 'trap-shy' and teach avoidance to others, making future trapping in that area far less effective. The only toxicant ever approved -- Kaput, a warfarin-based bait -- was blocked by a court injunction in 2017 after the feral hog meat processing industry and environmental groups both opposed it, though Texas and Oklahoma re-approved it in 2024 with restrictions. This problem persists because of a structural governance failure. Feral hogs are classified as an invasive species by wildlife agencies but are treated as a hunting resource by rural culture and the outfitting industry. Helicopter hog hunts generate tourism revenue. Feral hog processing businesses earn income from the meat. These economic interests actively oppose the most effective control methods -- widespread poisoning or disease-based biocontrol -- because they would eliminate the 'resource.' Meanwhile, the hogs are present in 73% of Texas farms and expanding into new states (they now occupy 35+ states). No single landowner can solve the problem because hogs range across property boundaries, and coordinated landscape-scale management requires cooperation among hundreds of private landowners, which rarely materializes. Federal funding through USDA Wildlife Services helps but covers only a fraction of the need.

Urban bed bug populations (Cimex lectularius) have developed resistance to pyrethroids, neonicotinoids, organophosphates, carbamates, and phenylpyrazoles -- essentially every chemical class available to pest control operators. Laboratory testing of field-collected strains has found pyrethroid resistance ratios exceeding 12,765x, meaning it takes nearly 13,000 times the normal dose of deltamethrin to kill resistant individuals. This is not a marginal decrease in efficacy; it is complete product failure. A 2025 global review in Entomological Research documented that resistance has been confirmed on every inhabited continent, with multi-mechanism resistance (target-site kdr mutations, metabolic detoxification via P450 enzymes, and penetration resistance through cuticle thickening) often stacked in the same population. The human cost falls disproportionately on low-income renters who can least afford it. A professional bed bug treatment costs $1,500-$6,000 depending on severity and method. Heat treatment -- the only reliable option against resistant strains -- requires specialized equipment that heats an entire unit to 120-140F for several hours and costs $3,000-$6,000 for a whole apartment. Low-income tenants in multi-unit housing cannot afford this, their landlords often refuse to pay, and the bugs reinfest from neighboring units within weeks anyway. People lose sleep, develop anxiety and PTSD-like symptoms, throw away furniture they cannot afford to replace, and face social stigma. Children in infested homes show measurable declines in school performance. Meanwhile, pest control operators who rely on chemical-only treatments face callback rates of 50%+ because the products simply do not kill resistant populations, destroying their margins and reputation. This problem persists because bed bugs have a unique biological advantage: they feed only on blood, so they encounter insecticides only through residual contact, which imposes strong selection for cuticle-thickening and metabolic resistance. Their flat body shape lets them hide in cracks as thin as a credit card, avoiding direct spray contact. They reproduce quickly (200-500 eggs per female lifetime) and have short generation times (6-8 weeks), so resistance alleles fix rapidly in populations. The insecticide industry has little financial incentive to develop new bed bug chemistries because the market is small compared to agriculture. Heat treatment works but requires $50,000+ in equipment investment per truck, and many pest control companies -- especially small operators serving low-income housing -- cannot afford the capital outlay. There is no coordinated public health response in most U.S. cities because bed bugs are classified as a nuisance pest, not a public health threat, so health departments have no mandate or funding to intervene.

Dicamba, a volatile herbicide used on genetically engineered soybeans and cotton, has been caught in a legal and regulatory whipsaw that makes it impossible for farmers to plan their weed management programs. Federal courts vacated EPA's registration in 2020. The EPA re-registered it in 2021. A federal court in Arizona vacated it again in February 2024. The EPA proposed re-approving it in July 2025 with loosened restrictions, eliminating the previous June cutoff date and dropping a 100-foot endangered species buffer. Environmental groups sued again in February 2026. Farmers planting dicamba-tolerant seeds have no certainty their primary herbicide will be legal by the time they need to spray. This regulatory chaos creates concrete financial damage. Dicamba-tolerant seed systems (Roundup Ready 2 Xtend, XtendFlex) cost farmers a technology fee of $10-15 per acre. If a court bans dicamba mid-season, farmers who already planted these seeds lose both the technology fee and their primary weed control tool, forcing emergency purchases of alternative herbicides that may not control resistant weeds. But the damage extends beyond the farmer who chose dicamba. Dicamba volatilizes in hot weather and drifts onto neighboring fields, damaging non-tolerant soybeans, orchards, vegetable gardens, and trees. The USDA estimated 15 million acres of soybeans were damaged by dicamba drift in 2018 alone. This has triggered thousands of complaints, neighbor-vs-neighbor lawsuits, and a Bayer settlement of $400 million. The social fabric of rural farming communities has been torn apart by a chemical that one neighbor uses and another's crops are destroyed by. This problem persists because of a fundamental conflict between EPA's registration authority and judicial review. The EPA registers pesticides under FIFRA based on risk-benefit analysis, but courts have repeatedly found that EPA failed to adequately assess drift damage and endangered species impacts. Each time EPA re-approves dicamba, environmental groups sue, and each time courts agree that EPA's analysis was inadequate. Meanwhile, farmers have already purchased seeds and planned their season around the product. The agrochemical companies (Bayer, BASF, Syngenta) continue to sell dicamba-tolerant seed systems because each re-approval, however temporary, generates billions in revenue. There is no mechanism in FIFRA for conditional or provisional registration that could give farmers certainty while the legal issues are resolved.

Palmer amaranth (Amaranthus palmeri) has evolved resistance to nine different classes of herbicides, including glyphosate, ALS inhibitors, PPO inhibitors, HPPD inhibitors, and most recently glufosinate. A 2025 study in The Plant Cell revealed the mechanism: Palmer amaranth carries dozens to hundreds of extrachromosomal circular DNA (eccDNA) copies of the EPSPS gene, and recent research shows these eccDNA molecules can rearrange to incorporate multiple resistance genes on a single element. This means a single plant can carry resistance to two or more herbicide classes simultaneously, and it passes this resistance to offspring at near-100% frequency. The practical impact on farmers is devastating. In Georgia, before glyphosate resistance emerged, Palmer amaranth management cost about $23 per acre. By 2010, 50-70% of Georgia's cotton acres required hand-weeding at $22-24 per acre on top of herbicide costs, bringing total weed management to $60-80 per acre -- a 475% increase. The weed grows 2-3 inches per day and can reach 6-8 feet tall. A density of just two plants per foot of row can reduce soybean yield by 79% and cotton yield by similar margins. Some Georgia and Arkansas farmers have abandoned fields entirely rather than fight it. University of Georgia researchers have explicitly stated that resistant Palmer amaranth "can run you out of the cotton business." This problem persists because Palmer amaranth evolves resistance faster than the agrochemical industry can develop new herbicides. The eccDNA mechanism is unique in the plant kingdom and allows resistance genes to amplify to hundreds of copies per cell, creating an essentially insurmountable dose-response barrier. Every new mode of action that Palmer amaranth encounters gets defeated more quickly than the last. Meanwhile, the herbicide development pipeline takes 10-12 years from discovery to market, and no genuinely novel mode of action for broadleaf weed control has been commercialized in decades. The weed is dioecious (separate male and female plants), which forces outcrossing and maintains genetic diversity, accelerating the spread of resistance alleles through populations. Hand-weeding is the last resort, but it is unsustainable at scale because agricultural labor is scarce, expensive, and hand-pulled Palmer amaranth can regrow from stem fragments left in the field.

Huanglongbing (HLB), commonly called citrus greening, is a bacterial disease spread by the Asian citrus psyllid (Diaphorina citri) that has destroyed Florida's citrus industry. Production has plummeted from 244 million boxes in 1998 to a projected 12 million boxes for 2024-2025 -- a 94% decline. Over 50 million trees have been killed. Florida citrus acreage has shrunk from 748,555 acres in 2004 to 274,705 in 2024, with groves abandoned or converted to housing developments and cattle pasture. The industry has lost $20 billion in revenue and 33,000 jobs. The core horror is that there is no cure for an infected tree. Once a psyllid feeds on an infected tree and then moves to a healthy one, the bacterium (Candidatus Liberibacter asiaticus) colonizes the phloem, blocking nutrient transport. The tree produces small, bitter, lopsided fruit that drops prematurely. Within 3-5 years the tree dies. A grower who has spent $10,000-15,000 per acre establishing a grove watches it become worthless with no remedy. Production costs have more than doubled because growers must spray insecticides intensively to suppress psyllid populations, apply foliar nutrition to keep sick trees alive longer, and replant at enormous expense -- only to see new trees get infected within 1-2 years. The economics have become so punishing that many multigenerational citrus families have simply quit. This problem persists because the Asian citrus psyllid is extraordinarily difficult to eliminate. It is tiny (3-4mm), highly mobile, and reproduces rapidly. Chemical control reduces populations but cannot achieve the near-total suppression needed to stop disease transmission -- a single infected psyllid can inoculate a tree. Biological control agents like Tamarixia radiata parasitoid wasps help but provide only partial suppression. Physical exclusion with mesh covers (CUPS) works for individual young trees but is impractical at commercial scale. The bacterium itself is unculturable in the lab, which has severely hampered research into antibiotics or resistant rootstocks. Despite two decades of intensive research and hundreds of millions in funding, no breakthrough treatment has emerged. Meanwhile, the psyllid has spread to California, Texas, and other citrus-producing states, threatening to replicate Florida's catastrophe.

The diamondback moth (Plutella xylostella) has developed documented resistance to 104 unique insecticide active ingredients across 1,099 recorded cases, making it the world's number-one insecticide-resistant pest. In the fall 2024 and spring 2025 growing seasons, growers across Arizona and California reported outright control failures against diamondback moth in broccoli, cabbage, and cauliflower fields. The moths now shrug off chlorantraniliprole (Coragen), cyantraniliprole (Verimark), emamectin benzoate (Proclaim), spinetoram (Radiant), and even Bt aizawai (XenTari) -- products that represented the last reliable lines of defense. This matters because brassica crops -- broccoli, cabbage, kale, cauliflower -- are a $3+ billion U.S. market, and diamondback moth is their primary insect pest worldwide. When control fails, growers don't just lose yield; they lose entire harvests. A single uncontrolled generation can destroy a field in days because the larvae bore into the growing points and heads, making the crop unmarketable. Growers respond by spraying more frequently and at higher doses, which accelerates resistance further, poisons beneficial insects that would otherwise provide natural control, and increases production costs by hundreds of dollars per acre. The global cost of managing diamondback moth is estimated at $4-5 billion per year, much of it wasted on chemicals that no longer work. This problem persists because of a structural mismatch between how insecticides are used and how resistance evolves. Growers plant brassicas year-round in warm climates like Arizona, California, and the Southeast, giving the moth continuous generations with no break in selection pressure. Each generation is roughly 3-4 weeks, meaning a single growing season can impose 8-10 rounds of selection on the same population. The moth also migrates long distances, so resistant genes spread rapidly across regions. Meanwhile, the pipeline for new insecticide modes of action has slowed to a trickle -- no genuinely new class has been commercialized for Lepidoptera in over a decade. Mating disruption and biological controls exist but are not widely adopted because they require more management expertise and don't offer the quick knockdown growers need when a crop is under immediate attack. The result is a treadmill: each new chemistry buys a few years before resistance catches up, and the interval keeps shrinking.

In 2024, the Consumer Financial Protection Bureau finalized a rule that would have prohibited medical debt from appearing on consumer credit reports, recognizing that medical debt is fundamentally different from other debt — it is rarely incurred by choice, pricing is opaque, and billing errors are rampant. But in July 2025, a federal court vacated the rule. The result: an estimated $194 billion in medical debt remains in active collection, 36% of US households carry medical debt, and 58% of all accounts in collections are for medical bills. Medical debt is the single largest category of debt in collections in the United States. This matters because medical debt on a credit report does not just represent money owed — it blocks access to housing, transportation, and economic opportunity. A patient with a $2,000 medical collection on their credit report may be denied an apartment lease, pay higher auto insurance rates, face a higher interest rate on a car loan, or be rejected for a small business loan. Studies show that over 60% of personal bankruptcies are tied to medical debt. The cruelty is that medical debt is often the result of billing errors (up to 80% of bills contain mistakes), surprise bills the patient could not have avoided, or insurance claim denials that were later overturned — but the credit damage occurs immediately and persists for years even after the underlying billing dispute is resolved. This problem persists because credit reporting agencies (Equifax, Experian, TransUnion) profit from comprehensive data collection — more data means more products to sell to lenders. Medical providers benefit from the coercive power of credit reporting as a collections tool: the threat of credit damage motivates patients to pay bills they might otherwise dispute. The court that vacated the CFPB rule held that the agency exceeded its statutory authority, and Congressional action to achieve the same result through legislation has stalled. Eleven states have enacted their own laws restricting medical debt credit reporting, creating a patchwork where a patient in Colorado is protected but a patient in Texas is not — for the same bill, the same debt, the same financial impact.

There is no federal law requiring healthcare providers to send a bill to patients within any specific timeframe. Timely billing requirements vary wildly by state and payer. Medicare requires claims to be filed within 12 months. Medicaid deadlines range from 90 days (Texas) to 365 days (Ohio). Commercial insurers set their own deadlines, ranging from 90 days (Cigna, UnitedHealthcare) to one year (some BCBS plans). But these are deadlines for the provider to bill the insurer — not the patient. A provider can bill the insurer within the filing deadline, have the claim denied or partially paid, and then send the balance to the patient months or years later. This matters because patients cannot budget for healthcare costs when bills arrive unpredictably, long after the service. A patient has surgery in January, receives and pays the hospital bill in March, believes the episode is financially closed, and then receives a separate bill from the anesthesiologist in September and a lab bill in November. Each bill restarts the patient's financial stress cycle. Late-arriving bills are also harder to dispute because the patient's memory of what happened has faded, medical records may need to be re-requested, and the patient may have already hit (or not yet hit) a new year's deductible, changing their cost-sharing calculation. For patients on fixed incomes or tight budgets, an unexpected $800 bill arriving 8 months after a procedure can trigger a cascade of financial harm. This problem persists because providers and billing companies benefit from flexibility. Complex cases involve multiple providers, each billing separately, and insurance adjudication can take weeks or months. Providers argue that timely billing mandates would force them to absorb costs when insurance processing is slow. Only a handful of states — notably Texas, which requires providers to send bills within 10 months — have enacted patient-facing timely billing laws with actual penalties for non-compliance. The absence of a federal standard means patients in most states have no legal protection against stale bills, and the medical billing industry has successfully lobbied against uniform deadlines by arguing that the complexity of multi-payer adjudication makes fixed deadlines impractical.

A study of hospitals in Dallas County found that chargemaster prices for identical items varied by staggering amounts: a 5-mg amlodipine tablet (a common blood pressure medication) ranged from $0.23 to $43.05 — an 18,617% difference. A partial thromboplastin time lab test ranged from $18.70 to $506.00 (2,606% difference). A circumcision ranged from $252.00 to $7,532.10 (2,889% difference). Nationally, 90th-to-10th-percentile chargemaster price ratios range from 3.2x to 11.5x for the same procedure. This matters because chargemaster prices are the starting point for what uninsured and out-of-network patients are billed. While insured patients with in-network coverage pay negotiated rates that bear little relation to the chargemaster, the approximately 27 million uninsured Americans and anyone who receives out-of-network care may be billed at or near the chargemaster rate. Even for insured patients, chargemaster prices affect out-of-pocket costs when a service is applied to a deductible or when cost-sharing is calculated as a percentage of the 'allowed amount,' which is itself derived from the chargemaster through opaque discount formulas. A patient needing a routine lab test has no practical way to learn, before the blood draw, that the same test costs $18 at Hospital A and $506 at Hospital B three miles away. This problem persists because hospitals set chargemaster prices unilaterally with no external review, no cost-justification requirement, and no legal formula. The chargemaster originated as an internal accounting tool and was never designed to be a consumer-facing price list. Hospitals have strong incentives to keep chargemaster prices high because they serve as the ceiling in negotiations with insurers — the higher the chargemaster price, the larger the 'discount' the hospital can offer while still achieving its target revenue. The price transparency rule was supposed to address this, but as noted, 65% of hospitals remain non-compliant. Even the compliant hospitals publish data in machine-readable formats that require technical expertise to access and interpret, making the data practically invisible to the average patient standing in an ER or scheduling a procedure.

Studies consistently find that 49-80% of medical bills contain at least one error, and the average hospital bill over $10,000 includes approximately $1,300 in overcharges. Yet fewer than half of patients who suspect a billing error actually contact their provider or insurer to dispute it. Of those who do, 74% get the error corrected — proving the errors are real and fixable, if only the patient can identify them. This matters because the primary document patients receive — the Explanation of Benefits (EOB) — is designed for payer-provider communication, not patient comprehension. It lists services as five-digit CPT codes (like '99214' or '73721') with minimal or no description of what the service actually was. A patient who had a knee examined cannot tell whether code 73721 (MRI without contrast) is correct or whether they were billed for code 73723 (MRI with and without contrast, which costs $400-$800 more). The EOB shows 'allowed amount,' 'provider charge,' 'plan payment,' and 'your responsibility' — but does not explain why the provider charged $3,200 when the allowed amount is $800, or what the patient can do about it. Patients receive multiple EOBs for a single visit — one for the doctor, one for the facility, one for the lab, one for the anesthesiologist — and are expected to reconcile these against provider bills that arrive on different dates with different account numbers. This problem persists because there is no regulatory requirement that EOBs include plain-language service descriptions that a patient can verify against their actual experience. CMS and state insurance departments regulate EOB content, but the regulations focus on financial accuracy (did the math add up?) rather than patient comprehensibility (can the patient tell if they were billed for the right thing?). Insurers have no financial incentive to make EOBs more transparent because confused patients who do not dispute bills save the insurer the cost of reprocessing claims. Providers have no incentive because clearer bills would increase dispute rates. The patient is the only party who benefits from comprehensibility, and the patient has no power to change the format.

The No Surprises Act's Independent Dispute Resolution (IDR) process was designed to settle payment disputes between out-of-network providers and insurers without involving patients. From mid-2022 to May 2025, 3.3 million disputes were filed. As of May 2025, 500,000 remained unresolved. In the first half of 2025, providers and insurers submitted nearly 1.2 million new cases — a 40% increase over the prior six months. The total cost of the IDR process has exceeded $5 billion. This matters because the system was supposed to protect patients by taking them out of the middle of billing disputes. Instead, it has created a parallel economy. Providers win 88% of IDR disputes, and the median payment determination is 459% above the qualifying payment amount (the insurer's calculated median in-network rate). This means the arbitration system is systematically awarding payments far above what insurers would have paid in-network, which raises overall healthcare costs that are ultimately passed back to patients through higher premiums. A few provider groups have filed tens of thousands of disputes each, treating IDR as a revenue optimization strategy rather than a last-resort dispute mechanism. Twenty percent of disputes submitted in early 2025 were ineligible — filed late, missing information, or otherwise invalid — further clogging the system. This problem persists because the IDR process was designed with a 'baseball-style' arbitration model where the arbiter picks one side's number, but Congress prohibited arbiters from considering Medicare rates as a primary factor, which tilts outcomes toward provider-submitted charges. The $50 per-dispute administrative fee (raised from the initial $50 but still nominal relative to the amounts at stake) makes it cheap to file speculative disputes. There is no penalty for filing ineligible disputes, no penalty for losing, and the per-case cost to the federal government of processing each dispute far exceeds the administrative fees collected. The system incentivizes volume over merit, and the backlog means legitimate disputes take months to resolve while the costs are socialized across all insurance purchasers.

Federal law has required hospitals to publish machine-readable files of their negotiated prices with every insurer since January 2021. As of June 2025, CMS had conducted over 5,149 compliance reviews and found that 65% of reviewed hospitals had received at least one warning notice or corrective action plan request. Only 21% of hospitals were fully compliant with all aspects of the rule. The OIG estimated that 46% of hospitals were not fully compliant based on its own audit sample. This matters because hospital price transparency was supposed to enable patients to comparison shop for medical services the way they compare prices for anything else. Without accurate, complete price data, a patient cannot know that Hospital A charges $4,500 for an MRI while Hospital B five miles away charges $800 for the same scan. The entire premise of the rule — that sunlight would create competitive pressure to lower prices — fails when the majority of hospitals either do not publish their data, publish incomplete data, or publish data in formats that are technically compliant but practically unusable. A GAO report found that CMS lacks the information to assess whether hospital pricing data is even complete or accurate. This problem persists because the enforcement mechanism is toothless. CMS fined only 3 hospitals in 2024 and 10 in 2025, with fines ranging from $32,301 to $309,738. For a hospital system generating hundreds of millions or billions in annual revenue, a $300,000 fine is a rounding error — far less than the revenue they would lose if patients could actually compare prices and choose lower-cost competitors. The rule requires hospitals to self-report their data with no independent verification, and the machine-readable file format requirements (changed in 2024 to require a CMS template) keep shifting, giving hospitals a perpetual excuse for non-compliance. The structural problem is that price transparency directly threatens hospital margins, and the penalty for non-compliance is orders of magnitude smaller than the financial benefit of opacity.

CMS's 2024 improper payment analysis found that CPT code 99214 — a standard outpatient evaluation-and-management visit for 'moderate complexity' — accounted for $459 million in overpayments. Sixty-three percent of those overpayments were caused by incorrect coding, overwhelmingly upcoding: billing for a higher-complexity visit than the clinical documentation supports. A separate code, CPT 99233, was responsible for over $490 million in overpayments, with more than 22% of all claims for that code being upcoded. The total improper payment rate across Medicare fee-for-service was 6.55%, amounting to over $28 billion. This matters because upcoding flows directly to patient cost-sharing. When a visit is coded as 99214 instead of 99213, the patient's copay, coinsurance, and deductible charges all increase. For an insured patient, the difference might be $20-$50 per visit. For an uninsured patient paying the full charge, it can be $80-$150 more. Multiply that by millions of visits and the aggregate patient overpayment is in the hundreds of millions of dollars. Most patients have no way to know they were upcoded because the codes appear as cryptic numbers on their Explanation of Benefits, with no plain-language description of what level of service they actually received. This problem persists because of a fundamental information asymmetry: the billing staff selects the code, and neither the patient nor the insurer was present during the visit to verify the complexity level. Documentation audits catch some errors after the fact, but CMS audits a tiny fraction of total claims. Meanwhile, billing staff face productivity pressure — higher codes mean higher revenue — and the penalty for upcoding is typically just returning the overpayment if caught, with no additional financial penalty in most cases. The 2021 E/M documentation guidelines simplified the rules but also made it easier to justify higher-level codes with less specific documentation, which contributed to the spike in 99214 usage and errors.

Medicare Advantage plans denied 3.2 million prior authorization requests in a single year. The HHS Office of Inspector General found that 13% of those denials were for services that met Medicare coverage rules — meaning those services would have been approved automatically under traditional Medicare. When patients or providers do appeal, 80.7% of denials are partially or fully overturned, which strongly suggests the initial denials were inappropriate. Yet fewer than 1% of denied claims are ever appealed. This matters because the math creates a perverse incentive. If a plan denies 1,000 claims and only 10 are appealed, even if all 10 appeals are overturned, the plan still avoided paying 990 claims. The OIG found that 18 of the 25 largest Medicare Advantage insurers had 'significant deficiencies' in their claims processing and appeals systems. For patients — predominantly elderly, often managing multiple chronic conditions — the appeals process requires filling out forms, gathering medical records, writing letters, and waiting weeks or months for a decision. Many patients simply give up, pay out of pocket, or go without the treatment. A patient denied a prior authorization for a necessary medication became septic, spent 39 days in the hospital, and then responded well to the medication — 70 days after it was originally prescribed. This problem persists because Medicare Advantage plans are paid a fixed capitated rate per enrollee, so every claim they deny is money saved. The penalties for inappropriate denials are weak — CMS conducts audits but rarely removes plans from the program. The appeals process places the burden entirely on the patient or provider, and the 80%+ overturn rate proves that the system functions as a friction-based cost control mechanism: deny first, pay only if the patient has the knowledge, resources, and persistence to fight back. Congress has not mandated automatic approval when denial rates or overturn rates exceed thresholds, so there is no structural feedback loop that punishes plans for systematic over-denial.

When a hospital system acquires a physician's private practice, the same doctor in the same office seeing the same patient for the same condition can suddenly generate a bill that is $200-$600 higher than before. The difference is a 'facility fee' — a charge for hospital outpatient services that the practice can now bill because it is technically a hospital department, even though it is still a standalone office building miles from the hospital campus. Between 2012 and 2024, the percentage of physicians employed by hospitals doubled to 55%. In Connecticut alone, hospitals collected over $1.4 billion in facility fees in 2024. This matters because patients are blindsided. A parent takes their 3-year-old to a follow-up appointment at a pediatric clinic and gets a $423 bill for 'outpatient hospital services' even though the child never set foot in a hospital. A retiree sees the same cardiologist they have seen for a decade and suddenly owes $350 more than last year because the practice was acquired by a health system. The patient did nothing different. The doctor did nothing different. The location did not change. But the billing entity changed, and that change triggers a second bill that the patient had no reason to expect. This problem persists because of a regulatory loophole in how CMS defines 'provider-based' departments. When a hospital acquires a practice and designates it as an off-campus provider-based department, it can bill both a professional fee (the doctor's charge) and a facility fee (the hospital's charge) for the same visit. Only 21 states have passed laws targeting facility fees, and many of those laws are narrow — requiring only signage in the office on the day of the visit, which is ineffective because patients have already scheduled and traveled to the appointment. There is no federal prohibition, and hospitals have strong financial incentives to continue acquiring practices precisely because facility fees represent pure margin on existing patient volume.

When you call 911, you do not choose your ambulance provider. The dispatcher sends whoever is closest. Nearly 80% of ground ambulance rides result in out-of-network bills because patients have zero ability to select an in-network provider during an emergency. The average surprise ambulance bill is $450, but bills of $2,000-$8,000 are common for rides under 10 miles. This matters because the No Surprises Act — the federal law specifically designed to eliminate surprise medical bills — deliberately excluded ground ambulances. Air ambulances are covered. Emergency room doctors are covered. Out-of-network anesthesiologists are covered. But the ambulance that brings you to the ER is not. This means a patient who does everything right — goes to an in-network hospital, sees in-network doctors — can still face a surprise bill for the ride there. And because ambulance services are often run by private equity-backed companies or municipal fire departments with opaque pricing, there is no way to comparison shop, no way to negotiate in advance, and no way to opt out. The reason this gap persists is structural: ground ambulance services are regulated by a patchwork of local, county, and state jurisdictions, making federal regulation complex. Congress punted the issue to an advisory committee (the GAPB), which recommended capping patient costs at $100 per ride in late 2023, but implementing that recommendation requires Congressional legislation that has not materialized. Meanwhile, 22 states have passed some form of ground ambulance protection, but those laws only cover state-regulated insurance plans. Sixty-three percent of Americans with employer-sponsored insurance have self-funded plans that are federally regulated under ERISA, meaning state protections do not apply to the majority of working Americans. The result is a deliberate, known hole in federal consumer protection that leaves tens of millions of patients exposed to bills they cannot predict, prevent, or negotiate.

Community gardens are expanding rapidly in post-industrial cities like Detroit, Cleveland, Baltimore, Newark, and Philadelphia — cities where vacant lots from demolished housing and shuttered factories are being converted to food production. These lots frequently sit on land with a century of contamination from leaded gasoline deposition, lead paint flaking from adjacent buildings, industrial emissions, and legacy pesticide use. A 2024 study published in Sustainability analyzing trace metal contamination in community garden soils across the United States found widespread exceedances of screening levels for lead, arsenic, cadmium, and zinc. Yet the EPA has no specific soil contamination standards for urban food gardens. The screening levels that do exist (like the 200 ppm residential lead standard) were designed to protect children playing in yards, not to assess the safety of food grown in contaminated soil — a fundamentally different exposure pathway because plants can bioaccumulate certain metals. This matters because community gardens serve the very populations most vulnerable to contamination: low-income families in food deserts who rely on garden produce for nutrition. A gardener in Cleveland's Slavic Village or Baltimore's Sandtown-Winchester may be growing tomatoes and leafy greens in soil with 500 to 2,000 ppm of lead and have no idea. Basic soil testing for heavy metals costs $30 to $100 per sample, and most gardeners need multiple samples to characterize a single plot. Some cities offer free or subsidized testing — New York State runs a Community Gardens Soil Testing Program, and the Cuyahoga Soil and Water District offers free basic tests — but lead screening costs an additional $30, and most programs are limited in scope, require applications, and have long turnaround times. The result is that the majority of urban community gardens in the U.S. have never been tested for heavy metals. This problem persists because community gardens fall through every regulatory crack. They are not regulated as agricultural operations (too small), not covered by residential soil standards (wrong exposure pathway), and not subject to commercial food safety regulations (produce is consumed by the growers, not sold commercially in most cases). No federal, state, or local agency is responsible for ensuring community garden soil is safe. University cooperative extension services can test soil but have no mandate or funding to proactively screen urban garden plots. The gap between the rapid expansion of urban food production and the complete absence of a regulatory framework for soil safety in these gardens means contaminated food is being grown and eaten right now in every major post-industrial American city.

When a contaminated site is cleaned up to a level safe for restricted use (e.g., commercial but not residential), institutional controls (ICs) are placed on the property to prevent future uses that could expose people to remaining contamination. These controls include deed restrictions, environmental covenants, groundwater use prohibitions, and activity restrictions. They are the most common 'remedy' at Superfund and state-regulated sites — cheaper than full cleanup, they allow contamination to remain in place as long as human exposure pathways are blocked by legal and administrative mechanisms rather than physical removal. The problem is that institutional controls are only effective if they are monitored, enforced, and communicated across property transactions — and the systems for doing this are broken. In California, the primary monitoring mechanism for land use covenants is self-reported annual inspections by the property owner (the 'obligated party'). Local planning and building departments are not required to notify the Department of Toxic Substances Control when a property owner requests a land use change that contradicts the restrictions in a covenant. When contaminated properties change hands — which they inevitably do over the decades or centuries that residual contamination persists — new owners frequently have no awareness of the restrictions. Title searches may miss environmental covenants if they were recorded improperly. Zoning changes can conflict with deed restrictions without triggering any alert. The result is that people build homes, daycare centers, and community gardens on land that was explicitly restricted to commercial or industrial use because the contamination underneath was never fully removed. This problem persists because institutional controls were designed as a cost-saving compromise — a way to close out sites without the expense of complete remediation — but the long-term stewardship infrastructure was never built. There is no national registry of institutional controls. State databases are fragmented, inconsistent, and often inaccessible to the public. Real estate transaction processes do not systematically check environmental restriction databases. Title companies are not required to search for environmental covenants in most states. The EPA and state agencies lack the staff and budget to conduct independent compliance inspections at thousands of IC-dependent sites. The fundamental assumption — that a legal document recorded in a county clerk's office will effectively prevent harmful land use for 50 to 100 years — is contradicted by everything we know about institutional memory, property turnover, and regulatory capacity.

Environmental remediation projects routinely fail because the technology selected does not match actual subsurface conditions. In-situ remediation — treating contamination in place rather than excavating it — is chosen for its cost savings, but what works at one site frequently does not work at another because subsurface geology, hydrology, geochemistry, and contaminant distribution are highly site-specific. A common failure mode is applying an anaerobic treatment technology in an aerobic environment, or injecting chemical oxidants into soil with high organic matter content that consumes the oxidant before it reaches the target contaminant. When an in-situ project fails, the typical re-remediation cost multiplier is 2 to 3 times the original project budget, because the failed treatment must be reversed or its byproducts managed before the correct technology can be deployed. This matters because remediation cost overruns do not just waste money — they extend the timeline during which communities are exposed to contamination. A project that was supposed to take 3 years and cost $2 million becomes a 10-year, $6 million effort while residents continue drinking from contaminated wells or breathing contaminated air. For brownfield redevelopment, failed remediation can collapse an entire real estate transaction: the developer's budget is blown, the lender pulls financing, and the site returns to vacancy. For Superfund sites funded by taxpayers, every dollar spent on re-remediation is a dollar not available for the 1,340 other sites waiting for cleanup. This problem persists because of economic pressure to minimize upfront site characterization costs. Adequate characterization — high-resolution soil sampling, groundwater flow modeling, bench-scale treatability testing — can cost $200,000 to $500,000 for a complex site. Clients and regulators push consultants to reduce these costs, which leads to insufficient data, which leads to wrong technology selection. The counter-intuitive rule of remediation cost management — that spending more on characterization almost always reduces total project cost — is well-established in the technical literature but systematically ignored in practice because characterization costs are visible and immediate while failure costs are uncertain and deferred. There is no industry-wide quality assurance mechanism that tracks remediation success and failure rates, so the same mistakes are repeated across thousands of sites with no feedback loop.

Before any brownfield can be redeveloped, a Phase I Environmental Site Assessment must be completed to identify potential contamination, followed by a Phase II ESA that involves actual soil and groundwater sampling to confirm or rule out contamination. Phase II assessments typically cost $10,000 to $30,000 per site but can exceed $100,000 for complex sites with multiple potential contaminant sources. A small city with 50 to 200 suspected brownfield parcels — a typical number for a post-industrial municipality in the Rust Belt, Appalachia, or the rural South — faces an assessment bill of $500,000 to $6 million before any cleanup or development can begin. This cost creates a catch-22 that keeps contaminated land idle for decades. Without a Phase II ESA, the extent of contamination is unknown, which means no lender will finance development, no insurer will write an environmental liability policy, and no developer will take on the risk. But without a committed developer or identified funding source, there is no economic justification for spending $30,000 to assess a single parcel in a weak real estate market where the remediated land might sell for less than the cost of assessment. EPA Brownfields assessment grants provide up to $25,000 per site for Phase II work, but these grants are competitive, oversubscribed, and require administrative capacity that many small municipalities lack. The result is thousands of contaminated parcels sitting vacant in cities that desperately need tax revenue, housing, and economic development. This problem persists because the environmental assessment industry is structured around individual site transactions, not portfolio-level municipal assessment. There is no standardized, lower-cost screening methodology that could triage a city's brownfield inventory to identify which parcels are likely contaminated and which are not. Cutting assessment scope to save money reduces investigation quality, creating a greater margin of error that leads to higher cleanup costs, legal liability, and regulatory penalties later. The EPA's 100% cost reimbursement for Phase II pre-purchase assessments (up to $25,000) helps but only covers one site at a time and requires the municipality to have already identified a willing buyer. Small cities with understaffed planning departments cannot manage the grant application, environmental consulting, and regulatory compliance processes simultaneously across dozens of sites.

Low-income communities and communities of color disproportionately host brownfield sites due to 20th-century redlining, discriminatory zoning, and patterns of industrial siting that concentrated hazardous land uses in neighborhoods with the least political power. When these sites are finally cleaned up and redeveloped, the resulting property value increases price out the residents who lived with the contamination for decades. A national assessment studying brownfield redevelopment from 2006 to 2015 found that in neighborhoods around 169 redeveloped brownfields in the mid-Atlantic, Southeast, and Southwest, Black households were displaced 61% of the time. Research from Chicago confirms that Black and Hispanic households benefit less than White households from brownfield cleanup, and renters benefit less than owners. The injustice compounds over time. Residents who endured the health consequences of living near contaminated land — elevated asthma rates, cancer risk, developmental delays in children — do not get to enjoy the health benefits of the cleanup because they are pushed out by rising rents and property taxes before remediation is complete. New York City's High Line, built on a former elevated rail corridor, is the canonical example: the conversion spurred luxury development that displaced small businesses and lower-income households from the surrounding neighborhood. The jobs created by brownfield redevelopment do not necessarily go to local residents, further concentrating benefits among newcomers rather than the community that bore the environmental burden. This problem persists because environmental remediation policy and housing policy operate in completely separate silos. EPA Brownfields grants fund assessment and cleanup but include no anti-displacement requirements. Local zoning decisions about what gets built on remediated land are made by planning commissions that are not required to consider displacement impacts. The EPA has begun studying this dynamic — conducting case studies in Indianapolis, Milwaukee, and Philadelphia — but has not translated findings into binding policy. Anti-displacement tools like community land trusts, inclusionary zoning, and property tax freezes exist but are rarely integrated into brownfield redevelopment plans because no federal program requires or incentivizes their use. The result is a system that cleans up contamination for the benefit of future wealthier residents while externalizing the health and displacement costs onto the original community.

Trichloroethylene (TCE) and tetrachloroethylene (PCE) — chlorinated solvents used for decades in dry cleaning, metal degreasing, and electronics manufacturing — persist in groundwater plumes beneath thousands of U.S. neighborhoods. These volatile organic compounds migrate upward through soil as vapor, penetrate building foundations through cracks and utility conduits, and accumulate in indoor air at concentrations that cause cancer, liver damage, and neurological harm. The EPA estimates that approximately 20% of U.S. residents live in areas where vapor intrusion could potentially affect indoor air quality. In California alone, the Department of Toxic Substances Control identified 214 sites that are known or suspected vapor intrusion hotspots. In Los Angeles, the former Rocketdyne rocket engine plant in Canoga Park and an industrial dry cleaner called Welch's in Lincoln Heights (shuttered in the 1980s) are still generating groundwater plumes that threaten residential buildings more than 35 years later. The human cost is measurable. At Camp Lejeune, where TCE and PCE contaminated base water from the 1950s through 1987, a study found thyroid, esophageal, and breast cancer rates 20% higher among personnel stationed there, and Parkinson's disease rates 70% higher. The EPA banned all uses of TCE in December 2024, citing Camp Lejeune and other sites as evidence of harm. But banning future use does nothing about the millions of gallons of TCE and PCE already in the ground. Residents living above these plumes often have no idea they are being exposed — symptoms like headaches, dizziness, and fatigue are non-specific and easily attributed to other causes. By the time cancer clusters are identified, exposure has been ongoing for years or decades. This problem persists because vapor intrusion assessment is expensive, technically complex, and governed by an inconsistent patchwork of state regulations. A 2024 review found that state approaches to vapor intrusion vary enormously — some states have mandatory screening requirements, while others have none. There is no federal requirement for vapor intrusion assessment at non-Superfund sites. Many contaminated dry cleaner and industrial sites are regulated under state voluntary cleanup programs that lack the resources or authority to mandate indoor air testing in nearby residences. Property owners and landlords have no obligation in most jurisdictions to test for or disclose vapor intrusion risks. Local planning and building departments are not required to check contamination databases before issuing permits for new construction above known plumes.

The Department of Defense has identified 723 military installations where PFAS-containing aqueous film-forming foam (AFFF) was used for firefighting training, potentially contaminating soil and groundwater at each site. As of March 2025, preliminary assessments and site inspections have been completed at 703 of these installations, with 586 advancing to the next phase of the CERCLA cleanup process. The estimated future cost for PFAS investigation and cleanup has more than tripled since 2022, reaching $9.3 billion in fiscal year 2025 and beyond, according to a February 2025 GAO report (GAO-25-107401). At Cannon Air Force Base in New Mexico, groundwater PFAS levels were found at 370 times what federal scientists consider safe, and at Fort Meade in Maryland, monitoring wells near former fire-training areas showed levels as high as 87,000 parts per trillion. The communities surrounding these bases bear the consequences. Residents near Cannon AFB had their private wells contaminated, and the Highland Dairy was forced to euthanize over 3,000 cows. In January 2025, the base was fined $98,780 for failing to report a PFAS spill within the required 24-hour window. Families living near these installations face elevated cancer risk, thyroid disease, and immune suppression, and many cannot sell their homes because contamination disclosure destroys property values. DOD's own cleanup budget increased by only $400 million between 2016 and 2021 while the overall cleanup backlog soared by $3.7 billion — a funding velocity that would take over 50 years to address the current scope. This problem persists because there is no proven, cost-effective technology to remove PFAS from soil at scale. PFAS are called 'forever chemicals' for a reason: the carbon-fluorine bond is one of the strongest in organic chemistry. Current approaches — excavation and landfilling, soil washing, thermal treatment — either move the problem elsewhere or are prohibitively expensive for the acreage involved. The EPA designated PFOA and PFOS as CERCLA hazardous substances in May 2024, which grants enforcement authority but does not create a technology that works. Meanwhile, DOD continues to resist full transparency: the GAO found that DOD needs to provide Congress with more complete cost information, and the department's own cost projections keep growing as the true extent of contamination at each base is revealed through remedial investigation.