Real problems worth solving

Annual appropriations to the EPA's Superfund program declined from approximately $2.6 billion in fiscal year 1999 to $537 million in fiscal year 2024 — a 79% nominal cut and roughly 65% in inflation-adjusted terms. During that same period, the sites remaining on the National Priorities List became harder and costlier to remediate because the simpler, cheaper sites were cleaned up first. As of March 2025, 1,340 active sites remain on the NPL, with 41 more proposed for listing. Only 459 sites have been deleted (fully cleaned up) since the program began in 1980. The Superfund excise tax on chemical and petroleum companies that originally funded the program expired in 1995 and was not reinstated until 2022. When it came back, revenue fell far short of projections: EPA expected $1.7 billion in chemical tax receipts for fiscal 2024 but collected only $472.8 million — less than a third of the estimate. The trust fund's unobligated balance sits at roughly $5.5 billion, but that money is not being spent at the rate needed. Communities living near these sites — disproportionately low-income and communities of color — continue to drink contaminated water, breathe contaminated air, and suffer elevated rates of cancer, neurological disease, and reproductive harm while waiting decades for cleanup that may never come. This problem persists because Superfund has a structural design flaw: it depends on either (a) identifying a 'potentially responsible party' who can be forced to pay, or (b) federal appropriations from Congress. For 'orphan sites' where no responsible party exists or is solvent, the program relies entirely on Congressional funding, which has been in secular decline for 25 years. The 2025 GAO report (GAO-25-108408) confirmed that staffing shortages at EPA regional offices and state environmental agencies are a compounding factor — even when money is available, there are not enough people to manage the cleanup projects. The Trump administration's 2025-2026 workforce reduction, which cut approximately one-third of EPA staff (over 4,000 employees, many with 30+ years of institutional knowledge), has further degraded the agency's capacity to manage existing sites, let alone take on new ones.

For decades, U.S. wastewater treatment plants sold or gave away sewage sludge (biosolids) to farmers as free fertilizer, a practice endorsed and even subsidized by state environmental agencies. That sludge contained PFAS — per- and polyfluoroalkyl substances — from industrial and household wastewater. The PFAS accumulated in soil, leached into groundwater, and entered the food chain through crops and livestock. Maine farmer Fred Stone of Stoneridge Farm in Arundel discovered in 2016 that sludge spread on his land from 1983 to 2004 under a state-sponsored program had left soil contamination as high as 800,000 parts per trillion of PFAS. He had to euthanize most of his dairy herd because PFAS levels in their milk exceeded safety thresholds. His family's blood tested at more than 20 times the national PFAS average. Stone lost his livelihood, was denied federal farm disaster aid, spent $22,000 on a water filtration system and $500 per month on ongoing tests, and now lives on welfare. He is not an outlier. Across Maine, farms that accepted state-approved sludge are discovering the same contamination pattern. A second farm in Fairfield was shuttered in 2020 after contaminated milk was traced to sludge application. The crisis forced Maine to become the first state to ban biosolid land application in 2022, but the ban created a new problem: municipalities now have nowhere to put their sludge, and landfills are filling up with PFAS-laden material that will leach into groundwater from a different location. This problem persists because of a fundamental liability gap. The farmers did not create the contamination — they accepted sludge that state agencies told them was safe. The wastewater utilities that produced the sludge did not manufacture the PFAS — they received it in influent from industrial and commercial sources. The PFAS manufacturers (3M, DuPont) are settling lawsuits but those settlements go to water utilities, not individual farmers. There is no federal program to remediate PFAS-contaminated farmland, no USDA mechanism to compensate farmers for lost agricultural productivity, and no established technology to remove PFAS from soil at agricultural scale. Sludge is still applied to approximately 5% of all U.S. crop fields in states without bans, meaning the contamination footprint continues to grow.

In January 2024, the EPA halved the recommended residential soil lead screening level from 400 parts per million to 200 ppm for the first time in 30 years (and to 100 ppm where multiple lead sources exist). Research from Indiana University-Purdue University Indianapolis estimates that roughly one in four U.S. households — approximately 29 million homes — now have yard soil exceeding the new threshold. The total remediation cost to address all affected properties ranges from $290 billion to $1.2 trillion. This matters because the new screening level is guidance, not an enforceable regulation, and applies primarily to Superfund and RCRA sites. Homeowners in older neighborhoods with legacy lead paint, leaded gasoline fallout, or proximity to former smelters have no federal mechanism to pay for soil replacement, which costs $10,000 to $30,000 per yard. Most of these homeowners are low-income families in formerly redlined neighborhoods who cannot afford voluntary remediation. Their children continue playing in contaminated dirt every day. Blood lead levels in children living in these neighborhoods remain 2-3 times higher than the national average, causing irreversible cognitive damage, lower IQ, and behavioral problems that compound across a lifetime of reduced earnings and health outcomes. This problem persists because the EPA's authority under CERCLA only triggers cleanup obligations when a site is listed on the National Priorities List or is part of a RCRA corrective action — mechanisms designed for industrial sites, not residential yards. There is no federal program analogous to the Lead Service Line Replacement program for soil. State voluntary cleanup programs exist but are drastically underfunded. The result is a regulatory announcement that reclassified millions of homes as contaminated without providing any pathway to actually fix them. In October 2025, the EPA further muddied the picture by issuing a directive raising the removal management level back to 600 ppm for Superfund sites, tripling the January 2024 benchmark and creating confusion about which standard applies where.

In February 2026, approximately 200 clerical workers at the San Francisco Superior Court launched an open-ended strike over chronic understaffing that had been festering since at least October 2024, when clerks staged a one-day walkout. The core complaint was not wages — it was that staffing levels were so low that overburdened clerks were making paperwork errors with life-altering consequences. Specifically, clerks reported that errors in processing release orders and sentence calculations resulted in people being kept in custody longer than they were supposed to be. Meanwhile, the San Francisco District Attorney filed 8,400 cases in 2025 compared to about 5,600 in 2021 — a 50% increase in caseload without a corresponding increase in clerical staff. Sacramento had cut court funding by $97 million statewide in 2024 due to budget concerns. The human cost of a clerical error in a courthouse is categorically different from a clerical error in almost any other workplace. When a clerk at a private company makes a data entry mistake, it might delay a shipment or miscalculate an invoice. When a court clerk makes a data entry mistake, a human being sits in a jail cell for days or weeks beyond their lawful release date. They miss work, their employer replaces them, their family does not know where they are, and they have no recourse because the 'error' is buried in a case management system that the person in custody cannot access or challenge in real time. The National Center for State Courts found that over 70% of courts reported staffing shortages, and 61% expect shortages to continue. This is not an isolated San Francisco problem — it is a systemic failure playing out in courthouses across the country. The structural cause is that court clerks are classified and compensated as low-level government administrative staff, despite performing work that directly affects people's liberty and legal rights. Starting salaries are typically $35,000-$50,000 in high-cost-of-living areas, making it impossible to recruit and retain competent staff. Courts compete for workers with every other employer in the market, but they cannot offer competitive salaries because their budgets are set by state legislatures that view court operations as a cost center to be minimized. When budgets get tight — as they did in California's $97 million cut — courts cannot raise prices or find new revenue; they simply do the same work with fewer people, and the errors compound. The work itself is also deeply unpleasant: processing violent crime cases, interacting with distressed litigants, navigating archaic case management software, all under the pressure of knowing that a mistake could keep someone in jail. The result is chronic turnover, institutional knowledge loss, and a workforce perpetually in crisis.

The federal judiciary charges the public $0.10 per page to access court records through PACER (Public Access to Court Electronic Records), the only comprehensive system for viewing federal case filings. While individual pages seem cheap, a single complex case can contain thousands of pages across hundreds of docket entries, meaning that following one case can cost hundreds of dollars. Journalists investigating judicial corruption, researchers studying sentencing patterns, legal aid organizations tracking systemic issues, and pro se litigants trying to find precedent for their own cases all face this paywall. A federal judge ruled in 2018 that PACER fees were being impermissibly used to fund unrelated court technology projects — the judiciary was overcharging for access and spending the surplus on internal IT systems that have nothing to do with public records access. The direct harm is that public court records — which are supposed to be public — are effectively private for anyone without a budget. A journalist investigating a federal judge's pattern of unusually harsh sentences cannot afford to download the sentencing documents from hundreds of cases. A legal aid attorney trying to find similar cases to help a client cannot spend $300 on PACER downloads when their organization operates on a shoestring budget. A pro se litigant trying to understand how judges in their district have ruled on similar motions is locked out of the information that would help them make their case. The result is an information asymmetry where large law firms with PACER budgets have comprehensive access to the judicial record, while everyone else operates in the dark. This is not a hypothetical access problem — it is a daily reality for thousands of people trying to engage with the federal court system. PACER persists as a paywall because the judiciary has become dependent on the revenue. The fee-funded model means the courts do not have to ask Congress for appropriations to run their electronic records system, which gives the judiciary independence but at the cost of public access. Bipartisan legislation (the Open Courts Act) has been introduced multiple times to eliminate PACER fees and fund the system through appropriations, but it has never passed because it would require Congress to actually appropriate the money — approximately $150 million per year — and no legislator wants to champion spending $150 million on 'making court records free' when the current system 'works.' The EFF, Free Law Project, and other organizations have campaigned for years to end the paywall, and the class action lawsuit resulted in a favorable ruling, but the judiciary has been slow to implement meaningful changes. The structural incentive is clear: the courts benefit from the revenue, Congress does not want to replace it, and the people harmed lack the political power to force action.

Filing procedures account for 45% of e-filing rejections in court systems, with missing documents, formatting issues, and incomplete forms being the top causes. For self-represented litigants — who now make up the majority in family law, housing, and debt cases — court e-filing portals present interfaces designed for attorneys, filled with legal jargon, jurisdiction-specific requirements, and form fields that assume knowledge of court operations. A single formatting error (wrong margin size, incorrect document type designation, missing cover sheet) results in rejection, often with a cryptic error message that provides no guidance on how to fix the problem. The filing deadline may pass while the litigant tries to figure out what went wrong. A missed filing deadline due to a rejected e-filing is not a minor procedural hiccup — it can be case-ending. In many jurisdictions, failing to file a response by the deadline results in a default judgment. For a tenant fighting an eviction, this means losing their home. For a parent in a custody dispute, this means losing time with their children. For a debtor, this means wage garnishment. The irony is that e-filing was supposed to make courts more accessible by eliminating the need to physically go to a courthouse during business hours. Instead, it created a new digital barrier that is in some ways worse than the physical one, because at least at a courthouse window, a clerk could point out errors in real time and explain how to fix them. The root cause is that court technology is procured and designed for the court's internal workflow, not for the end user. E-filing systems are built to make life easier for clerks processing filings, not for litigants submitting them. Courts do not conduct usability testing with self-represented litigants, do not iterate on user feedback, and do not measure filing success rates by user type. The procurement process selects vendors based on compliance with technical specifications, not on user experience quality. As a result, e-filing systems have the usability of 2005-era enterprise software while the people forced to use them have the expectations of 2025-era consumer apps. Until courts treat litigants as users whose experience matters, e-filing will remain a barrier dressed up as an improvement.

In eviction courts across the United States, landlords are eight times more likely to have legal representation than tenants. Only about 4-12% of tenants have a lawyer, compared to 83% of landlords. The impact of this representation gap is not subtle: having a lawyer reduces the probability of a possessory judgment (losing one's home) by 38.5 percentage points and the probability of a warrant of eviction being issued by 38.1 percentage points. In New York City, where a right-to-counsel program exists, 72-93% of represented tenants were able to remain in their homes, depending on the borough. In jurisdictions without such programs, unrepresented tenants face a system designed by and for lawyers, with no guide. The 'so what' runs straight to homelessness. An eviction judgment does not just mean losing a current apartment — it creates a record that follows the tenant for years, making it nearly impossible to rent from any landlord who runs a background check. A single eviction filing (even if the tenant wins) can appear on screening reports and result in automatic denials. For families with children, eviction means school disruption, which correlates with lower academic achievement and higher dropout rates. For working adults, the instability of homelessness or doubling up with family makes it harder to maintain employment. The economic cost of a single eviction to public services — emergency shelter, healthcare, child welfare — far exceeds the cost of providing a lawyer, which is why right-to-counsel programs consistently show positive ROI. But only a handful of cities have implemented them. The structural reason this persists is that eviction courts were designed as landlord debt-collection mechanisms, not as forums for resolving housing disputes fairly. The procedures, timelines, and default rules all favor the filing party. Cases move on timelines of days or weeks — far too fast for a tenant to find, qualify for, and engage a legal aid lawyer, even if one is available. Legal aid organizations are chronically underfunded and can only serve a fraction of eligible clients. The federal Legal Services Corporation's budget has been essentially flat in real dollars for decades. Meanwhile, landlord attorneys handle evictions at volume — filing hundreds of cases per month with template pleadings — making it economically efficient for landlords but structurally impossible for tenants to compete without representation.

Queens County, New York — one of the most linguistically diverse places on Earth — has just 41 staff interpreters to cover 160 languages spoken by litigants in its courts. Statewide, the New York court system lost 69 interpreters (a 23% decrease) between 2019 and 2025, with New York City alone losing 66 staff interpreters (a 27% decrease). When an interpreter is unavailable for a scheduled hearing, the case gets adjourned — often for weeks or months. For a defendant in custody, this means additional time in jail. For a family in a custody dispute, it means more months of uncertainty about where their children will live. The concrete harm is not just delay — it is procedural injustice during the hearings that do proceed. When courts cannot get a qualified interpreter for a less common language, they sometimes proceed with an unqualified interpreter or a family member, which introduces errors, omissions, and conflicts of interest into the record. A litigant who does not fully understand the proceedings may agree to terms they do not comprehend, waive rights they do not know they have, or fail to present critical evidence because they cannot communicate it. In family law and immigration cases, these communication failures can result in a parent losing custody or a person being deported. The right to an interpreter exists on paper, but when the interpreter pool is 41 people deep and the language demand is 160 languages wide, the right is functionally unenforceable. The shortage persists because court interpreting is a highly specialized, poorly compensated profession with no clear career path. Certified court interpreters must pass rigorous exams — in some states, the pass rate is under 20% — but starting salaries are often $40,000-$55,000, well below what bilingual professionals can earn in the private sector. Courts rely heavily on per diem freelance interpreters, who cost $50-$150 per hour but are unreliable for scheduling because they take the highest-paying gig available on any given day. California launched a $6.8 million pilot program to address this, but the scale of investment is minuscule compared to the scope of the problem. The fundamental issue is that courts treat interpretation as an operational expense to be minimized rather than a constitutional requirement to be guaranteed.

The number of federal civil cases pending for more than three years rose from 18,280 in March 2004 to 81,617 in March 2024 — a 346% increase. Civil case filings in U.S. district courts rose 22% in 2024 alone. Meanwhile, Congress has not passed comprehensive judgeship legislation since 1990, even as district court filings have increased more than 35% since then. The Judicial Conference recommended creating 66 new permanent district court judgeships in March 2023, but Congress has not acted. The average time from filing a civil case to trial is now over two years nationally, and in many overworked districts it stretches to three or four years. A three-to-four-year wait for trial is not an inconvenience — it is a denial of justice for the party who cannot afford to wait. A small business suing a larger competitor for patent infringement or breach of contract needs resolution within months to survive; waiting four years means the business may be bankrupt before it ever sees a courtroom. An individual suing an employer for wrongful termination needs the case resolved while they can still find comparable employment and before their savings run out. The delay itself becomes a weapon: well-resourced defendants know that stretching a case out increases the plaintiff's costs and desperation, making them more likely to accept a lowball settlement. Justice delayed is not just justice denied — it is justice auctioned to the party with deeper pockets. The structural cause is straightforward: creating federal judgeships requires an act of Congress, and judgeships have become politically toxic because each new judge is a lifetime appointment that the party in power gets to fill. Neither party wants to create judgeships that the other party's president will appoint. So the judiciary starves while Congress plays political chicken. The Judicial Conference's recommendation process is rigorous and nonpartisan — based on weighted caseload analysis accounting for case complexity, senior judge contributions, and magistrate assistance — but its recommendations have been ignored for over three decades. The result is that the federal court system is running on a staffing model designed for the caseload of 1990, while handling the caseload of 2024.

In a practice known as 'sewer service,' process servers file affidavits swearing they personally delivered lawsuit papers to defendants — when in fact they threw the papers away or never visited the address at all. The New York Attorney General sued American Legal Process and 35 debt collection law firms, seeking to overturn over 100,000 default judgments obtained through fraudulent service. The Minnesota AG filed a similar suit against TJ Process Service. Studies have found that personal service to the defendant was actually achieved in only about 6% of cases in some jurisdictions. The other 94% of defendants may never have known they were being sued. The consequences are devastating and often invisible to the victim until it is too late. A person goes to use their debit card and it is declined — their bank account has been frozen by a judgment creditor they never heard of, for a debt they did not know was in litigation. They cannot pay rent, buy groceries, or fill a prescription. When they try to contest the judgment, they face the burden of proving they were never served, which requires hiring a lawyer they cannot afford for a case they did not know existed. The legal system is designed to trust the process server's affidavit, so the defendant is fighting uphill against a sworn document. Many people simply give up and accept the garnishment, which can take 25% of disposable income for years. Sewer service persists because the incentive structure rewards it. Process serving companies are paid per service, creating pressure to maximize volume. Debt collection law firms file thousands of cases per month and need quick default judgments to be profitable — they have no incentive to ensure defendants actually receive notice, because a contested case costs them money. Courts lack the resources to independently verify service, and judges process default judgment motions in bulk without questioning whether the defendant truly received notice. The penalty for sewer service, when caught, is typically a fine that is a fraction of the profits generated. Until courts implement independent verification of service — such as GPS tracking, photographic proof, or mandatory follow-up contact — the economic incentives will continue to favor fraud over due process.

Only 11 states have any program to send reminders for civil court dates, and not all of those programs are statewide. In the remaining 39 states, a person sued in civil court receives an initial summons — often months before the hearing — and then nothing. No reminder text, no email, no phone call. If they miss the hearing, the court enters a default judgment against them. In debt collection cases, this means automatic liability for the full amount claimed, plus fees, plus the creditor's right to garnish wages and freeze bank accounts. The person never had a chance to present their side. This matters because default judgments are not edge cases — they are the primary way debt collection lawsuits are resolved. When the vast majority of defendants in debt collection suits never appear, it is not because they agree they owe the money. Many dispute the amount, or the debt has been sold to a collector who cannot prove the original terms, or the statute of limitations has expired. But none of that matters if the defendant does not show up. A default judgment is a financial death sentence for low-income people: wages garnished means rent cannot be paid, which means eviction, which means homelessness or housing instability. The entire cascade starts because nobody sent a text message. Courts have not adopted reminder systems for the same reason most government services lag behind consumer technology: there is no institutional incentive to do so, and the people most harmed — low-income defendants in debt collection cases — have no political power to demand change. Courts are funded by legislatures that see text message reminder systems as a cost, not a savings, because they do not account for the downstream social costs of default judgments (emergency housing, public benefits, lost tax revenue from garnished workers). The technology is trivial — any SaaS company can send automated reminders — but court IT budgets are controlled by slow-moving government procurement processes, and judges have historically viewed their role as passive adjudicators, not active case managers responsible for ensuring parties actually show up.

Of the 2.49 million family law, probate, and unlimited civil hearings in California between April 2023 and March 2025, an estimated 1.78 million — 71.7% — had no verbatim record because no court reporter was present. The California Supreme Court itself ruled in Jameson v. Desta that the absence of a verbatim record will be 'frequently be fatal' to a litigant's ability to have claims of trial court error resolved on the merits on appeal. In practice, this means that for nearly three-quarters of civil hearings in the largest state court system in the country, the right to appeal is effectively dead on arrival. The 'so what' chain here runs deep. A parent loses a custody hearing because the judge misapplied the law. Without a transcript, the appellate court has no way to know what happened at trial — what evidence was presented, what the judge said, what objections were raised. The appeal gets denied not because the parent was wrong, but because there is literally no record to review. That parent loses custody of their child not on the merits, but because of a staffing shortage. Multiply this by 1.78 million hearings and you have a systemic denial of due process at industrial scale. The California Constitution guarantees due process and equal protection, but these guarantees are hollow when the infrastructure to enforce them does not exist. The stenographer workforce has declined 21% over the last decade, with only about 23,000 remaining nationally. The profession requires 2-4 years of specialized training with a high attrition rate — some programs report graduation rates below 15%. Courts cannot simply hire their way out of this shortage because the pipeline of new reporters is a trickle. Meanwhile, electronic recording — used routinely in 33 out of 35 states surveyed — faces resistance from the court reporter lobby, which has fought legislative efforts to allow digital alternatives. California dedicated $6.8 million to a five-year pilot program to hire more reporters, but this is a rounding error against the scale of the shortage. The structural problem is that courts built their entire appellate system around a profession that is disappearing, and the political and institutional inertia preventing a switch to digital recording means millions of litigants lose their appeal rights while stakeholders debate.

When courts replace their aging, often decades-old case management systems, the leading replacement — Tyler Technologies' Odyssey platform — has repeatedly caused wrongful arrests and unlawful detentions due to data migration failures and system defects. In North Carolina, within four days of eCourts launching in Mecklenburg County in October 2023, 66 people were held in jail longer than they should have been because the eWarrants module failed to properly track bond postings and release conditions. In Harnett County, clerks had to suspend operations entirely, shutting down multiple courtrooms because the system made case processing impossible. A federal class action lawsuit was filed alleging due process violations, and the court allowed the case to proceed to discovery. This matters because every hour someone spends wrongfully detained is an hour of lost wages, missed childcare, potential job loss, and psychological harm. People who are detained pretrial — even briefly — are more likely to plead guilty just to get out, regardless of actual guilt. The downstream consequences compound: a guilty plea means a criminal record, which means harder employment, housing, and credit access for years. The constitutional right to liberty is not abstract — it is the difference between a person making it to work on Monday or losing their apartment. This problem persists because courts are locked into a near-monopoly. Tyler Technologies dominates the court case management market, and courts lack the technical expertise to evaluate or negotiate these implementations effectively. The company has faced similar lawsuits over an eleven-year period (2011-2022) in Texas, California, Tennessee, and Indiana — including a $5 million settlement in Shelby County, Tennessee — yet continues winning new contracts because there are so few alternatives. Courts are not software companies; they cannot build their own systems, and the procurement process rewards the vendor with the longest track record, even when that track record includes repeated failures. The result is a cycle where the same vendor keeps getting hired, keeps failing during rollouts, and keeps settling lawsuits — while real people sit in jail cells they should not be in.

Across the United States, bike lanes simply end. A painted lane runs for eight blocks, then stops at a sign that reads "Bike Lane Ends" or "Share the Road" — and the cyclist is suddenly in a 35-40 mph travel lane with no transition, no merge zone, and no warning to motorists that a cyclist will be entering their lane. In Washington DC, the bike lane on New Jersey Avenue ends abruptly, forcing cyclists to ride unprotected through Capitol South before reconnecting with bike infrastructure at Independence Avenue. This pattern repeats in every American city: bike lanes are built in segments where right-of-way is available, with no requirement that they connect to each other. The danger is not hypothetical. When a cyclist merges from a protected or semi-protected facility into a general travel lane, they are making a lane change into faster traffic while being the smallest, slowest, least-visible vehicle on the road. Drivers in the travel lane are not expecting a cyclist to appear from the right because, for the preceding blocks, cyclists have been separated from them. The speed differential at the merge point is typically 15-25 mph, and the cyclist has no acceleration capability to match traffic speed. This is the inverse of a highway on-ramp: instead of an acceleration lane that lets you match speed before merging, the cyclist gets a painted line that evaporates. The problem persists because there is no federal standard requiring safe termination treatments for bike lanes. The MUTCD (Manual on Uniform Traffic Control Devices) provides guidance on bike lane markings but does not mandate specific merge transition designs. Cities build bike lanes project by project, each project ending at its funded boundary, with no requirement to connect to the next segment. The result is a network of disconnected fragments — what advocates call "bike lanes to nowhere" — where the most dangerous moments of a ride are exactly the points where the infrastructure abandons you.

When a cyclist falls or is pushed under the side of a large truck, they are dragged under the rear wheels. This is one of the most lethal crash types in urban cycling — the fatality rate for cyclist-truck side-impact crashes exceeds 50%. Side guards are simple metal panels or rails mounted between the front and rear wheels of trucks that prevent a person from going under the vehicle. Research shows they reduce cyclist fatalities by 55-75% and pedestrian fatalities by 20-27%. The European Union mandated blind-spot assistance systems and side guards for all newly type-approved heavy commercial vehicles starting July 2024. In the United States, there is no equivalent federal requirement. NHTSA has conducted cost-benefit analyses but has not issued a rule. A handful of cities have passed local ordinances — Boston was the first U.S. city to require side guards on city-contracted trucks — but these only apply to a small fraction of the truck fleet. The vast majority of trucks operating on U.S. urban streets have completely open sides. The structural reason the U.S. lags is the fragmented regulatory landscape. Vehicle safety standards are set federally by NHTSA, but cities and states cannot mandate vehicle equipment standards that conflict with or exceed federal rules (under federal preemption doctrine). So even a city that experiences multiple cyclist deaths from truck side-impact crashes cannot require all trucks entering the city to have side guards — only a federal rule can do that. The trucking industry has lobbied against side guard mandates, citing cost ($1,000-$2,500 per truck for installation) and weight (which affects cargo capacity). Meanwhile, the League of American Bicyclists has documented NHTSA's persistent failure to address cyclist safety in its regulatory agenda. The result is a regulatory gap where a proven, inexpensive safety device that is mandatory in Europe remains optional in the country where cyclist fatalities are rising fastest.

Nearly 50% of cyclist fatalities happen during nighttime hours, with the highest risk window between 6 PM and midnight. This is not simply because it is dark — it is because road lighting infrastructure is designed for motorists, not cyclists. Street lights are spaced and angled to illuminate the travel lane at driver eye height (approximately 4 feet above the road surface), but bike lanes sit at the edge of the roadway where light levels drop to a fraction of the main lane. A cyclist riding in a bike lane at night is often in a shadow zone between two streetlights, invisible to turning drivers. The problem compounds because cyclists themselves cannot see road hazards — potholes, debris, glass, grate slots — in time to avoid them at cycling speed. A car's headlights illuminate the road 200-300 feet ahead; a bicycle headlight illuminates 30-50 feet ahead. At 15 mph, a cyclist has roughly 1.5 seconds of reaction time to a pothole visible only in their own headlight beam. Research shows that road lighting has a clear positive effect on cycling rates, with the association being strongest among potential and less experienced cyclists — exactly the demographic cities need to attract to make cycling infrastructure investments pay off. The structural reason bike lanes remain dark is that lighting standards (like the IESNA RP-8 standard for roadway lighting) specify minimum illumination levels for vehicle travel lanes and pedestrian walkways but have no specific category for bicycle facilities. When a city's public works department installs or maintains street lighting, they are checking compliance against vehicle-lane minimums. The bike lane is not in the specification, so it is not in the budget. Adding dedicated bike lane lighting would require either new light poles (expensive, requires underground wiring) or retrofitting existing poles with additional lower-mounted fixtures angled toward the bike lane — a solution that is technically straightforward but has no institutional champion because it falls between the transportation department (which builds bike lanes) and the public works department (which manages lighting).

Storm drain grates with slots running parallel to the direction of travel are wheel traps for bicycles. A road bike tire is typically 23-28mm wide; a parallel grate slot is typically 25-30mm wide. When a cyclist's front wheel drops into one of these slots, the bike stops instantly and the rider is catapulted over the handlebars. In 2007, a man was paralyzed after his tire got caught in the metal grating of the Montlake Bridge in Seattle. These crashes happen regularly but are dramatically underreported because most result in road rash and bruises rather than hospital visits, and police reports rarely attribute the crash to infrastructure. The fix is well-known and simple: cross-hatched or perpendicular-bar grate designs that prevent tire entrapment have been available for decades, and the U.S. Bureau of Reclamation published a bicycle-safe grate inlet study identifying safe designs. New installations in most cities now use bike-safe designs. But cast iron storm drain grates have a functional lifespan of 50-75 years, and cities have no systematic program to proactively replace old parallel-slot grates before they reach end of life. Replacement only happens when a grate is damaged, or when a road is fully reconstructed. The Bicycle Coalition of Greater Philadelphia surveyed Center City and found dozens of unsafe inlet grates still in active use. The structural problem is that storm drain grates are managed by water and sewer departments, not transportation departments. The water department's mandate is stormwater management — they care about flow capacity, not bicycle safety. There is no cross-departmental process that flags parallel-slot grates for replacement based on cycling safety criteria. A grate that drains water perfectly well will stay in place for decades, even if it has already caused crashes, because no one in the responsible department has a reason to replace it.

Bike-share theft is not petty crime — it is an operational crisis that threatens the financial viability of entire systems. Philadelphia's Indego bike-share lost at least 550 bikes between 2021 and 2023, valued at more than $1.3 million. At peak, thieves stole 14% of the entire fleet in a single year — three times the budgeted loss rate. In 2024, the vandalism escalated: thieves physically broke bikes out of docking stations, rendering five stations so damaged they had to be removed entirely. Indego's general manager called it an "unprecedented level of vandalism" that posed an "existential" threat to the system. The downstream effects compound. When stations are removed or emptied by theft, coverage gaps appear in the network, making the system less useful for remaining riders. Ridership in affected neighborhoods drops. The system becomes less equitable because the stations most targeted tend to be in lower-income areas where alternative transportation options are already scarce. Insurance premiums rise. Maintenance budgets get diverted from routine service to emergency repairs. Bikes that should be available for paying customers are instead circulating on the black market or stripped for parts. The problem persists because bike-share docking stations were designed for convenience, not security. The locking mechanisms are standardized and well-known, making them vulnerable to simple tools. GPS trackers add cost and weight, and thieves quickly learn to disable them. Law enforcement treats bike theft as a low-priority property crime — stolen bike-share bikes are rarely recovered through police action. Indego eventually hired private security — essentially repo men — to recover stolen bikes, which added yet another operating cost. The structural issue is that bike-share operators bear 100% of the theft cost while receiving zero help from the criminal justice system, and the economics only work if theft stays below 5% of fleet per year. Above that threshold, the business model breaks.

In cities across the United States, delivery drivers routinely park in bike lanes to make drop-offs. When a cyclist encounters a blocked bike lane, they must merge into moving traffic to go around the obstruction — exactly the scenario bike lanes are designed to prevent. This is not an occasional nuisance; it is the default condition of urban bike lanes. Studies show that short-term commercial loading is the most common source of bike lane obstructions, and the typical obstruction lasts less than five minutes. The real problem is that enforcement policies are designed to accommodate the obstruction rather than prevent it. The Philadelphia Parking Authority, for example, gives someone blocking a bike lane with posted "No Parking" signage a 25-minute grace period before issuing a ticket, and will never issue a ticket if the person is in the vehicle and agrees to move. Since most obstructions last under five minutes, this policy means effectively zero enforcement. Even in cities with stricter rules on paper, enforcement requires a parking officer to physically witness the violation, identify the vehicle, and issue a ticket — a process so labor-intensive that it cannot scale to the thousands of daily obstructions across a city. Chicago's Smart Streets pilot, launched in late 2024, represents the first serious attempt to automate enforcement: AI-powered cameras on CTA buses and at fixed locations automatically detect and ticket vehicles parked in bike and bus lanes. The program generated 11,723 warnings and 1,620 citations in its first months. But it only covers a small downtown area, and delivery companies have lobbied aggressively for exemptions, arguing that their business model depends on curbside access. The structural problem is that cities have designed their curb management around the assumption that motor vehicles have priority access to the curb, and bike lanes are painted on whatever space is left over. Until cities redesign curb allocation to give loading zones and bike lanes separate, non-overlapping space, delivery vehicles will keep parking in bike lanes because it is the rational, low-cost choice.

Delivery workers in New York City — the roughly 65,000 app-based couriers for DoorDash, Uber Eats, and Grubhub — rely on e-bikes with lithium-ion batteries that need daily charging. Most of these workers live in shared apartments or rooming houses where they charge batteries overnight, stacking multiple batteries on power strips in cramped rooms. When a lithium-ion battery with a damaged cell enters thermal runaway, it becomes a blowtorch that cannot be extinguished with water. Since 2022, these fires have killed 30 people in New York City and injured over 400, making e-bike batteries the leading cause of fatal fires in the city. The core issue is not that lithium-ion batteries are inherently dangerous — UL-certified batteries from reputable manufacturers rarely catch fire. The issue is that delivery workers, who earn $10-15 per hour after expenses, cannot afford $800-1,200 certified replacement batteries. Instead, they buy $200-300 uncertified batteries from street vendors or online marketplaces, often with cells salvaged from other devices, no battery management systems, and no thermal protection. These batteries are the ones that catch fire. New York City launched a battery-swapping pilot in 2025 with 25 outdoor cabinets in high-delivery-traffic neighborhoods, and a trade-in program where workers can exchange uncertified e-bikes for certified ones. But the scale is woefully inadequate: 25 cabinets for 65,000 delivery workers means roughly one cabinet per 2,600 workers. The structural reason this gap exists is jurisdictional fragmentation — the app companies that profit from delivery do not own the bikes, do not employ the workers (who are classified as independent contractors), and bear no liability for the batteries. The city is left to solve a fire safety crisis created by a business model that externalizes equipment costs onto the lowest-paid workers in the gig economy.

Every morning, bike-share riders flow from residential neighborhoods into downtown cores. Every evening, they flow back. This tidal pattern means that by 9 AM, downtown stations are full (no docks to return bikes) and residential stations are empty (no bikes to ride). The system operator must physically truck bikes from full stations to empty ones, and then reverse the process in the afternoon. This operation — called rebalancing — is the single largest cost in running a bike-share system. In Arlington, Virginia, rebalancing accounted for approximately 80% of total operating costs for the dock-based bike-share system. Operators must maintain fleets of trucks, hire drivers, and run logistics operations that essentially duplicate the transportation service the bikes are supposed to replace. For dockless systems, the problem is even worse: without fixed stations, bikes accumulate in random clusters, creating sidewalk clutter that generates public complaints and regulatory fines, while other areas have zero bikes available. In Toronto, 83% of Bike Share Toronto's fleet moves into the downtown core during the morning rush, creating bottlenecks where users wait up to 20 minutes to dock their bikes. This problem persists because bike-share pricing does not reflect the cost of rebalancing. A $4 ride generates the same revenue whether it travels with or against the tidal flow, but the ride that goes with the flow costs the operator $10-15 in rebalancing labor to undo. No major bike-share system has implemented directional pricing (cheaper to ride against the flow, more expensive to ride with it) because operators fear it would confuse users and reduce ridership. So the cross-subsidy continues, and rebalancing trucks burn diesel to move bikes that humans just moved under their own power.

Cities install protected bike lanes — separated from traffic by physical barriers like bollards, planters, or parked cars — and celebrate them as the gold standard of cycling safety. But the protection ends exactly where it matters most: at intersections. When a cyclist riding in a protected lane reaches a cross street, the physical barrier disappears, and they enter an unprotected conflict zone where right-turning vehicles cross their path. This is called the "right hook," and it accounts for 60% of all bicycle-motor vehicle crashes at intersections. The Insurance Institute for Highway Safety (IIHS) found that cyclists in protected bike lanes collide with vehicles most often at intersections or junctions with driveways and alleys. Urban intersections account for one-third of all crashes between bicyclists and motor vehicles and 43% of bicyclist fatalities in the United States. A 2025 study on intersection treatments found that only 30.7% of drivers yield to cyclists at lateral-shift intersection designs, meaning nearly 70% of drivers do not yield when they are supposed to. This is not a marginal risk — it is the primary way cyclists in protected lanes get killed. The problem persists because intersection treatments are expensive, space-constrained, and politically contentious. A fully protected intersection — with corner refuge islands, setback crossings, and separate signal phases — requires 15-20 feet of additional right-of-way at each corner, which usually means removing parking, narrowing vehicle lanes, or acquiring private land. Most cities lack the budget, political will, or engineering staff to redesign every intersection along a protected corridor. So they build the easy mid-block protection and leave the intersections unprotected, creating a false sense of security that may actually increase risk by encouraging cyclists to ride faster into conflict zones they assume are safe.

In most U.S. cities, conventional bike lanes are painted directly adjacent to parked cars, placing cyclists squarely in the "door zone" — the 3-to-4-foot arc where a car door swings open. When a driver or passenger opens their door without checking for cyclists, the cyclist either slams into the door at speed or swerves into moving traffic to avoid it. This is called "dooring." Dooring accounts for 19.7% of all reported bicycle crashes in Florida, 16% in Santa Barbara, and 5% in Boston, with over 800 dooring incidents reported annually in New York City alone. In February 2024, a 64-year-old Citi Bike rider was killed on Broadway near Lorimer Street in Williamsburg when a driver opened a car door into his path, sending him into traffic where a second vehicle struck him. At least 24 cyclists have been confirmed killed by dooring nationwide, though the real number is almost certainly far higher because most dooring crashes go unreported or are miscategorized in police reports. The reason this persists is that door-zone bike lanes are cheap. Painting a stripe on existing asphalt costs a city roughly $5,000-$15,000 per mile, while a physically protected lane costs $100,000-$500,000 per mile. Cities count these painted lines toward their "miles of bike infrastructure" metrics for federal grants and political credit, even though the design itself creates the hazard. The AASHTO guide has recommended a minimum 5-foot buffer between parked cars and bike lanes since 2012, but most cities grandfather in existing designs because retrofitting would require removing parking spaces — which triggers fierce political opposition from merchants and residents. So the painted line stays, and cyclists keep getting doored. Structurally, the root cause is that bike lane design standards are advisory, not mandatory. There is no federal requirement that bike lanes be built outside the door zone. Cities can and do build door-zone lanes and still receive federal transportation funding. Until funding is conditional on meeting minimum safety geometry, the cheapest design will keep winning.

Courts routinely order people on probation to complete substance abuse treatment, anger management, cognitive behavioral therapy, parenting classes, or other mandated programs. These programs are overwhelmingly offered during weekday business hours at approved provider locations, which are often only in the county where the person was sentenced. A person working a 7 AM-4 PM warehouse shift who is ordered to attend a Tuesday/Thursday 10 AM substance abuse group must either miss work twice a week or miss class and face a violation. There is rarely an evening or weekend option. If the person lives in a rural county, the nearest approved provider may be 30-60 miles away, adding hours of unpaid travel time and gas costs to every session. The cascading consequences are devastating. Missing a mandatory class is a technical violation. Getting fired for missing work means you cannot pay supervision fees, which is another violation. Losing income means you cannot pay for the class itself — which costs $25-$75 per session and is rarely covered by the court. You now have three potential violations from a single structural conflict that was built into your conditions at sentencing. The Pew Charitable Trusts found that probation conditions 'meant to support behavior change can burden more than benefit,' with treatment and programming requirements identified as among the most common sources of compliance failure — not because people refuse to participate, but because the logistics make participation impossible for anyone with a full-time hourly job. This problem exists because the treatment provider ecosystem was not built to serve people on probation. It was built to serve people with insurance and flexible schedules. Courts order treatment without verifying that it is available at times and locations compatible with the person's work and transportation situation. Providers have no obligation to offer evening or weekend sessions — their contracts with courts guarantee a steady stream of mandated clients regardless of scheduling convenience. And the system has no feedback loop: the judge who ordered the treatment never learns that the only available slot conflicted with the person's shift, and the officer who files the violation report for nonattendance does not have the authority to approve an alternative schedule. The person in the middle has no power to change any of it.

A Pew Charitable Trusts analysis of data from Oregon and South Carolina found that among people who were on probation for one year without being arrested, more than 90% could have spent less time on supervision with no impact on recidivism. Had supervision been limited to the minimum term needed to reduce reoffending, the average probation length in South Carolina would have dropped from 26 to 18 months and in Oregon from 24 to 14 months. The highest risk of reoffending occurs in the first few months; after that, additional supervision time produces diminishing returns that approach zero. Yet people routinely serve probation terms of 3, 5, 10, or even 20+ years. Some states impose lifetime supervision for certain offenses. Every additional month of unnecessary supervision is another month where a person can accumulate a technical violation for a missed appointment, a failed drug test, or an unpaid fee — and be sent to prison for it. Long terms do not make communities safer; they make supervision failure statistically inevitable. A person who successfully completes 36 months of probation but misses one check-in in month 37 goes to prison, wiping out three years of compliance. Meanwhile, that person occupied a slot on an officer's caseload for three unnecessary years, taking time and attention away from people in the high-risk early months who actually needed intensive supervision. The taxpayer cost is enormous: 3.6 million people are on probation or parole in the U.S., and each one costs $3,000-$5,000/year to supervise. This persists because probation length is set by statute and judicial norms, not by evidence. Legislators set long maximum terms to appear tough on crime. Judges impose long terms as a hedge against uncertainty — 'better safe than sorry.' There is no systematic mechanism for early discharge when someone is doing well. Some states have earned compliance credits, but they are bureaucratically cumbersome and officers rarely initiate the process because reducing caseloads is not incentivized — department budgets are tied to supervision population. The evidence base for shorter terms is robust, but the political incentive structure rewards length over effectiveness.

In states like Georgia, Tennessee, Alabama, and Mississippi, courts contract with private, for-profit companies to supervise people convicted of misdemeanors — often offenses as minor as traffic violations, public intoxication, or shoplifting under $50. The person is placed on probation not because they are dangerous but because they could not pay their fine at sentencing. The private company then charges $30-$60/month in supervision fees, plus enrollment fees, drug test fees ($15-$35 each), and late payment penalties. A person who owed a $200 traffic fine can accumulate $1,000+ in probation fees over the course of a year-long supervision term. Senator Elizabeth Warren's 2024 investigation found that private probation companies generate hundreds of millions of dollars annually from this model. The cruelty is structural: the less money you have, the longer you stay on probation, and the more you pay. If you could have paid the $200 fine on the day of sentencing, you would have walked away. Because you could not, you were placed on probation, which costs you $60/month, which you also cannot afford, which extends your probation, which generates more fees. People have been documented paying 5-10x their original fine in probation fees alone. The companies have been caught threatening jail time for people who cannot pay, failing to inform people about community service alternatives to fees, and continuing to collect fees after probation terms have legally expired. This persists because private probation is a textbook regulatory capture problem. The companies lobby state legislatures to maintain the offender-funded model. Courts use private probation because it costs the government nothing — all costs are borne by the supervised individual. Judges rarely see the downstream fee accumulation because the company handles everything after sentencing. The people trapped in the system are disproportionately Black (1 in 23 Black Americans is under some form of supervision vs. 1 in 81 white Americans), politically voiceless, and too poor to hire attorneys to challenge the fees. Human Rights Watch, the ACLU, and now Congressional Democrats have documented these abuses, but the industry remains entrenched because it serves the financial interests of both the companies and the cash-strapped courts that contract with them.

Standard immunoassay urine tests used in probation drug testing are known to produce false positives from over-the-counter medications including ibuprofen, pseudoephedrine (Sudafed), and diphenhydramine (Benadryl), as well as poppy seeds, certain B vitamins, and even some hand sanitizers. Hair follicle tests — increasingly used because they detect substances over a longer window — have documented racial bias: dark hair binds more drug metabolites due to melanin content, producing higher readings for Black and Latino individuals independent of actual drug use. A 2024 investigative report in Georgia found that parents were losing custody of their children based on hair follicle false positives from environmental contamination. When a person on probation tests positive, the process is not 'innocent until proven guilty.' The officer files a violation report. The person may be detained immediately pending a revocation hearing. Confirmatory GC-MS testing — which can distinguish a false positive from actual drug use — costs $100-$200 and is not automatically ordered in most jurisdictions. The probationer must request it, often pay for it themselves, and wait days or weeks for results while sitting in jail. If they cannot afford the confirmatory test, the initial false positive stands as evidence of a violation. A person who took Advil for a headache can end up in prison because they could not afford a $150 lab test to prove they did not use drugs. This problem persists because the probation system treats drug testing as infallible compliance verification rather than as an imperfect screening tool. Officers are trained to enforce conditions, not to evaluate the scientific reliability of test results. The testing companies have no liability for false positives — they sell the cheapest possible test because volume matters more than accuracy in a system processing millions of tests per year. Courts defer to 'positive test result' as objective evidence without examining the error rates of the specific assay used. And the fundamental asymmetry remains: the person with the least resources bears the entire burden of disproving a flawed test.

If you are on probation in Texas and get a job offer in Arizona, you cannot just move. The Interstate Compact for Adult Offender Supervision (ICAOS) requires your home state to submit a formal transfer request, the receiving state to conduct a 45-day investigation, and both states to agree before you can legally relocate. During those 45-60 days, you must remain in your current jurisdiction, report to your current officer, and hope the receiving state accepts you. If Arizona rejects the transfer — which states can do for any reason, including caseload capacity — you lose the job and remain trapped in a state where you may have no employment, no housing, and no family support. This matters because employment is the single strongest predictor of successful reentry. When someone on probation gets a legitimate job offer in another state and the bureaucracy takes two months to process paperwork, the employer hires someone else. The person remains unemployed, cannot pay their supervision fees, misses a payment, and gets a technical violation. The system that was supposed to help them reintegrate has now actively prevented them from doing the one thing most likely to keep them out of prison. For people whose families live in other states — a spouse who relocated, aging parents who need care — the restriction is not just economically devastating but personally cruel. This problem exists because ICAOS was designed to protect states from liability, not to serve the people being transferred. Each state's compact office operates independently with its own staffing levels and processing priorities. There is no expedited track for employment-based transfers. The 45-day investigation window exists because receiving states want to verify residence plans and assess risk, but the process was built for an era of paper files and postal mail and has never been modernized for urgency. The person on probation has no ability to appeal a rejected transfer, no right to expedited processing, and no recourse if the delay costs them the opportunity that would have changed their life.

When a judge sentences someone to probation, they typically impose a list of 'standard conditions' that apply uniformly to every probationer in the jurisdiction. The Prison Policy Initiative analyzed standard conditions across 76 jurisdictions and found that people are routinely given 10-20 conditions that have nothing to do with their specific offense. A person convicted of a financial crime gets the same drug testing requirements as someone convicted of drug possession. A person with no history of violence gets the same curfew as someone with assault charges. The conditions are not tailored — they are photocopied. The real damage is that these conditions frequently conflict with each other. You must maintain employment, but you also must attend three weekly treatment sessions that are only offered during business hours. You must not leave the county, but the only affordable housing you can find is in the next county. You must pay $100/month in supervision fees and $200/month for mandatory counseling, but you also must complete 100 hours of unpaid community service that takes time away from the job you need to earn the money. Each condition, viewed in isolation, seems reasonable. Stacked together, they create an impossible compliance puzzle where failing any single piece can send you to prison. People do not fail probation because they are dangerous — they fail because the conditions are designed as if they have unlimited time, money, and transportation. This persists because judges face no consequences for imposing excessive conditions — only for being seen as lenient. Standard condition lists are administrative conveniences inherited from decades of accretion: each time a high-profile failure occurs, a new condition gets added to the standard list, but conditions are never removed. Defense attorneys rarely challenge conditions at sentencing because the alternative is incarceration, so clients accept anything. And probation officers — not judges — end up with de facto authority to interpret vague conditions like 'maintain good behavior' or 'avoid persons of disreputable character,' giving them subjective power to revoke probation for conduct that is not criminal.

The recommended caseload for a probation or parole officer is 50 cases for general supervision and 20 for intensive or high-risk cases. In practice, officers routinely carry 65-75 active cases, and in some jurisdictions the number exceeds 100. The UK Probation Service is operating at 79% of its target staffing level, short 1,479 officers, and the proportion of high-risk cases that require qualified officers has nearly doubled from 12% to 22% between 2021 and 2024. In that same period, the service went from meeting 50% of its performance targets to meeting just 26%. When an officer has 75 cases, a monthly check-in lasts 10-15 minutes. There is no time to ask about housing stability, mental health, employment barriers, or family dynamics. The visit becomes a compliance audit: Did you pay your fees? Did you pass your drug test? Are you at the right address? This is not supervision — it is surveillance. The person on probation gets no help navigating the dozens of conditions imposed on them, and the officer has no capacity to provide it. When that person inevitably fails a condition, the officer files a violation report because that is the only tool available in a 10-minute interaction. The officer who wanted to help people is now just processing paperwork for re-incarceration. This persists because probation is chronically underfunded relative to its scope. The U.S. has 3.6 million people on probation or parole — more than the entire prison and jail population combined — but community supervision receives a fraction of corrections budgets. Hiring and retaining officers is difficult because the pay is low ($40,000-$55,000 in most states), the emotional toll is high, and burnout drives annual turnover rates above 25% in many departments. Officers report that paperwork — not client interaction — is their primary source of occupational stress, meaning the system demands documentation over human connection.