One behavioral parameter is missing from a chain that transportation engineers have already built two-thirds of. It costs roughly $40,000 to produce, and it unlocks between $1.2B and $8.3B in annually recoverable loss.
THE PROBLEM
When a lane closes in congested traffic, the efficient behavior is for both lanes to stay full to the taper and then alternate — the zipper merge. Minnesota DOT field data shows roughly 40% shorter backups. One Michigan work zone went from six miles of congestion to three, returning 15–25 minutes to every driver in the queue. Almost nobody does it, and the reason is not ignorance. It is an assurance failure. Most drivers would privately prefer the zipper because it is faster for them. They believe other drivers regard late merging as cheating. So they merge early, and a minority actively block the open lane — a behavior that manufactures the very hazard it claims to police. The blocking makes late merging genuinely unsafe, which manufactures the evidence for the belief that produced it. The drivers enforcing the norm are generating the behavior they are punishing. Every public campaign to date has been built to inform drivers. In one MoDOT simulator study, over 60% of subjects had never heard of the zipper merge, and when shown current signage only 42% read it correctly — 26% said they would simply follow the cars around them. An information campaign aimed at an assurance problem fails by construction.
WHAT IS ALREADY KNOWN, AND WHAT IS NOT
The economic chain is standard and nearly complete:
Attributable burden = (A) congested lane-closure exposure × (B) throughput loss per unit of non-compliance × (C) delay, fuel and crash cost per unit of throughput loss
- Term A is computable today. State lane-closure permits combined with commercial probe data.
- Term C is computable today. EPA MOVES emissions modeling and Texas Transportation Institute 2024 valuations of $24.01 per person-hour and $80.16 per truck-hour.
- Term B does not exist. It requires an observed compliance distribution — how far from the taper drivers actually merge, and how often the open lane is deliberately obstructed. We have found no published measurement of this distribution. The gap is structural rather than technical. Transportation research funds facility-level decisions — should this taper get a dynamic sign? — so the existing literature measures treatments at treated sites. The best example, a Michigan dynamic late-merge evaluation, found a 15.7% emissions reduction during congested peak over an 11-week deployment. That is a device evaluation, not a population parameter. No agency has a line item for the economic burden of a driver belief, so nobody has produced one.
WHAT THIS STUDY PRODUCES
- Term B, measured. Dashcam and inertial observation of merge distance, lane utilization ratio, and obstruction incidence across active I-35 Capital Express work zones — the corridor whose core segment (I-35, US 290N to SH 71) TTI’s 2024 Top 100 edition (2023 conditions) ranked the #1 truck-congested segment in Texas and #3 overall; in the 2025 edition (2024 conditions) it ranks #5 for trucks and #9 overall.
- The Merge Gap. A survey instrument measuring private preference against perceived public preference. Prediction published in advance with its falsification condition; if the gap is under 15 points the mechanism is wrong and the null result is published.
- A reusable parameter. Term B plugs into every simulation model already in the literature. One measurement, retroactively multiplied across an entire existing field. THE MERGE GAP · AUX LABS LLC PAG E 2 O F 2
WHAT FULL COMPLIANCE IS WORTH, ANNUALLY
RECOVERABLE PER YEAR LOW MID HIGH
I-35 Austin corridor $1.7M $5.6M $11.4M
United States $1.2B $4.1B $8.3B
US lives / year 18 52 106
US serious injuries / year 1,138 3,412 6,882
Fuel not burned (gallons / yr) 27.2M 91.8M 187M
CO2 avoided (tonnes / yr) 242k 816k 1.66M
Each figure is a published anchor multiplied by a stated assumption, bracketed rather than point-estimated. Anchors: TTI Texas Top 100, 2025 edition (2024 conditions), I-35 Austin segment US 290N–SH 71: annual congestion cost $206,497,330 (truck-only $40,243,425) and annual delay of 913,926 person-hours per mile — approximately 7.2M person-hours across the 7.93-mile segment, a derived figure (per-mile × length), not a printed one; TTI Urban Mobility Report, Austin urban area: 21,134 thousand gallons (≈21.1M) of annual excess fuel consumed (2024; 20,372 in 2023); FHWA work-zone fatality counts (~875/yr nationally, Texas consistently highest of any state); EPA 19.6 lbs CO2 per gallon. Assumptions: work zones = 10–25% of corridor delay; merge behavior = 8–22% of that; merge-related share of work-zone fatalities = 8–22%, of which 25–55% avertable. The fatality band is the least certain, because crash records code “unsafe lane change” and stop. Establishing that attribution is part of what this study produces. A statewide tier is deferred to a future revision, pending a published statewide anchor.
WHY THIS PROBLEM IS UNUSUALLY TRACTABLE
• No opposition. No party on earth has a financial • Falsifiable in months. A registered numerical
interest in early merging. There is no counter-lobby prediction that resolves against field data.
and no incumbent who loses. • Guidance already exists. Agencies have published
• Two-thirds pre-built. The engineering literature the correct behavior for years. Nothing needs to be
supplies everything except the behavioral term. legislated — only assured.
THE ASK
$40,000 · six months Deliverables: Term B measured and published open; the Merge Gap instrument and dataset; a peer-reviewed methodology paper; and a segmentation model specifying which message moves which driver population.
Larger partnerships — claims-linked crash attribution, multi-corridor deployment, or signage specification work — are available to discuss, but are not required for this phase to produce its result.
WHO CARRIES THE LOSS TODAY
Insurers hold work-zone rear-end and sideswipe claims, and hold the data that would close crash attribution. State transportation agencies are scored on corridor performance and work-zone fatalities. Road contractors lose crew members at the taper. Freight carriers absorb delay valued at more than three times person-time. Autonomous vehicle operators face a deployment constraint: systems tuned to the efficient merge inherit the same social penalty human late-mergers face. Signage and ITS vendors supply the equipment any remedy requires.
HOW THIS IS BEING FUNDED, AND WHY WE ARE PUBLISHING IT
This proposal is going out on the same day, in identical form, to four kinds of funder: a government research program, a philanthropic grantmaker, a corporate risk partner, and a founder-discretionary source. The protocol is published in advance. We are recording time-to-decision and cost-of-entry at each door, coding every delay by cause — statutory, procedural, capacity, or discretionary. No outcome is predicted. The intended finding is a replicable instrument that any researcher can run on their own project. The underlying question is one that shapes whether problems of this kind get solved at all: who will pay to fix a diffuse public good, and how quickly? We hope the answer is that public, philanthropic, and private capital move together. If it turns out that a single discretionary funder is dramatically faster, that is worth knowing precisely, and worth publishing. Every counterparty is informed of this at first contact.
Imran Hafiz — Founder, Aux Labs LLC · Austin, TX · imran@auxlabs.co Duke University, Sanford School of Public Policy. Seven years at VICE / VIRTUE, most recently Publisher of VIRTUE Intelligence. Co-author of the peer-reviewed Jin-Hafiz Disinformation Index (Jin et al., Cureus, 2021; PMID 34211810). Aux Labs LLC is a research lab drawing on mechanism design, signaling theory, and behavioral science.