Eleven Claims
Tom Wright's public case against Penn Station through-running, examined claim by claim, with the strongest defensible reading of each argument tested against the documentary record.
The interview circulated on social media as a short explainer for general audiences, and the conclusion offered is that through-running is operationally unsuitable for Penn Station. The Regional Plan Association holds a multi-year advocacy contract with Amtrak under which the Build Gateway Now Coalition is managed; Amtrak appointed the RPA CEO to co-chair the Station Working Advisory Group; and RPA authored the April 2026 Penn Station through-running report that reaches conclusions consistent with the public case presented here.
Six Patterns Across the Eleven Claims
The interview relies on recurring analytical patterns: passenger imbalance is converted into train-capacity constraint, present-day demand is treated as fixed, and through-running is framed as a distant future option rather than an operating model to test against current capital plans.
Passenger ratio treated as train ratio
The central operational claim takes a real passenger imbalance and converts it into a train-capacity argument. The 10:1 figure is broadly accurate as a description of morning-peak rail passengers entering versus leaving the Manhattan CBD, and it may even understate the imbalance during the single peak hour. It does not follow that train movements are 10:1. Penn must clear arriving equipment regardless of load. Through-running’s capacity case rests on shorter platform occupancy, fewer terminal turnbacks, and converting necessary outbound movements into revenue service.
Demand treated as static
The geographic and demand claims rest on the assumption that current ridership patterns and employment locations are eternal. The argument that there are not many people wanting to travel between particular suburbs ignores induced demand from new connectivity, employment relocation enabled by improved transit access, and the long record of infrastructure shaping land use. The Elizabeth line generated millions of trips per month that did not previously exist as documented demand.
The straw-man of suburb-to-suburb travel
The interview reframes through-running's value proposition as primarily about one-seat rides between specific suburban origin-destination pairs, then dismisses the case on the basis that those pairs generate insufficient demand. Through-running advocates argue something different. The actual benefits are operational: reduced dwell time, increased frequency on each branch, eliminated turnaround conflicts at the throat, and capacity gains independent of any rider's particular trip. Through-running benefits accrue regardless of whether any given passenger ever rides through Penn.
Geographic exceptionalism as dismissal
A uniquely-New York argument is invoked to dismiss international comparisons. London works because of two things specific to London. The unstated implication is that no other global comparison applies. Paris, Berlin, Munich, Zurich, Tokyo, Stockholm, and Stuttgart all operate through-running regional rail systems with diverse spatial geographies. The London-is-special argument is not actually about London. It closes off the comparative frame.
The future-step deferral
“I’d love to do through-running as a future step.” The phrase converts agreement on principle into deferral in practice. Earlier Penn Station governance research described through-running as a potential capacity and regional-integration strategy constrained primarily by institutional barriers. Later capital plans preserve the same posture: through-running is acknowledged as possible, but placed behind terminal expansion, new tunnels, new governance arrangements, or an undefined future operating phase.
Monocentric framing hides secondary gravity
Midtown remains the dominant regional employment center, but the daytime-population evidence shows that the region is not a simple suburb-to-Manhattan funnel. Downtown Brooklyn, Long Island City, Newark, New Haven, White Plains, Stamford, and New Brunswick all show measurable daytime gravity. The through-running question is therefore not whether these places equal Midtown. They do not. The question is whether Penn Station operations should continue treating them as peripheral when they already function as job centers, reverse-commute destinations, or cross-core anchors.
The region is not Midtown-only.
Daytime population gives a practical test of Wright’s geographic claim. It estimates where people are during the workday by combining resident population, workplace workers, and resident workers. The result does not show a region of equal centers. It shows a dominant Manhattan core surrounded by secondary hubs with different planning roles.
Destination-strength hubs
Downtown Brooklyn, Long Island City, Newark, New Haven, White Plains, Stamford, and New Brunswick show measurable daytime job gravity. These places are the strongest candidates for reverse-commute and cross-core service testing.
Origin-heavy hubs
Jamaica and Paterson are large and important, but their daytime-population signal points more toward origins, transfers, and mixed-use activity than pure job importing. That distinction improves the argument because it avoids treating all secondary hubs alike.
Planning implication
The appropriate public test is not whether every suburb-to-suburb pair has large current demand. It is whether through-running improves access to existing and latent job markets across the regional rail network.
The Eleven Claims, Examined
Each claim is presented with its public wording, the operating premise it implies, the strongest defensible version of the argument, and an assessment against the documentary record.
"The demand in our region is so heavily peaked, in the morning there are 10 people coming in for every one person going out. And in the evening it reverses and most of the demand is going out and not as much as coming in."
Penn's commute pattern is so directionally asymmetric that any service model premised on bidirectional passenger flow becomes uneconomic.
As a passenger-count claim, the 10:1 figure is defensible. NYMTC's 2019 Hub Bound data shows severe rail-only directional asymmetry into the Manhattan CBD. During the 8:00–9:00 AM peak hour, the New Jersey rail sector carried 27,223 inbound passengers against 2,490 outbound passengers, or 10.9:1. The Queens sector, driven primarily by LIRR, measured 33,540 inbound against 1,840 outbound, or 18.2:1. The 60th Street sector measured 35,826 inbound against 2,035 outbound, or 17.6:1. Across all suburban and intercity rail sectors, the combined peak-hour ratio was 15.2:1. Across the broader 7:00–10:00 AM peak period, the rail-only ratio was approximately 11.3:1.
The numerical claim should not be dismissed as invented or exaggerated. It is broadly accurate for rail passenger volumes entering versus leaving the Manhattan CBD during the morning peak, and it is corroborated by Penn Station platform activity. The 2022 Penn Station Master Plan Alternatives Report recorded 60,820 AM peak-hour alighting passengers at Penn Station against 4,610 boarding passengers, a 13.2:1 alighting-to-boarding ratio.
The operational problem is the inference drawn from the statistic. The 10:1 figure describes people, not trains. In the New Jersey sector during the same 8:00–9:00 AM hour, 27,223 inbound passengers were carried on 25 trains, while 2,490 outbound passengers were carried on 12 trains. The passenger ratio was roughly 11:1, but the train ratio was roughly 2:1. The difference matters because through-running pairs physical trainsets, not equivalent passenger loads.
Passenger asymmetry does not mean trains can remain at Penn. Equipment must clear the platform whether it leaves full, half-full, lightly loaded, or empty. The through-running question is therefore not whether outbound reverse-peak demand equals inbound peak demand. It is whether necessary outbound train movements can be made faster, less conflict-prone, and more useful by carrying passengers instead of turning back or deadheading.
NYMTC Hub Bound Travel Data Report 2019 documents the rail-sector inbound and outbound passenger counts by Manhattan CBD cordon. MTA Penn Station Master Plan Alternatives Report, April 2022 documents Penn Station AM peak-hour alighting and boarding volumes. October 2024 Penn Station Capacity Expansion Feasibility Study confirms that roughly 12 New Jersey commuter trains already turn back at Penn Station to provide reverse-peak revenue service during the AM peak hour.
"Through-running requires you to kind of equalize those flows. Every train that comes in has to become a train going out during the peak hour."
Through-running structurally requires symmetric inbound and outbound trains, which makes it incompatible with directionally peaked demand.
In any continuous through-running operation, a trainset that arrives must continue out the other end. Over a peak hour the count of inbound and outbound trains through a station must roughly match. This is structurally true and is the condition that distinguishes through-running from terminal operation.
The premise is correct but the conclusion drawn from it does not follow. Equipment must move out of a terminal station regardless of operating model. Trains do not vanish at the bumper block. Under terminal operations, an inbound trainset is reversed and dispatched outbound either in revenue service or as a deadhead. The question is not whether the trainset moves outbound but whether that movement carries passengers and serves through-demand or moves empty as a positioning move.
Through-running converts a movement that is operationally necessary into a movement that is also revenue-productive. The requirement Wright identifies as a constraint is in fact a description of what already happens at Penn under terminal operation, except that today the outbound move is burdened by terminal turnaround time, throat conflict, and dwell penalty.
NYMTC Hub Bound Travel Data Report 2019 shows that passenger imbalance and train imbalance are different metrics. October 2024 Penn Station Capacity Expansion Feasibility Study documents that approximately 12 commuter trains from New Jersey already turn back at Penn Station to provide reverse-peak revenue service during the AM peak hour. RPA's April 2026 report, "New York Penn Station Constraints and Considerations for Meeting Future Demand" uses dwell and platform assumptions that should be tested against through-running-specific operating scenarios.
"That train going out, it's basically going to be a nearly empty train going out and it might be replacing a full train that would have been coming in."
Reverse-peak through-running trains will run empty and will displace fully loaded inbound trains, producing net capacity loss.
If tunnel capacity were the binding constraint and every reverse-peak slot crowded out an inbound slot, then an empty reverse-peak train would represent a real opportunity cost relative to a fully loaded inbound train using the same tunnel slot. The thought experiment makes physical sense as stated.
The argument takes a real passenger imbalance and applies it to the wrong operating variable. Under-river tunnel capacity, measured in trains per hour through the North River Tunnels, is one constraint. Penn Station platform and throat capacity, measured by how quickly trains can be received, unloaded, loaded, and dispatched, is another. Through-running does not make two Hudson River tunnel tracks into four. It changes how productively Penn's platform time and outbound equipment movements are used.
The empty-reverse-train framing assumes one-for-one substitution at the tunnel level. That is not how the current Penn operating problem works. Arriving trains must be cleared from the platform. Some already turn back in reverse-peak revenue service, while others move to yards or storage locations. A through-running plan asks whether those physical train movements can continue forward with shorter dwell and fewer terminal conflicts.
The 10:1 passenger imbalance may actually simplify the platform exchange in a through-running model. If roughly 1,100 people alight and roughly 200 board, the dominant process is deboarding. The lightly loaded outbound movement does not require the same platform time as a balanced 1,100-off / 1,100-on exchange. Under that operating logic, asymmetry is not automatically a capacity penalty. It can be a condition that enables rapid clearance.
NYMTC Hub Bound Travel Data Report 2019 shows that the peak-hour passenger ratio and train ratio diverge sharply: roughly 10.9:1 by passenger count in the New Jersey sector, but roughly 2:1 by train count. October 2024 Penn Station Capacity Expansion Feasibility Study documents existing reverse-peak turnbacks from New Jersey. MTA Penn Station Master Plan Alternatives Report, April 2022 documents the alighting-heavy platform condition at Penn Station.
"It actually reduces capacity that way. ... The simple truth is that through-running here it might actually reduce capacity in the overall system because you're running empty trains in the off peak direction that are replacing full trains that would have been coming in."
The net effect of through-running on Penn Station system capacity is negative.
If one accepted both the equalized-flow premise and the empty-reverse-train premise as binding, the conclusion of net capacity loss would follow. The argument is internally consistent given its assumptions.
This conclusion depends on converting a true passenger-ratio observation into a train-capacity conclusion. The passenger imbalance is real. The train imbalance is much smaller. During the 8:00–9:00 AM New Jersey sector peak, the underlying rail data shows a passenger ratio of roughly 10.9:1 but a train ratio of roughly 2:1. That distinction is enough to prevent the 10:1 statistic from carrying the operational conclusion by itself.
The relevant question is not whether reverse-peak trains carry fewer people. They do. The relevant question is whether those trains, which must clear Penn in some form, occupy platforms longer under through-running than under terminal operation. RPA and the railroads argue that Penn's narrow platforms and stairways force sequential deboarding and boarding, producing long dwell times. Through-running advocates argue that the small reverse-peak boarding load can occur within the deboarding window if platforms, vertical circulation, dispatching, and service rules are designed around rapid station work.
The net-capacity conclusion therefore remains assumption-dependent. If the model assumes long terminal-style dwell, existing platform constraints, current agency boundaries, and little operating reform, through-running will perform poorly. If the model treats the outbound leg as a necessary train movement with minimal boarding load and short platform occupancy, the same 10:1 imbalance can support rather than defeat the operating case.
NYMTC Hub Bound Travel Data Report 2019 provides the passenger and train-count basis for separating demand asymmetry from train movement asymmetry. MTA Penn Station Master Plan Alternatives Report, April 2022 documents the alighting-heavy AM peak condition at Penn Station. October 2024 Penn Station Capacity Expansion Feasibility Study and RPA's April 2026 report, "New York Penn Station Constraints and Considerations for Meeting Future Demand" provide the competing dwell-time and station-capacity claims that require public testing.
"The Elizabeth line brilliantly creates through running and you can connect all these places. Well London has kind of two things. One is it has more central business districts and places where you want to go. Two, they're all on strung along one line."
London works because of geographic conditions specific to London, which therefore do not generalize to New York.
London does have multiple commercial centers and the Elizabeth line does serve the City of London, the West End, and Canary Wharf along a roughly east-west alignment. To that limited extent the geographic observation is accurate, and a city's spatial structure does shape what kind of regional rail service generates ridership.
The framing is selective. Through-running regional rail is not a London phenomenon. Paris RER. Berlin S-Bahn. Munich S-Bahn. Zurich S-Bahn. Stockholm pendeltåg. Tokyo's through-running JR and private-rail operations. Stuttgart S-Bahn. These systems serve cities with diverse spatial geographies, including monocentric, polycentric, linear, and clustered urban forms. The claim that through-running depends on a London-specific geography requires erasing every other city that operates the same service model.
The east-west alignment of the Elizabeth line was not a found feature of London's geography. It was a deliberate routing decision made over decades of planning to connect the commercial centers. The rail infrastructure shaped the geography it served as much as it conformed to it.
London also operates Thameslink, a north-south through-running line, on a different corridor entirely. The two-things framing collapses to one example and treats it as the general case.
Published comparative transit literature on Paris RER, Berlin S-Bahn, Munich S-Bahn, Zurich, Stockholm, Tokyo, and Stuttgart through-running operations. Heywood's 2016 MIT thesis, Multi-Scale Regional Transportation Governance, notes that several worldwide examples of through-running exist and identifies Paris Châtelet-Les Halles as a successful high-volume comparison point for Penn Station. The Thameslink record also shows a long-running north-south London through-service built around capacity, frequency, and station-throughput objectives rather than one narrow spatial template.
"In New York City it's really Midtown Manhattan. ... [Ariel] It's always Midtown and maybe downtown and that's that. [Wright] And that's about it."
New York is a monocentric region with employment concentrated in Midtown and modestly in Lower Manhattan, with no other significant employment destinations.
Midtown Manhattan is by a wide margin the largest employment concentration in the region. Lower Manhattan is the second. No other cluster in the metro area approaches them in scale. To the extent that where commuters need to go is dominated by these two locations, the monocentric description captures most of the AM-peak inbound flow.
The dismissive “and that's about it” reads as rhetorical erasure rather than empirical description. Long Island City has become a major employment and mixed-use center. Newark has a large downtown worker base and direct regional rail position. Downtown Brooklyn is a major office, civic, educational, and medical district. Jersey City and Hoboken host major financial-services and technology employment. Stamford, White Plains, New Haven, and New Brunswick are regional employment centers in their own right. These places do not equal Midtown. They do not need to equal Midtown to matter.
The daytime-population evidence makes the monocentric claim too crude. The Manhattan core remains the benchmark, but Downtown Brooklyn, Long Island City, Newark, New Haven, White Plains, Stamford, and New Brunswick all show measurable daytime gravity. Some are major job importers; others are balanced mixed-use centers; others, including Jamaica and Paterson, function more as origins or transfer-rich urban nodes. That variation matters. It shows that the region is not a flat field of suburbs feeding a single center, but it also avoids the opposite error of pretending every hub has the same role.
The argument is also circular. Through-running enables the polycentric employment growth the dismissal claims is impossible. Reverse-commute connectivity is itself a precondition for office markets outside Manhattan. Without it, Newark stays smaller than it would be, Long Island City stays smaller than it would be, and the monocentric pattern reproduces itself. With it, the geography that the rebuttal requires becomes economically possible.
The Sunnyside Yard housing-and-transit proposal, which would create a major new employment-and-residential cluster directly served by through-running infrastructure, is itself evidence against the monocentric claim. The same planning world promotes development of new centers while allowing the rail network that would serve them to be dismissed because those centers are not yet Midtown-scale.
Bureau of Labor Statistics employment data for Long Island City, Newark, Jersey City, Hoboken, and Downtown Brooklyn. NYC Department of City Planning employment density studies. Sunnyside Yard housing-and-transit planning documents. NY–NJ–CT daytime population atlas estimates daytime population as resident population plus workplace workers minus resident workers, distinguishing the Manhattan benchmark from secondary hubs with net daytime gain, balanced mixed-use hubs, and origin-heavy urban nodes. Heywood's 2016 MIT thesis, Multi-Scale Regional Transportation Governance, frames through-running as a strategy that could change commuting sheds for Northern New Jersey and Long Island and open new business markets and economic development opportunities. October 2024 WSP/FXC feasibility study Section 5.2.1.2 identifies western Queens turnback infrastructure relevant to the Long Island City and Sunnyside Yard corridor.
"The few commuters who actually are, say, interested in going from New Brunswick to Ronkonkoma say. First of all there's not a ton of people. There's not a lot of demand for that."
The number of commuters traveling from one outer suburban node to another outer suburban node on the opposite side of Manhattan is too small to justify reorganizing the rail network.
Direct origin-destination flows between any specific pair of New Jersey and Long Island municipalities are indeed small. New Brunswick to Ronkonkoma is not a major commuting flow today and would not become a primary corridor under any plausible operational model.
The example is a straw-man of through-running's actual value proposition. Through-running is not primarily justified by direct suburb-to-suburb flows. The benefits are operational: reduced dwell at Penn, higher frequency on each line, eliminated terminal turnaround, and capacity gains that accrue regardless of any particular passenger's origin and destination.
The daytime-population evidence also shows why the New Brunswick-to-Ronkonkoma example is analytically weak. A through-running test should not be built around one deliberately thin origin-destination pair. It should test a market basket of cross-core and reverse-commute access changes: New Jersey to Long Island City and Downtown Brooklyn; Long Island and Queens to Newark; Manhattan residents to Stamford, New Brunswick, White Plains, and New Haven; and same-corridor trips that become more frequent once terminal dwell and turnback conflicts are reduced. The relevant planning unit is the regional accessibility field, not a single comic-distance pair.
To the extent through-running does generate new origin-destination demand, the relevant flows are reverse commutes between Manhattan and outer-borough or New Jersey employment centers, same-side trips that pass through Manhattan en route to a destination further along the corridor, and induced demand from employment relocations that improved connectivity makes economically feasible. The cherry-picked extreme of New Brunswick to Ronkonkoma functions as an argument by minimization, not analysis of the actual demand patterns through-running would serve.
Even on the cherry-picked example, demand is endogenous to supply in transit. The Elizabeth line generated trips that did not previously exist as documented demand. Asking what current ridership is between a pair of nodes in a network that does not connect them produces a small number, not a meaningful measure of latent demand.
Elizabeth line ridership growth data from Transport for London. Origin-destination studies for Newark-Manhattan, Long Island City-Manhattan, and other reverse-commute flows. NY–NJ–CT daytime population atlas identifies a market basket of secondary employment and mixed-use hubs more appropriate for testing through-running demand than any single suburban pair. ReThink Studio Regional Unified Network Vol. 1, pages 18-19 outlines the demand-following-supply argument with budget figures of $29.4 billion for the full RUN scope and $1.1 billion for Penn Station-only operational change.
"They don't care if it's a one seat ride to any of the other branches. They only care about the line that takes them to where they want to go."
Riders value a one-seat ride only to their specific destination, not generic network connectivity, and through-running's main appeal is generic connectivity.
Riders do prefer one-seat rides to their actual destination over multi-transfer trips. Generic connectivity to many destinations is a weaker preference than direct service to the one destination a particular commuter cares about. The behavioral observation is correct.
The behavioral observation is correct and the inference drawn from it is wrong. Through-running advocates do not primarily promise generic connectivity to many destinations. They promise higher frequency on each line, faster trips overall, reduced dwell, and capacity that supports more service to each rider's actual destination. The riders of the New Brunswick branch under through-running operations get more trains per hour to New Brunswick, not a useless connection to Ronkonkoma.
The argument substitutes a weak version of the through-running case for the actual case, then dismisses the weak version. That version of the argument does not engage with the operational benefits that form the core through-running case.
Daytime-population evidence strengthens the correction. The relevant rider benefit is not a one-seat ride from every branch to every other branch. It is improved accessibility to a set of real employment and activity centers: Midtown, Lower Manhattan, Downtown Brooklyn, Long Island City, Newark, Stamford, White Plains, New Haven, New Brunswick, and other rail-adjacent hubs. That is an accessibility test, not a slogan about one-seat rides.
October 2024 WSP/FXC feasibility study documenting frequency and capacity gains under through-running operations. ReThink Studio RUN proposal articulating per-line frequency improvements as the primary rider-facing benefit. NY–NJ–CT daytime population atlas identifying secondary hubs appropriate for a regional accessibility test.
"That service is never going to be that attractive to you. ... And the service is just never going to be very frequent."
Through-running services at Penn cannot achieve frequencies that would make them competitively attractive to riders.
If through-running combinations were poorly designed, with mismatched line pairings or poorly utilized fleet rotation, they could indeed produce lower per-line frequency than independent terminal operations. Operational design matters and a bad design could produce a worse outcome than the status quo.
The claim treats frequency as a property of through-running rather than as an operational design choice. Properly designed through-running increases frequency on each branch. Reduced platform dwell allows more trains per platform per hour. Reduced fleet idle time at terminal layovers allows more revenue trips per day from the same equipment. Eliminated terminal-throat conflicts at Penn allow more conflict-free movements across the diamond.
The frequency claim is unsupported by any analysis the public can review. RPA's April 2026 report does not present a frequency comparison between a through-running operating model and the current terminal model. Wright's assertion that frequency is never going to be very frequent is forecasting a result based on an analysis no one has shown.
The 2024 WSP/FXC feasibility study reaches the opposite conclusion. The frequency benefit is one of the core operational gains the study documents.
October 2024 WSP/FXC feasibility study Section 5.2.1.2. RPA's April 2026 report, "New York Penn Station Constraints and Considerations for Meeting Future Demand" absence of a frequency-comparison analysis.
"What are you more likely to do? I'm going to take the first express train into New York City. I'd like to have a beautiful Penn station where I can make an easy transfer. And then I will take the next train out to Ronkonkoma."
A reconstructed and beautified Penn Station with easy transfers is a sufficient response to the network connectivity case for through-running.
A high-quality Penn Station with well-designed transfer concourses, clear wayfinding, and pleasant pedestrian environment is genuinely valuable. Many passengers will continue to make terminating trips and benefit from a better station experience. Aesthetic and functional improvements to the existing terminal are not in themselves a failure of policy.
This claim recasts an operational question as a station-experience question. The choice presented is narrower than the policy question. It is between operational change that delivers capacity benefits within years at roughly one billion dollars for the Penn-only scope, and a physical reconstruction that delivers an improved waiting environment at fourteen-and-a-half to nearly seventeen billion dollars while preserving terminal-only operation. The beautiful-Penn framing trades the operational case for a station-experience case, then accepts an order-of-magnitude cost difference as if the comparison were neutral.
The April 2022 Penn Station Master Plan Alternatives Report describes a southern expansion that is being designed to permit eventual through-running by means of new tunnels constructed at some future date. The expansion preserves terminal-only operation in the interim. The temporal horizon for future tunnel construction has been estimated by some analysts at the 2080s. This is what beautiful Penn with easy transfer is being proposed in lieu of. It is not merely a station upgrade. It is a multi-decade deferral of operational change, financed at a price point that makes the deferral structurally hard to revisit once committed.
The transfer experience itself does not address the binding capacity constraints. A passenger with a smooth concourse transfer is still subject to the dwell penalty at the terminal, the throat conflict at the diamond, and the per-line frequency ceiling those constraints impose. The argument substitutes wayfinding and aesthetics for operational performance.
April 2022 Penn Station Master Plan Alternatives Report Section 6.1 documenting nine-track configuration with five tracks set lower for future Sunnyside extension, and total cost range of $14.9 to $16.8 billion. ReThink Studio RUN Vol. 1 pages 18-19 documenting Penn-only scope at $1.1 billion. 2015 Amtrak SLD Penn South memo documenting that the southern expansion preserves stub-end terminal capacity at 19 to 22 TPH.
"I'd love to do through-running as a future step. But the higher priority day has got to be creating more capacity under the Hudson River."
Through-running is desirable in principle, and should be considered after the Hudson tunnel capacity expansion is complete.
Hudson tunnel capacity is a real and binding constraint. The Gateway Program's new tunnels add genuine inbound and outbound capacity that no operational change can replicate. Sequencing infrastructure investment is a legitimate policy question and there is room for reasonable disagreement about the order in which capacity-relevant projects should proceed.
The framing presents a false choice. Hudson tunnels and through-running are not substitutes. They are complementary. New tunnels expand under-river capacity. Through-running maximizes the productivity of platform infrastructure, both existing and new. The two interventions answer different constraints and a regional rail strategy that needs both does not have to choose.
The Gateway tunnels are funded and proceeding regardless. The decision in front of the Gateway Development Commission and the master developer procurement is not whether to build new tunnels. It is whether the southern expansion of Penn at fourteen-and-a-half to nearly seventeen billion dollars is the best use of public funds when through-running operational change at roughly one billion dollars for the Penn-only scope would deliver capacity benefits within years and shape how new Gateway tunnels are operated when they open.
The future-step framing has a long public record. Heywood's 2016 Penn Station governance thesis described through-running as a potential capacity and regional-integration strategy, already demonstrated at small scale by the Meadowlands Train to the Game, while identifying institutional barriers as the central obstacle. The 2026 RPA report treats through-running as a future possibility. The southern expansion currently being procured by Amtrak under the master developer P3 treats through-running as a future operating model the new station will eventually accommodate. Each iteration preserves the idea while allowing major capital commitments to proceed without near-term operating change.
The cost of perpetually deferring operational change is itself enormous. Climate considerations make additional decades of automobile-dependent commute patterns increasingly costly. Demographic and employment shifts are reshaping the region in ways that operational change could accommodate sooner if undertaken now. Technology is advancing in ways that could change what counts as feasible. Promising the operational change in some indefinite future and committing the funding to physical infrastructure today is the structural choice through-running advocates have been objecting to for decades.
Heywood, Rebecca J. (2016), Multi-Scale Regional Transportation Governance: Evaluating Cooperation and Decision-making at New York Penn Station, identifying through-running as a potential Penn Station capacity and regional-integration strategy, noting the Meadowlands Train to the Game as a small-scale precedent, and locating the major barriers in institutional cooperation, labor, politics, and governance. April 2022 Penn Station Master Plan Alternatives Report Section 6.1 identifying through-running as a future operating state the southern expansion would accommodate. RPA's April 2026 report, "New York Penn Station Constraints and Considerations for Meeting Future Demand". ReThink Studio RUN Vol. 1 pages 18-19 Penn-only scope at $1.1 billion. NY–NJ–CT daytime population atlas identifying secondary employment and activity centers that should be included in any origin-destination, accessibility, transfer-penalty, and latent-demand test.