At roughly 0405 UTC on 2 May 2000 — a few minutes past midnight, Eastern time — the accuracy of every civilian satellite navigation receiver on the planet improved by roughly a factor of ten. An error term maintained in the Global Positioning System’s ground segment was set to zero, and the change propagated across the entire satellite constellation over an interval of about a minute or two.[12]
Some four million civilian receivers were in service worldwide at that moment, supporting a market for GPS goods and services then valued at about $8 billion.[33] Every one of those receivers got better overnight without a firmware update, a service call, or an invoice.
It’s rare that someone can press a button and make something you already own worth more. Clinton administration fact sheet, 1 May 2000[33]
That is the pivot on which this entire history turns, and its shape is worth noticing before going back for the explanation. The most consequential moment in the commercialization of satellite navigation was not an invention. It was the withdrawal of a deliberate impairment — a policy artifact, rendered in software, sitting on top of an asset the American taxpayer had already paid for as an instrument of war.
Which raises the question that organizes everything before it: why was the civilian signal broken on purpose in the first place?
II · Three answers to a submarine’s question
A system that never hears from its customers can serve four of them or four billion
The system was born as a subsystem of nuclear targeting, and it is impossible to understand its later economics without understanding that first customer. A Polaris ballistic missile submarine cannot shoot accurately if it does not know precisely where it is. Its inertial navigation system drifts; its periscope star trackers cannot see through weather.
President Eisenhower personally pressed Admiral Arleigh Burke on this accuracy problem in 1957, and the answer arrived from an unexpected direction — two physicists at the Johns Hopkins Applied Physics Laboratory who, on the Monday after Sputnik launched, stuck a wire into a shortwave receiver and recorded what one of them called “an absolutely gorgeous Doppler shift.” Within a week, a colleague had inverted the logic: if you can find a satellite from a known ground station, you can find the ground station from a known satellite.[13]
Navy · Transit
Find the ground from the satellite
The world’s first satellite navigation system. Under Eisenhower, only ARPA was authorized to develop military satellite systems — so the Polaris office made the case that submarine position error was a missile accuracy problem, and ARPA paid.[13] A two-dimensional fix, available only when a satellite happened to pass, taking ten to sixteen minutes to compute, accurate to a quarter of a mile or so, and degraded by a fifth of a nautical mile for every knot of unknown velocity.[42]
The cheapest way to fund a universal service is to describe it as necessary for a submarine carrying nuclear missiles.NRL · Timation · 31 May 1967
Put the clock in orbit
Roger Easton proposed measuring signal travel time instead of inferring position from Doppler on the ground. Timation I, an 85-pound satellite with a gravity-gradient boom, proved a passive user could determine range — and that precise time could be transferred through the broadcast itself.[14] Successive satellites carried quartz, then rubidium, then cesium; by NTS-2, Easton’s team had verified the relativistic offset correction every satellite still applies.[14]
Timing, not position, is the true primitive. The user is silent.Air Force · Program 621B
Four ranges, one frequency
A classified 1966 system study — not declassified until 1979, six years after the decisions it shaped — laid out four simultaneous ranging measurements to solve for four dimensions at once.[42][45] Spread-spectrum pseudorandom codes let every satellite broadcast on the same center frequency and made continuous three-dimensional fixes possible for fast-moving aircraft and, candidly, for munitions.
A receiver that transmits nothing can be multiplied without multiplying cost.Lonely Halls · Labor Day weekend 1973
Three services, three answers, three budgets. In August 1973 the program went before the Defense System Acquisition Review Council and failed — Black Thursday. Over the weekend that followed, about a dozen officers and engineers met in a nearly deserted Pentagon and synthesized the three approaches into one: four-satellite passive ranging, the Air Force’s code-division signal structure, space-hardened atomic clocks, and a constellation of 24 satellites in twelve-hour orbits.[42][44][45] In December 1973 the program was approved as NAVSTAR GPS, with a Phase One budget of roughly $150 million covering the initial satellites, launches, a master control station with six monitor stations, seven kinds of user equipment, and an eighteen-month test program at Yuma.[45]
Two details from that approval matter more than the money. The architecture contained two signals from the outset — an encrypted precise code for authorized military users and a coarse acquisition code available to everyone — and the program manager, Bradford Parkinson, testified to Congress that the receiver signal specification was being offered to the public. Students at Leeds locked a civil receiver onto the first satellite within twenty-four hours of first broadcast in 1978. The signal was not, however, guaranteed.[42][45]
The Air Force paid for a system whose most numerous beneficiaries would be Army infantry and Navy aviators, and it never much liked the arrangement. Parkinson likened it to asking the richest household on the block to fund the entire high school. As late as 1979 the Air Force was still trying to cancel development; civilian leadership in the Pentagon overruled the attempt.[43][45] The program office’s internal motto was a two-line summary of the whole seat-and-consumption bargain before anyone had a phrase for it: drop five bombs in the same hole, and build a cheap set that navigates.[43]
Nobody in that room could have justified the appropriation with “everyone on Earth will always know where they are.” Fewer bombs per target was a budget line. Ambient global positioning was not.
III · The corpse that opened the door
A promise, unlike an affordance, can be held against you
On 1 September 1983, Korean Air Lines Flight 007 — a Boeing 747 flying from New York to Seoul via Anchorage — drifted hundreds of kilometers off course on inertial navigation, crossed into Soviet airspace, and was destroyed by a Soviet interceptor. All 269 people aboard died, among them a sitting U.S. congressman. Reagan’s address four days later dwelt on the fact that the airliner had flown a straight-line course at 30,000 to 35,000 feet for two and a half hours, that the pilot had reported a position more than a hundred miles from where he actually was, and that only civilian airliners fly that way.[16]
Twelve days after the shootdown, the White House announced the policy that created civilian GPS as a commitment rather than an accident. The Deputy Press Secretary’s statement of 16 September 1983 declared that the President had determined the United States was “prepared to make available to civilian aircraft the facilities of its Global Positioning System when it becomes operational in 1988,” a system that would give civilian airliners three-dimensional positional information.[38] The Reagan Library files the press release under a plain title: Civilian Airlines to be Equipped with GPS.[37] Two decades later, the U.S. delegation to the United Nations would list the same event as a founding milestone of the civil framework: in 1983, Reagan offered free civilian access to help ensure aviation safety around the world.[20]
A coarse civil signal already existed in the architecture; the receiver specification had already been offered publicly a decade earlier.[42][45] KAL 007 did not invent civilian access. What it did was convert an engineering affordance into a presidential promise — and a promise, unlike an affordance, can be held against you. That distinction is what makes the following seventeen years dramatic rather than merely technical.
Specified permission · not the Erlanger measurement
The offer came with a governor
The precise code stayed encrypted for authorized users. The coarse code, the one being pledged to the world’s airlines, was subject to Selective Availability: an intentional degradation of the public signal, implemented globally through the satellites themselves, specified so that a civilian was guaranteed only to be somewhere within a hundred-meter radius of where the receiver said.[25][36]
Precision was not a physical property of the system. It was a permission, issued in two tiers. The threat model was pure Cold War: an adversary using American satellites to aim at America. The mitigation was to make every fishing boat, ambulance, surveyor and rental car on Earth a hundred meters vaguer. The Soviets, for their part, were building GLONASS and did not appear to be waiting for permission.
IV · How Selective Availability lost before it was switched off
A tariff on a good with no marginal cost
Consider what Selective Availability actually was in economic terms. The fixed costs of the system — satellites, atomic clocks, launches, master control station, worldwide monitor network, spectrum allocation, geodetic reference frame — were sunk, and sunk against a weapons appropriation. The marginal cost of an additional civilian user was, and remains, exactly zero, because the receiver is passive. Into that structure the government inserted an artificial quality ceiling in order to protect a monopoly on precision.
That is a tariff on a good with no marginal cost. And tariffs of that kind invite arbitrage.
-
01 · Promise versus governor
A public pledge the degradation quietly contradicts
Reagan’s 1983 offer made civilian access a commitment. Selective Availability left the pledged signal a hundred meters vague. A promise can be held against you; an engineering affordance cannot.[38][20]
-
02 · Desert Storm
The military became a customer of the channel it had blurred
Soldiers wrote a commercial manufacturer that desert navigation was “an absolute nightmare.” The program office awarded a contract to Trimble Navigation in seven days, with deliveries beginning in thirty, ultimately procuring thousands of small lightweight receivers built for the civil market.[17]
-
03 · Public corrections
One budget to inject the error, another to cancel it
The Coast Guard’s differential network delivered one-to-three-meter accuracy plus integrity monitoring. The FAA built WAAS; Transportation was still requesting $92 million a year for it in its fiscal 2026 request.[11][21] After May 2000, differential accuracy did not much change — corrections no longer needed to be broadcast as frequently, because a large, fast-varying component of what they had been cancelling was the deliberate error itself.[21]
-
04 · Other skies
Allies begin building their own constellations
A global fog protects a monopoly only while no one else turns on the lights. The architecture that could not meter users also could not stop them from listening to someone else’s satellites.
The formal machinery caught up in 1996, when the first U.S. GPS policy committed the government to providing the Standard Positioning Service worldwide, continuously, free of direct user fees — a commitment Congress wrote into law the following year — and set in motion the decision to zero Selective Availability, with discontinuation promised by 2006 and annual assessments beginning in 2000.[20][32]
The 2000 assessment came back with the answer that made the hinge inevitable, and it was not a moral answer. The Secretary of Defense recommended discontinuation. Threat assessments concluded that setting the error to zero “would have minimal impact on national security.” And the decisive sentence in Clinton’s statement is the one about substitution, not generosity: the United States had “demonstrated the capability to selectively deny GPS signals on a regional basis when our national security is threatened.”[32] The subsequent official line was even plainer — the military would pursue regional denial capabilities in lieu of global degradation.[19][21]
The Pentagon did not surrender control of precision. It swapped a crude global instrument for a targeted local one, and in doing so it discovered that the global instrument had been costing it more than it was worth. The most economically consequential deregulation in modern history required no rocket, no appropriation and no ceremony — a parameter, changed overnight, while the country slept.
V · The ledger
Nobody has ever paid for GPS. Everybody has paid for a way to listen.
Here is the whole argument in one glance. Every line above the receiver row is public and sunk. The only private line is the cheapest one. U.S. law and policy require the civil service to be provided free of direct user fees, and there are no plans to privatize it; Congress appropriated over $2 billion for the core program in a single recent fiscal year.[11]
An estimated $1.4 trillion in U.S. private-sector benefit since the 1980s.[2][5] A modern multi-constellation chip runs about $3.[45]
VI · The cascade
The value accrued at the top of the stack
Everything after the hinge follows from the ledger. Once the fixed cost is public and sunk, and the signal is free, nobody downstream can defend a margin on access. The only defensible margin is on the application layer. That is why the map is free and the ride costs money.
Precision became sufficient without augmentation
The government’s own guidance after May 2000 drew the line explicitly: safety-critical navigation and sub-meter surveying still needed differential corrections, but a trucking company tracking and managing assets would now find the basic civil signal adequate on its own.[21] That sentence is the birth certificate of commercial fleet telematics.
Regulation, not the market, put receivers in pockets
The FCC’s Enhanced 911 requirements were arriving just as the degradation lifted. Removing Selective Availability boosted accuracy enough that GPS “could become the method of choice” for implementing the 911 location requirement, likely simpler and cheaper than radio tower triangulation.[33][35] The government sank the space segment as a weapons program, then legislated the consumption device into existence.
within 40 billionths of a second
Timing turned out to be the product
The same fact sheet noted that the time signal’s accuracy improved to within forty billionths of a second, and predicted adoption of GPS as the preferred means of acquiring Coordinated Universal Time for synchronizing electrical power grids, cellular towers, telecommunications networks and the Internet — far less costly than maintaining private atomic clocks.[33][35] When economists later tried to quantify the value of GPS to the American private sector, they examined ten sectors and found that finance, electricity and telecommunications sat alongside precision agriculture and location-based services; the enabling condition was not the map pin but a reliable, extremely precise timing signal.[2][5] NIST has since had to publish a formal evaluation of how far critical infrastructure timing systems now depend on it.[24]
about $1 billion a day
Machines, not people, became the primary consumers
In agriculture, GPS stopped being consulted and started steering: farm planning, field mapping, soil sampling, tractor guidance, variable-rate application, yield mapping — and the ability to work through rain, dust, fog and darkness, when a human operator cannot see the rows.[22] A GPS outage is estimated to cost the United States about $1 billion a day, and as much as 50 percent more than that if it lands during the April and May planting season.[5]
Of the roughly $1.4 trillion in U.S. private-sector benefit attributed to GPS since it became available for civil and commercial use, most has accrued in the most recent decade, on the back of miniaturization, commoditized devices and ubiquitous wireless service.[2][5]
VII · The vessel reforms
The seat is still military. The meter has returned by other hands.
In April 2026 the last satellite of the GPS III series went up on a Falcon 9 from Space Launch Complex 40 at Cape Canaveral, at 2:53 a.m. Eastern on the 21st — the fourth consecutive GPS mission reassigned from a grounded Vulcan to SpaceX, executed after a launch-provider pivot in under seven weeks.[8][10] Space Vehicle 10 completed a 32-satellite active constellation and carried an optical crosslink demonstration, a new Digital Rubidium Atomic Frequency Standard, a NASA laser retroreflector array, and the program’s first 3D-printed omni antenna.[8][26]
It also carried M-code, which gives the warfighter positioning three times more accurate and eight times more resistant to jamming than the legacy constellation; the follow-on GPS IIIF block will add Regional Military Protection — steerable beams delivering over sixty times the anti-jam capability to qualified military users.[8][26] Read that alongside the 2000 decision and the architecture of control becomes clear. The Pentagon never gave up asymmetry. It stopped degrading civilians and started privileging soldiers. Same asymmetry, better implementation.
Meanwhile the seat is exactly as governmental as it ever was, and considerably more geographically explicit than most of its six billion users would guess: a master control station at Schriever Space Force Base with an alternate at Vandenberg, four ground antennas, six Space Force monitor stations, eleven National Geospatial-Intelligence Agency monitor stations, and tracking sites at Thule, Kaena Point, Ascension, Diego Garcia, Kwajalein, Guam, Osan, Bahrain, and U.S. embassies in Ecuador and Uruguay.[7] The American taxpayer pays for all of it; the civil service remains free of direct user fees; there are no plans to privatize.[11]
The promise is still not finished
The legacy civil signal, L1 C/A, is the only one delivered as a fully operational civil signal. L2C is set healthy but formally pre-operational — “use at risk.” L5, the safety-of-life signal that aviation and maritime users have been designing around for well over a decade, is set unhealthy and pre-operational. Making it fully operational requires 24 orbital slots with L5 capability, which means the GPS IIIF satellites are essential.[7][23] That is the state of play as reported to the Civil GPS Service Interface Committee in April 2026. Six billion people depend daily on a service whose operator has never finished promising it.
The meter has returned
Imposed not by Washington but by whoever owns a transmitter. EASA has issued the fourth revision of its safety bulletin on GNSS outages and alterations, describing jamming and spoofing that has grown in severity and sophistication since February 2022 around conflict zones.[27] EASA and EUROCONTROL published a joint European action plan in February 2026, prompted by a letter from thirteen member states, to contain the threat for at least three years.[30] The International Chamber of Shipping produced a free poster for bridge officers and asked the IMO to circulate it.[28] In January 2026, fourteen North Sea and Baltic coastal states called on the maritime community to treat GNSS interference as a direct threat to safety.[29] A handheld jammer covering the civil frequency retails for about $33.[31]
So the Pentagon spent the 1990s ensuring that a Soviet targeteer could not use American satellites against America, and the 2020s discovering that a truck driver with a mail-order transmitter can degrade an international airport.
The remedy, characteristically, runs back through the ledger. When the Department of Transportation proposed repealing the National Timing Resilience and Security Act in its fiscal 2026 budget request, its reasoning was that no single backup can meet the diversity of critical-infrastructure requirements, and that for the federal government to procure or fund a specific solution would be “inefficient, anti-competitive, and potentially harmful to the existing market for back-up/complementary PNT services.”[11] The government that gave away the signal now declines to buy the fallback, on the grounds that doing so would distort the private market in fallbacks. Whatever else it is, that is a consistent position.
One last detail, offered without comment. The April 2026 briefing in which Mission Delta 31 reassured the world’s mariners and aviators about signal integrity, described its 6 billion customers, and explained why L5 is still not operational, records that the unit has partnered since September 2025 with multiple Department of War organizations, and closes with the line that the Department of War and the U.S. Space Force remain committed to delivering GPS modernization requirements.[7] The renaming began as an executive order in September 2025;[55] the Pentagon has since formally asked Congress to codify it at an estimated cost of about $52.5 million,[59] and the House attached the change to its version of the annual defense bill on a 216–212 vote.[56]
That is the finished shape of the thing. A weapons program, approved in December 1973 for about $150 million because fewer bombs per target was a defensible budget line, now flies 32 satellites at 20,200 kilometers, is tracked from eleven countries and two hemispheres, costs the taxpayer over $2 billion a year, is operated by a service whose department is in the middle of renaming itself for war, charges nobody anything, and is consumed daily by roughly six billion people through a component that costs about three dollars and draws milliwatts.[7][11][45] The seat is military and permanent. The meter is a chip. Between them sits the single most valuable subtraction any government has ever performed — and the reason almost nobody remembers it is that on the morning of 2 May 2000, the only thing that changed was that everything worked better, and no one was sent a bill.