If GPS switched off worldwide for a week, your phone map would be the smallest problem. GPS is above all a clock: its satellites carry atomic clocks that time-stamp mobile networks, power grids and payments, and as backup clocks drift, those systems degrade over hours to days. Official studies put the cost at about $1 billion a day in the US and £7.6 billion for a week in the UK. A rider would still get home, by sun, Polaris, a compass, paper maps and asking at dhabas, with errors of a few kilometres per hundred.
- 31 satellites in GPS were healthy in April 2026, orbiting 20,180 km up.
- 1 microsecond of clock error is 300 m of position error, which is why GPS is a timing system first.
- £7,644 million is the estimated UK loss from a 7-day outage; 87.6% falls on emergency services, road and shipping.
- NavIC can't step in: with 3 working satellites in July 2026, India's own system was below the 4 needed for positioning.
- 4 minutes of clock error equals 1° of longitude, about 108 km at Bengaluru's latitude.
Calculated in this post from the Earth turning 15° of longitude per hour (Wikipedia: Longitude), so 4 minutes is 1°, and 111.32 km per degree at the equator (same source) times cos 13°, Bengaluru's latitude (Wikipedia: Bengaluru).
This is a tier-2 what-if, so I go through it as a system failure: what GPS really provides, what fails in the first hour, day and week, how much it costs, and then the old-school navigation a rider would fall back on, with the error maths for each method.
What does GPS actually do besides maps?
It broadcasts time. Each satellite carries atomic clocks (caesium and rubidium standards) and sends its time and position; a receiver measures how long each signal took to arrive and solves for its own place and the exact time. Four satellites give four unknowns: x, y, z and the receiver's clock error.
The speed of light makes timing everything:
So one microsecond wrong is 300 m wrong. Relativity makes the satellite clocks, seen from the ground, run about 38 µs a day fast; left uncorrected, that alone is \( 38\times10^{-6} \times c \approx 11 \) km of error per day. GPS is designed to correct it.

Has a GPS failure happened before?
Yes, a timing one, for about 12 hours in 2016, and it showed how far GPS reaches beyond maps.
On 26 January 2016, operators retiring SVN 23, the oldest satellite in the GPS constellation, uploaded ground software with an error. Positioning kept working, but the UTC time message broadcast by GPS shifted by 13 microseconds. Timing receivers across Europe raised alarms; one UK timing firm reported the disruption lasting about 12 hours for those caught in the middle of it, and the effect moved round the globe as the affected satellites orbited. Telecom networks, energy and defence users were hit, and it was not limited to one manufacturer's equipment. The incident revived calls for a terrestrial backup such as eLoran, at a time when only the Anthorn station in the UK was still broadcasting in Europe.
Inside GNSS: GPS glitch caused outages, fueled arguments for backup (2016)Thirteen microseconds is 3.9 km of equivalent distance error. Nobody's map moved, but clocks that trusted GPS were suddenly wrong by an amount telecom networks notice.
Calculated in this post from the 13 microsecond shift (Inside GNSS, 2016) times the speed of light, 299,792,458 m/s (Wikipedia).
What breaks first in a week without GPS?
Navigation breaks at once; timing breaks slowly. Systems that need time keep running on a local backup clock, called holdover, until it drifts too far.
Holdover maths
A clock with fractional frequency error \( y \) drifts by \( \Delta t = y \times T \) over time \( T \). Over one day (86,400 s), with these example inputs:
- Ordinary quartz, \( y \approx 10^{-6} \): about 86 milliseconds a day.
- Good oven-controlled quartz, \( y \approx 10^{-10} \): about 8.6 µs a day.
- Rubidium, \( y \approx 10^{-11} \): about 0.9 µs a day.
These \( y \) values are typical orders of magnitude I'm using for illustration, not specs. The point: a network that needs microsecond timing runs out of holdover in hours to days, not weeks.
| Time after switch-off | What fails | What still works |
|---|---|---|
| Minutes | Phone maps, ride-hailing, delivery apps, fleet tracking, drone navigation | Mobile calls, payments, power |
| Hours | Ships and planes fall back to radar, radio aids and inertial systems; surveying stops | Most timing, on holdover |
| 1 day | Systems without good holdover start to degrade; the UK study assumes good holdover covers 24 hours | Rubidium-backed networks |
| 1 week | Losses pile up in emergency response, road freight and shipping | Aviation, which has non-satellite backups |
See the data as a table
| 7-day loss (£ million) | |
|---|---|
| Emergency services | 3533.9 |
| Road | 1659.9 |
| Maritime | 1500.2 |
| Other sectors | 583.0 |
| Agriculture | 223.6 |
| Rail | 139.9 |
| Aviation | 4.0 |
Source: London Economics for the UK Government, The economic impact on the UK of a disruption to GNSS (2023) · Finance shows nil loss because its timing systems are assumed to have enough holdover.
The study models an instant, UK-wide, standalone loss of all satellite navigation as a reasonable worst case, drawing on public data and interviews with more than 20 stakeholders. It estimates GNSS benefits at £13,622 million a year, a 24-hour outage at £1,424 million, and a 7-day outage at £7,644 million. Emergency services (£3,534 m), road (£1,660 m) and maritime (£1,500 m) take 87.6% of the weekly loss; aviation loses only £4 m because aircraft carry backups, and finance shows nil because of holdover. Resilient backup positioning and timing could cut the loss by almost half. The limit: one country, pre-2020 data, and the authors warn the results may have a short shelf life.
Read the summary (free)RTI estimated the benefits of GPS across 10 US sectors, including telecoms, electricity, financial services, agriculture and location services, at about $1.4 trillion (2017 dollars) since civilian use began in the 1980s. A widespread outage would cost about $1 billion a day, and up to 50% more if it lasted weeks during the April–May planting season, when farmers rely on precision guidance. The limit: NIST itself says to read these as rough orders of magnitude, because many sectors have never been without GPS to measure the difference.
Read the summary (free)Could India switch to NavIC?
Not in 2026. India's own regional system, NavIC, needs at least four working satellites for positioning; in a written Lok Sabha reply reported on 30 July 2026, the government said only three were providing services (IRNSS-1B, IRNSS-1I and NVS-01), after IRNSS-1F reached the end of its mission life in 2026.
- Timing: still available from NavIC with three satellites.
- Positioning: not available as a standalone service.
- Other systems: Galileo, GLONASS and BeiDou are separate constellations, so a GPS-only failure would leave multi-system phones working. This post imagines GPS alone gone; a total GNSS loss is the UK study's case.
How would a rider navigate without GPS?
The way riders did before smartphones: paper, compass, sun, stars, odometer and people. Each has an error you can calculate.
Latitude from Polaris
Polaris sits within about a degree of the celestial pole, so its altitude above the northern horizon equals your latitude. In Bengaluru (13°N) it is only about 13° up, low and often lost in haze; in Leh (34°N) it is high and easy.
- Measure. A fist at arm's length covers roughly 10°; a cheap inclinometer app or a protractor and string does better.
- Convert. One degree of latitude is about 111 km, so a 1° reading error is about 111 km north or south.
Calculated in this post from Earth's radius of 6,371 km (NASA): 2π × 6,371 ÷ 360 ≈ 111 km per degree.
Longitude from the clock
The Earth turns 360° in 24 h, so 15° an hour, or 1° every 4 minutes. Local solar noon (shortest shadow) compared with a watch set to IST gives your longitude.
Calculated in this post from 15° of longitude per hour (Wikipedia: Longitude): 360° ÷ 24 h = 15° an hour, so 1° every 4 minutes.
- Width of a degree. \[ w = 111.32 \cos\varphi \ \text{km} = 111.32 \cos 13^\circ = 108.5\ \text{km} \]
- Clock error. A watch 1 minute off is 0.25° off: \( 0.25 \times 108.5 = 27 \) km.
- Practical limit. Judging the exact moment of the shortest shadow by eye is hard, so expect tens of kilometres. Fine for "am I near Nagpur", useless for a turn-off.
Dead reckoning: compass plus odometer
Heading error \( \alpha \) over distance \( d \) gives a sideways error of \( d \sin\alpha \). With a 5° compass error, every 100 km puts you \( 100 \sin 5^\circ = 8.7 \) km off line. On roads this barely matters, because the road corrects you; it matters off-road and in fog.

Use this to check what you just read: how far off would you be?
Navigation error without GPS
The rider's no-GPS kit
- Paper maps: a state road atlas and a printed list of towns and highway numbers for the day.
- Compass: mounted away from the steel and magnets.
- Odometer and trip meter: reset at every junction you're sure of.
- A watch set to IST: for solar-noon longitude.
- Milestones: India's highway stones give distance to the next town; they are the best GPS-free navigation aid in the country, a true analogue backup.
- People: fuel-pump staff and dhaba owners know the road better than any app.

How sensitive is the damage to the outage length?
Strongly, and not linearly. The UK numbers show the first day costs £1,424 million, while the average day of a week costs about £1,092 million, because some losses are front-loaded and some systems adapt; but beyond the holdover window, timing failures start to stack up.
Calculated in this post from the UK study's 7-day loss of £7,644 million (London Economics, 2023) divided by 7 days: about £1,092 million a day.
- One hour: a nuisance; most holdover clocks don't notice.
- One day: the UK study's £1.4 billion; navigation-heavy services hit hardest.
- One week: £7.6 billion in the UK; in the US, about $7 billion at RTI's $1 billion a day.
- Planting season: up to 50% worse in the US, per RTI, because tractors steer by satellite.
Verdict: what happens if GPS goes dark for a week?
A week-long GPS outage is a realistic scenario: a 2016 software error already shifted GPS time by 13 µs for about 12 hours. The cost is large but bounded, around $1 billion a day in the US and £7.6 billion a week in the UK, because aircraft carry backups, many receivers use Galileo and BeiDou too, and good holdover clocks bridge the first day. For a rider it is entirely survivable: Polaris for latitude within a degree, solar noon for longitude within tens of kilometres, a compass for heading, and milestones and people for the rest.
GPS is a clock that happens to tell you where you are. Turn it off and the clocks notice before the riders do.From the working above
What's new
- Only three NavIC satellites active; government admits standalone navigation gap Business Today, Jul 2026
- DGCA orders GPS spoofing reports within 10 minutes after Delhi airport incidents The Tribune, Nov 2025
- Perseverance gets a GPS-like way to fix its own position on Mars NASA JPL, Feb 2026
Planning without an app: fuel range planning, the solo touring safety plan with check-ins that don't need tracking, and reading mountain weather. Sapta's own trackers and mounts are listed on the gear page.
What if next?
- What if you rode non-stop until the bike or you gave up? Without a map, the first thing to fail is the rider's judgement.
- What if the monsoon never came one year? The satellites that would watch the drought unfold.
- What if you rode through a monsoon on Titan? A moon with no GPS, no sun to steer by, and methane rain.
Questions people ask
What would happen if GPS stopped working?
Navigation apps would fail at once; timing-dependent systems such as mobile networks and grids would run on backup clocks and degrade over hours to days. RTI estimated about $1 billion a day of losses in the US.
Does the banking system depend on GPS?
Many financial systems use GPS time for time-stamping. The UK study estimated nil finance losses over a week because those systems are assumed to have enough holdover and backup timing.
Can I navigate by the stars on a motorcycle?
Roughly. Polaris's height gives your latitude to about a degree, and solar noon against an accurate watch gives longitude to within tens of kilometres. Paper maps and road signs do the fine work.
Is NavIC a backup for GPS in India?
Not fully in 2026. With three working satellites in July 2026 it could provide timing but not standalone positioning, which needs at least four.
Has GPS ever failed?
On 26 January 2016, a ground software error during the retirement of satellite SVN 23 shifted GPS time by 13 microseconds for about 12 hours, triggering alarms in telecom and energy systems.
Sources
- London Economics for UK Government: The economic impact on the UK of a disruption to GNSS (2023; £7,644 m for 7 days, £1,424 m for 24 h, sector table, assumptions)
- NIST: Economic Benefits of the Global Positioning System to the U.S. Private Sector (RTI) (2019; $1.4 trillion, $1 billion a day, 50% higher in planting season)
- Inside GNSS: GPS glitch caused outages, fueled arguments for backup (2016; SVN 23, 13 µs, about 12 hours)
- Wikipedia: Global Positioning System (31 healthy satellites April 2026, 20,180 km orbit, 38 µs a day relativistic offset)
- Business Today: Only three satellites left active, NavIC standalone navigation gap (Jul 2026)
- The Tribune: DGCA asks pilots and controllers to report GPS spoofing within 10 minutes (Nov 2025; 465 incidents Nov 2023 to Feb 2025)
- NASA JPL: Perseverance now autonomously pinpoints its location on Mars (Feb 2026)
