Gujarat, Plainsnet (simulated), blue fibre
Town primary school, Kheda
A connection that just works. The provider is paid for every day, and the funder never has to chase anyone.
- Days lit
- 6
- Days returned
- 0
- Paid so far
- 72
Paid for being ready. Not for working.
India’s auditor checked BharatNet, the programme running fibre to every village council. Most connections were marked ready; a third of those were working.
- 96.7%
- of 226,000 village councils marked service ready
- 33.2%
- of those connections actually operational
- 17 days
- average to restore a fibre fault; cuts caused 48% of downtime
CAG Report No. 19 of 2026, performance audit of BharatNet, figures as of November 2025, as reported by The Policy Edge.
Nobody counts whether school internet works.
India’s school census records whether 1.47 million schools have electricity, water, toilets and computers, and whether each of them works. For internet it records only whether a connection exists.
The state-by-state pictureElectricity
1,377,945 have it, 1,352,061 work
Drinking water
1,461,032 have it, 1,456,968 work
Girls’ toilet
1,421,205 have it, 1,372,881 work
Computers
951,868 have it, 852,684 work
Internet
933,987 have it. Working: not recorded.
Out of 1,471,473 schools. Outline: schools that have the facility. Fill: schools where it works. Source: UDISE+ 2024-25, Ministry of Education, Table 2.5 (pp. 35–36).
Schools live with the same gap. A connection is paid for when it is installed, or when a month ends, and nobody in the school can say “it was down all week” in a way a payment system will hear. The provider is paid either way, so a cut cable can wait 17 days.
Roll Call makes the school’s own speed test the payment trigger. A day the school was online over the paid-for line earns that day’s fee. A day it was not earns nothing, and the money goes back to the funder. Fixing a fault fast becomes the cheapest thing a provider can do.
How one school day is decided
Three parties who need not trust each other. Only the contract moves money.
- 1
Step 1, by the school
The school measures
A probe in the school runs an M-Lab ndt7 test several times a day against the nearest M-Lab server: download, upload, round trip. Giga uses the same protocol to measure schools.
- 2
Step 2, by the school
The probe signs
A key only the school holds signs the numbers and the M-Lab test id. Change one digit and the signature breaks.
- 3
Step 3, by the attester
The line is identified
The attester looks up which network the test ran over, and refuses a reading sent from any other. A test over a phone hotspot earns the contracted provider nothing.
- 4
Step 4, by the contract
The contract pays
The day is split into parts, and a test has to pass in each. After a late window and a challenge window anyone can settle it: every part passed pays in full, some pays that share, none returns the day to the funder.
The latest real reading on chain, 1 h 49 min ago
Probe 0: a cloud server in India, day 10
ndt-nql67_1791477899_00000000002EC174|ndt-nql67_1791477899_00000000002ED143
- Download
- 50 Mbps
- Upload
- 51 Mbps
- Round trip
- 17 ms
- Network
- AS31898
- Result
- passed
Six schools, six things a real term goes through
Each simulated school follows a script built from what actually breaks school connections. The contract decides every mark you see on its strand.
Uttarakhand, Ridgeline Fibre (simulated), orange fibre
Hill primary school, Chamoli
A landslide cuts the fibre on day 8 and nobody comes to fix it. The probe stays on and signs “no connection” every school day. After five absent days in a row the funder ends the term and takes back every day not yet served, plus part of the provider’s bond.
- Days lit
- 7
- Days returned
- 3
- Paid so far
- 84
Assam, Brahmaputra Broadband (simulated), green fibre
River-island school, Majuli
Floods close the school twice; the headteacher marks those days closed before testing starts. Later a two-day power cut switches the probe off: nothing reports, the days are refunded, and none of it counts against the provider.
- Days lit
- 5
- Days returned
- 1
- Paid so far
- 56
Jharkhand, Sal Valley Net (simulated), brown fibre
Ashram school, Gumla
The link is up but slow. A day is split into three parts and a test has to pass in each; on most days only one or two parts reach 10 Mbps, so the provider is paid a third or two thirds.
- Days lit
- 5
- Days returned
- 1
- Paid so far
- 40
Odisha, Deltaline (simulated), slate fibre
Coastal school, Kendrapara
When the line drops, a teacher runs the test from a phone hotspot. The speed is fine, but it came over another network, so the contracted provider earns nothing for that day.
- Days lit
- 5
- Days returned
- 1
- Paid so far
- 60
Rajasthan, Thar Links (simulated), violet fibre
Government school, Barmer
On day 10 the readings look perfect, but their test ids are not in M-Lab's archive. The funder challenges the day, the arbiter upholds it, the day is voided and the provider loses part of its bond.
- Days lit
- 6
- Days returned
- 0
- Paid so far
- 72
Visitors who ran a real test
Each strand is a term somebody opened from Be the school: three two-minute days, contracted to their own network, real M-Lab tests.
Two real probes keep the real calendar
Probe 0 is a server in a data centre, the easy case: four real M-Lab tests a day. Probe 1 is a laptop on a home broadband line during school hours, which is off some days: its silent days show what a real school device does. Their days settle two days after they end, the challenge window a funder needs to check M-Lab’s archive, so their first marks appear on October 27.
PaidReturned
Run it at a real school today
Nothing above is a mock-up. A school, a funder and a provider can run a term on the same contract with their own keys.
- Step 1
The school starts a probe
Any computer that stays on in the school opens the probe kit. It makes a key, tests once in each part of the day, and signs offline reports when the line is down.
Open the probe kit - Step 2
The funder opens a term
With any browser wallet: the probe’s address, the provider, the arbiter, what a working day means, the money. Test dollars come from a built-in faucet.
Open a term - Step 3
The provider accepts
From its own wallet on the term page, posting its bond. From then on each day settles itself, and every term prints a statement for funders who pay in their own currency.
See every term
Who has to agree before money moves
No single party can pay itself. Each role is a separate key, and the contract refuses any term where one key holds two roles.
School probe
Sign what it measured, or that it could not get online. Mark a day closed before any test runs: holiday, flood, exams.
Cannot pay anyone, or vouch for which network it used.
Attester
Say which network a test ran over. Record readings the probe signed.
Cannot change a number the probe signed, or count one test twice.
Funder
Lock the term’s money. Challenge a finished day, staking a deposit it loses if wrong. End the term after repeated absent days.
Cannot take back a day the provider earned, unless the arbiter agrees.
Provider
Accept the terms and post a bond. Collect for every working day, the same week.
Cannot be paid for a day the probe did not see working.
Arbiter
Decide a challenge. Upheld voids the day and cuts the bond; silence lets it lapse.
Cannot move money any other way.
Anyone
Settle finished days. Read every reading. Check test ids against M-Lab’s archive.
Cannot move money any other way, or change a mark.
How school connections are paid for today
Roll Call does not replace the people who build networks. It changes what triggers the payment.
Milestone contracts
Pays when
A connection is installed and declared ready; monthly fees after that.
When the line is down
Usually nothing changes. An audit may find it years later.
Giga Connectivity Credits
Pays when
A provider earns credits for connecting a school, redeemable for subsidies or tax relief.
When the line is down
Rewards connecting. Quality is measured separately with Giga Meter.
Roll Call
Pays when
Each school day the school’s own M-Lab tests pass, over the contracted network.
When the line is down
That day’s money returns to the funder within the week. Repeated outages end the term and cut the bond.
Giga is the UNICEF and ITU initiative to connect every school. Its measurement tool, Giga Meter, runs the same M-Lab ndt7 protocol as Roll Call, so a small adapter on the school device could sign Giga Meter’s results with the school’s probe key instead of running a second test.
Take the register yourself.
Your browser becomes the school’s probe. The house opens a three-day term contracted to your network, you run real M-Lab tests, and you watch the contract light or cut each two-minute day. No wallet, no sign-up.