The science & precedent floor

Why a pool, and why now

AnTIE's design rules trace to a measured run on a live gateway, three national data-exchange precedents, an open-detector system, an agency inventory, a licence clause that rules out riding a foreign crowd, and a transit-data commons Malaysia already adopted — the evidence for an exchange Malaysia owns.

What the record actually says

One measured experiment, three proven national-exchange precedents, one detector network, one licence reality and one transit precedent Malaysia already set.

0.24→0.40

The density crossover, measured on the running system

Before asking anyone to join, we ran the exchange's central claim against the live fusion engine. One member's 3 pings left a trunk segment honest-grey (HTTP 404); a second member's 3 lit it to confidence 0.24; four members mixing slow readings raised it to 0.40. The gate counts total samples on the road — so pooling opens it with zero code change.

AnOmbak production gateway · localhost · 2026-11-03

X-Road

Neutral, co-owned data exchange is proven statecraft

Nations already run sovereign data exchanges between rivals and agencies: X-Road's open-source layer links tens of thousands of organisations; TN-ITS standardises authoritative road-data exchange between EU authorities (Germany ran its first service in Aug 2026); the UN's Transport Data Commons hosts open national datasets. 'Owned by all, held by none' is not idealism — it is a deployed pattern.

X-Road (Estonia) · TN-ITS · UN Transport Data Commons

40,000

A state can publish its own road data as open infrastructure

California streams real-time speeds from ~40,000 freeway detectors into a public system researchers and apps build on. It is the fixed-sensor model: authoritative, open, government-run. AnTIE's move is to pair that idea with the mobile crowd a static network never had — and to do it in Malaysia, on Malaysia's own data.

Caltrans PeMS · Performance Measurement System

0 APIs

Malaysia funds the sensors, and never the glue

The national ITS blueprint has prioritised real-time journey info for years; the Transport Management Center runs ~140 variable message signs and automated counters; KL and Putrajaya pay for AI signal control and adaptive junctions. Yet not one open, machine-readable, per-segment road-flow interface exists — the output of all that paid sensing ends at control-room screens. Inventory done, fusion missing.

Malaysian ITS Blueprint 2019–23 · TMC · DBKL/KKJ contracts

1-way

Riding a foreign crowd is a legal dead end

Waze's city programme is a one-directional barter under a non-negotiable public-authority agreement — you give them closures, they show your drivers their crowd, and you may not hand the feed on. Google's terms bar caching and derivative datasets. There is no lawful seam to build a national map on someone else's users. The only honest path is a pool Malaysia owns.

Waze for Cities · Google Maps Platform ToS

30 s

Malaysia already ran this play once — for buses

Public-transport data became an open commons the state adopted and every app got better on top of — positions published every thirty seconds under CC BY 4.0. That is the exact precedent AnTIE repeats for road flow: turn a paid, siloed asset into an open national data layer, and let differentiation live above it.

GTFS-Realtime · CC BY 4.0 open data

Finding → rule

So the exchange is built this way

from: 0.24→0.40

Pool before you paint

A thin market cannot colour its roads honestly, and no single app can out-market arithmetic. The gate counts samples on the road, not per app — so the only truthful way to turn grey roads live is to pool the crowds. The exchange is the pooling; the honesty gate stays exactly as strict.

from: X-Road / TN-ITS

Neutral exchange, not a privileged operator

Because competitors and agencies must both trust it, the referee is a co-owned entity with a deterministically reproducible reference implementation and a protected fork right — the X-Road / TN-ITS posture. Neutrality is a testable property (run it, diff the output), never a clause to take on faith.

from: 40,000 detectors / 0 APIs

Expose what the state already paid for

Agency cameras, adaptive signals and operator incidents enter fusion as authoritative indications the crowd verifies — Malaysia's ITS estate becomes day-one coverage on instrumented corridors. Improve existing infrastructure, literally: no new hardware, the missing product is the open per-segment layer.

from: 1-way Waze barter

Own the seam, rent no one's map

A foreign feed can be withdrawn, repriced, or simply not be passable on — Waze for Cities proves it. Sovereignty is architecture: the segment registry is open and Malaysian-owned, the fusion runs in-country, and any external licensed feed is a short-TTL scaffold the community is designed to outgrow.

from: GTFS commons

Open by default, share-alike forever

Following the GTFS precedent, fused outputs carry a share-alike national-data-commons licence and the registry stays open — so no member, including the founding contributor, can enclose what the crowd built. The licence is the reason a competitor signs up and the reason the government adopts it.

from: Charter Art. 4

Privacy welded into the protocol

Identities, trajectories and plates are rejected at the schema layer, not by policy; pseudonyms are salted per member and rotate; coordinates coarsen before storage; k ≥ 3 before anything publishes. The exchange could not re-identify a driver even under order — the inverse of the model that got the incumbent banned.

Why here, why now

The Malaysian numbers

~71%
of corridors live with five crowds pooled

Modelled at typical thin-app density on the production gate: 1 member colours 0.8% of roads, 5 members 71.5%, 8 members 96.2%. The shape — not the exact number — is the pitch; the real figure arrives when members go live.

140
message signs + counters the state already runs

Variable message signs and automated counters, plus AI signal control in KL and Putrajaya on eight-figure contracts. The hardware is funded; the exposure of its output as open per-segment road data is the missing product — and it is the exchange's, not another agency's.

0
open per-segment road-flow feeds today

Every fixed-sensor and every operator's control room ends at a screen. AnTIE's honest claim is not a running service — it is that the fusion maths, the gate and the measured crossover already exist, and only the pool around them is proposed.

30 s
live bus positions, already open in Malaysia

GTFS-Realtime vehicle positions published under CC BY 4.0 every thirty seconds are the in-market proof that Malaysia can and does run an open transport-data commons. Road flow is the same move, one layer down.

A rush hour, honestly described

6:15pm on the Sprint Highway: a stretch that would be grey in any one app is amber now, because four thin crowds pooled enough samples to clear a gate none of them could on their own — and the confidence rose, not fell, as they did. Upstream a line of brakes forms a pattern the wave model admits across members, so a hazard is flagged before the queue shows; a driver from a different app confirms it, and that cross-company agreement is worth more than a dozen in-app tips. An agency camera's closure rides in as an authority indication the crowd verifies. No identity crossed the wire, no trajectory was stored, and a road nobody sampled stayed honestly grey.

The engine this exchange generalises is the one already running for drivers today — in AnOmbak.

See the engine proven in the app — AnOmbak→

Ready to build something intelligent?

Let's engineer your next product — from idea to launch.