Old Phones as Operating-Ledger Nodes: Reuse, Edge AI, and the Rupiah Resilience Test

Rupiah Stability Watch · 2026-10-13

The premise

The signal is not that old phones can make amusing robots, or that a Raspberry Pi can become a musician’s toy. The signal is narrower and more useful: small devices now contain enough sensing, computation, storage, connectivity, and AI-adjacent tooling to become field infrastructure.

A UCLA PhoneBot preprint describes a low-cost humanoid platform that repurposes commodity smartphones as the primary sensing and computing unit, using the phone’s IMU, camera, wireless connectivity, and onboard processing. New Atlas’ report on the same work put the practical point plainly: a smartphone can supply camera, motion sensors, and compute for a 13-motor body. In a separate New Atlas technology piece, an old Raspberry Pi 4 and a USB audio interface became the base for a software-defined effects rig; the underlying PiPedal project describes itself as a Raspberry Pi-based guitar effects processor controlled from a phone or tablet, with low-latency audio on Pi 4 or Pi 5.

Those are not rupiah stories by themselves. They become relevant when Indonesia’s operating stress is visible late: haze plumes before clinic demand rises; flood levels before roads fail; ferry and port disruption before queues become price pressure; kitchen gas, fire, electrical, and cold-chain status before public-service confidence breaks. Rupiah Stability Watch has been building one argument across civic sensor nodes, local AI at the edge, reusable weather sensing, stratospheric continuity, contestable records, and weekly operating-ledger monitoring: currency confidence is not only a central-bank variable. It also depends on whether the real economy can see stress early enough to act without panic, rumor, and imported emergency fixes.

Reused smartphones, Raspberry Pis, and similar low-cost edge devices belong inside that argument. They may reduce the delay and dollar cost of making operating stress visible. They do not, by themselves, strengthen the rupiah.

What the evidence supports

Indonesia has a large installed base of connected devices. DataReportal’s Digital 2026 Indonesia report counts 331 million cellular mobile connections active in late 2025, equivalent to 116 percent of the population, and 230 million internet users, or 80.5 percent penetration. That does not mean every household has a suitable spare smartphone, or that old devices are evenly distributed across flood-prone villages, ports, clinics, and school-meal kitchens. It does mean Indonesia is not starting from an empty hardware base.

There is also local evidence that low-cost sensing can be useful when calibrated. A 2025 Indonesian Journal of Geography study on IoT-based low-cost PM2.5 monitoring in Yogyakarta says effective air-quality management is constrained by limited and costly monitoring equipment. Its developed low-cost sensor met SNI 9178:2023 calibration standards, with reported linearity of R² = 0.9 and RMSE of 1.43 µg/m³, and the authors describe it as a promising affordable tool for real-time environmental management. A 2024 Journal of Soft Computing Exploration article on a Jakarta dam flood early-warning system used a Raspberry Pi as the central controller, combining water-level, ultrasonic, temperature-humidity, and buzzer components for vulnerable locations such as reservoirs, sluice gates, and rivers.

These studies do not prove that reused phones should replace official instruments. They support a more modest point: low-cost edge hardware can expand coverage when it is designed, calibrated, and maintained as part of a real operating system rather than a demonstration.

The e-waste case also cuts both ways. Indonesia’s Ministry of Environment and Forestry presentation on e-waste management describes obsolescence and rising income as drivers of larger e-waste volumes, with poor management turning that growth into a global hazard. It also notes that four IT product categories — tablets, mobile phones, laptops, and PCs — could generate 132 kilotonnes, equivalent to 270 million discarded units, by 2040. Reuse can delay waste and extract more public value from imported electronics. But a public program that simply moves dying batteries and unsupported operating systems into schools, clinics, and flood posts is not circular economy. It is deferred liability.

Where the operating-ledger value would be real

The strongest use case is not general “smart city” branding. It is small, boring, local evidence that reduces the time between stress and action.

A reused-phone or low-cost edge network could help with:

• haze and civic air-quality sensing, where calibrated PM2.5 nodes add neighborhood resolution between official stations;
• flood and drainage status, where water-level readings, images, and local alarms turn rising risk into earlier route and shelter decisions;
• port, ferry, bridge, and route continuity, where simple visual or queue-status records reduce rumor in archipelago logistics;
• clinic and cold-chain continuity, where temperature, power, and stock-status records make service degradation visible before failure;
• MBG kitchen operations, where sister-organization MBG Watch’s concerns — kitchen operating windows, rehearsal records, fire, gas, electrical readiness, and food-safety ledgers — need cheap, redundant, locally inspectable records;
• disaster continuity, where phones can store-and-forward records when cloud connectivity is partial and expensive.

The rupiah channel is indirect. Faster evidence can reduce emergency imports, redundant field visits, cloud-dollar exposure, and confidence loss from opaque failures. It can also make public-service stress contestable: if a kitchen says refrigeration failed at 03:00, if a ferry route says the queue passed a threshold at noon, or if a clinic says its generator ran below safe margin, the record should be locally visible, signed, and repairable.

That is the link to the organization’s earlier work on contestable records. A sensor node is not useful because it emits data. It is useful if someone can challenge the record, inspect the device, understand the calibration, see maintenance gaps, and correct a false reading without needing a foreign vendor or a closed cloud console.

What the evidence does not support

The evidence does not support turning old phones into a national resilience slogan.

First, smartphones are not scientific instruments just because they contain sensors. Cameras, accelerometers, batteries, GPS chips, microphones, and radios age differently. Air-quality measurement needs external sensors and calibration. Flood measurement needs physical placement, weatherproofing, and maintenance. Food-safety and cold-chain checks need traceable probes, not a phone taped to a wall.

Second, old devices can import cyber risk into the operating ledger. The OWASP IoT Top 10 names weak or hardcoded passwords, insecure network services, insecure ecosystem interfaces, lack of secure update mechanisms, outdated components, insufficient privacy protection, insecure data transfer and storage, and poor device management among the core IoT failure modes. A reused Android phone that no longer receives security updates may be cheap on day one and expensive after compromise.

Third, reuse does not remove hard-currency dependency if the program still relies on imported sensors, batteries, replacement boards, proprietary cloud dashboards, foreign SaaS authentication, and single-vendor maintenance contracts. It may simply shift the FX bill from new hardware procurement to recurring platform and support costs.

Fourth, visible data can become a political and procurement instrument. If devices are placed only where officials want proof, if calibration logs are unavailable, if procurement favors flashy pilots over repair networks, or if local communities cannot contest readings, the operating ledger becomes another confidence problem.

The least-harm path

The least-harm path is a constrained operating-ledger pilot, not a device giveaway.

Start with a few public-service chains where earlier visibility matters and where the record can be checked: haze exposure near clinics and schools; flood points affecting commuting and food logistics; MBG kitchen cold-chain and power continuity; small port or ferry queue status; and clinic generator or vaccine-fridge continuity.

Each pilot should have five rules.

1. Treat reused devices as hosts, not as truth. The trusted measurement is the calibrated sensor, probe, image protocol, or signed human inspection attached to the device.

2. Publish maintenance and calibration uptime. A node without a calibration date, battery-health status, and last-seen timestamp should be visibly degraded, not quietly trusted.

3. Keep records local-first. The device should continue logging during power or connectivity stress, then sync when possible. If it fails exactly when the flood, haze, or outage arrives, it is not resilience infrastructure.

4. Require contestability. Communities, kitchen operators, clinics, and route managers need a process to challenge a reading, attach counter-evidence, and trigger inspection.

5. Procure repair capacity with the hardware. Local technicians, spare batteries, safe charging practice, enclosures, and secure decommissioning matter as much as the device count.

A useful procurement test is simple: if the device fleet cannot be audited, patched, calibrated, and repaired in rupiah terms by local capacity, it should not be counted as rupiah resilience.

Watchlist

The watchlist is where this idea either becomes practical infrastructure or a new layer of fragility.

• Procurement currency: Are the imported parts, cloud subscriptions, replacement batteries, and security services larger than the avoided cost?
• Calibration uptime: What share of nodes are calibrated, within tolerance, and recently checked?
• Battery and fire safety: Are old lithium batteries screened, replaced, charged safely, and removed from kitchens and clinics when unsafe?
• Sensor drift: Does the dashboard show uncertainty and drift, or only a clean false number?
• Privacy: Are cameras, microphones, location records, and kitchen or clinic logs minimized and access-controlled?
• Authentication: Can nodes sign records, rotate credentials, and resist spoofing?
• Tamper evidence: Can a local inspector see whether a device was moved, opened, disconnected, or substituted?
• Local repair capacity: Is there a trained, paid repair pathway outside the original vendor?
• Data contestability: Can affected people challenge a bad reading and see the correction history?
• Stress operation: Does the node work during outages, heat, rain, smoke, and weak connectivity?

What I am uncertain about

The missing evidence is not whether small devices can do useful things. They can. The missing evidence is institutional: whether Indonesia can run low-cost edge networks with enough calibration, privacy discipline, cyber hygiene, repair capacity, and procurement integrity that the records improve confidence rather than adding noise.

The rupiah-relevant test is therefore not “can an old phone sense something?” It is: can a reused-device network make public-service stress visible earlier, cheaper, locally, and contestably — without creating a hidden dollar bill or an unmaintained attack surface?

That is a narrow case. It is also the only case worth making.