In the late 1940s, the British government shipped tractors and bulldozers to Tanganyika to clear vast tracts of land for groundnut production. The machines broke down in hard, abrasive soils. Spare parts were thousands of miles away, and the rains didn't cooperate. The scheme was abandoned within a few years, at a cost of tens of millions of pounds.

Seven decades later, the pattern looks familiar. Water pumps powered by children's merry-go-rounds. Laptops delivered to schools without trained teachers. Improved cookstoves left unused in corners. Each arrived with a compelling story, and each ran into the same wall: the context it was designed to ignore.

Whether inappropriate technology fails is no longer in question. The evidence on that is fairly settled. The real question is why the sector keeps making the same mistake, decade after decade, despite thick files of evaluations and lessons learned. Answering it means looking past the technologies themselves to the institutions that select, fund, and reward them. It also means looking at what successful technologies have done differently.

The Tanganyika groundnut scheme was not an anomaly. Across post-independence Africa and South Asia, governments and donors launched mechanization drives built around imported tractors. Many state-run tractor hire services ended up with large shares of their fleets idle, waiting on spare parts, fuel, or trained mechanics. The machines themselves were sound. The supply chains, repair skills, and field conditions they assumed simply did not exist where they were deployed.

The PlayPump offers a more recent version. The idea was elegant: children spin a merry-go-round, which pumps water into a storage tank. Hundreds were installed across southern Africa in the 2000s with high-profile backing. Field assessments later found the pumps cost several times more than conventional hand pumps. They also needed near-constant play to deliver enough water, and communities found them hard to repair. In some villages, adults (often women) ended up pushing the wheel themselves.

Education technology followed the same arc. One Laptop per Child promised to transform learning by putting low-cost computers in children's hands, and Peru became one of its largest adopters. A randomized evaluation by the Inter-American Development Bank found that the program sharply increased computer access. It produced no detectable gains in math or reading. The laptops arrived. Teacher training, curriculum integration, and technical support largely did not.

Even technologies that work in the lab can fail in the household. Rema Hanna, Esther Duflo, and Michael Greenstone ran a long-term randomized study of improved cookstoves in Odisha, India. They found that early reductions in smoke exposure faded as households used the stoves irregularly and stopped maintaining them. In all four cases, the technical design performed as intended. What failed was the fit between the object and the daily lives of the people expected to use it.

Takeaway

Inappropriate technology rarely fails because the engineering is bad. It fails because the design assumes a world of spare parts, skills, and habits that isn't there.

If the failures are well documented, why do they recur? Part of the answer lies in what development organizations can count and show. A tractor, a water pump, or a laptop is a tangible output. It can be photographed, tallied in a results framework, and reported to a board or parliament. Better maintenance systems, stronger local markets, and changed household routines are harder to see and slower to verify.

Project cycles compound the problem. Most donor-funded projects run three to five years, and success is typically measured at closeout. The costs of inappropriate technology tend to arrive later, after the implementing partner has left and the budget for spare parts has run out. Commonly cited estimates suggest that roughly a third of rural hand pumps in sub-Saharan Africa are not working at any given time. Installation gets funded. Operation and repair often fall between institutions.

William Easterly frames this as a feedback problem. In a market, a product that doesn't serve its users loses customers. In aid, the people using a technology are rarely the people paying for it. Their dissatisfaction has to travel a long way to reach decision-makers, if it arrives at all. In Easterly's terms, planners design solutions from the top and answer to funders. Searchers look for what works on the ground and answer to users.

Career and funding incentives add a final layer. Launching a new initiative brings attention, budget, and professional recognition. Quietly maintaining an old one rarely does. Innovation narratives also suit fundraising: a novel device makes a better story than a revised maintenance contract. None of this requires bad faith. Well-intentioned staff responding rationally to their institutions' reward structures can, together, produce the same failure again and again.

Takeaway

When a development failure keeps recurring, look for the incentive that rewards it. Outcomes are shaped less by what people intend than by what their institutions measure.

The idea of appropriate technology is not new. In 1973, E.F. Schumacher's Small Is Beautiful made the case for "intermediate technology." These were tools more productive than traditional methods, yet cheap and simple enough to be owned, operated, and repaired locally. The evidence gathered since then points to a few characteristics that separate technologies that take root from those that end up abandoned.

First, maintainability within local capacity. International Development Enterprises promoted the treadle pump in Bangladesh, and KickStart later did the same in East Africa. The pump was built from locally available materials and could be fixed by village mechanics. It spread to more than a million farm households in Bangladesh, largely through private manufacturers and dealers rather than free handouts. When repairs depend on imported parts or foreign technicians, the clock starts ticking on the day of installation.

Second, fit with existing behaviour and felt needs. M-Pesa succeeded in Kenya partly because it built on things people already did, such as sending money home through relatives and bus drivers or transferring airtime. Research by Tavneet Suri and William Jack estimated that access to mobile money lifted around two percent of Kenyan households out of poverty. The technology was new. The need it served was old and clearly felt.

Third, real signals of demand and ongoing feedback. Willingness to pay can show whether a productive technology meets a genuine need, but the evidence warns against applying this test everywhere. Jessica Cohen and Pascaline Dupas found that charging even small prices for malaria bed nets sharply reduced uptake without improving usage. Either way, the broader principle holds. Technologies succeed when their designers keep listening after deployment and adjust to what users actually do.

Takeaway

Appropriateness is not a property of the device but of the system around it: who fixes it, who wants it, and who notices when it stops working.

The history of inappropriate technology transfer is less a story about machines than about institutions. Tractors, pumps, laptops, and stoves each failed for local reasons. They were chosen for systemic ones: visibility, short project cycles, and weak feedback from users to funders.

The evidence points to a practical discipline. Before introducing any technology, ask three questions. Who will maintain it in year five? Do users already feel the need it addresses? How will the program find out if it stops working? Test small, measure outcomes rather than outputs, and budget for operation as seriously as installation.

The best technology for a given context is often the one that looks least impressive in a donor report, and keeps working long after anyone stops counting.