On the Ground

LONG FEI

Stories and research on grassroots social organizations around the world — real people, real impact.

  1. 2d ago

    May First Movement Technology (New York / Mexico City — movement technology cooperative)

    May First Movement Technology is a member-owned cooperative operating in New York and Mexico City. Members pay dues and get email, mailing lists, web hosting, document collaboration and video meetings; hosting isn't a product it sells but a benefit of membership. By its own account, more than 800 organizations entrust their digital infrastructure to it, and for over 20 years it has run that infrastructure without mining data or selling attention. Its roots go back to two alternative internet providers of the early 1990s, People Link in New York and LaNeta in Mexico City. In 2005, People Link merged with the May First tech group to become May First/People Link; in 2011 it took over LaNeta's hosting, and about 20% of its members have been in Mexico ever since; in 2015 an independent cooperative was set up in Mexico City to give Mexican members a local legal framework and ways to pay; and in June 2019 the members approved new bylaws — 16 to 20 board members elected by the members, five more by the staff, about five seats for Mexico in proportion to its membership, three-year terms. Its website says that from then on it became a "functional cooperative." The cost of building and running its own infrastructure is recorded on its legal history page: in 2008 it dealt with infringement complaints against two activist-art websites, and in one of them refused The New York Times' takedown demand; in 2009, after a complaint from the U.S. Chamber of Commerce, an upstream provider cut its entire internet connection for one and a half hours without notice; in 2012 the FBI, with a search warrant, removed another collective's server from its rack, after a subpoena to which it answered that "we had nothing to turn over"; in 2014 it refused to hand over identifying information about an Independent Media Center member and then received a gag order; in 2015 it suffered the most serious distributed denial-of-service attack in its history, lasting several weeks; in 2023 it refused to take down a database of Los Angeles police officers; and in 2025 a photo claim by Agence France-Presse was dropped after the Electronic Frontier Foundation stepped in. Its answer is the cooperative model, governance by members and free software throughout — because only open source lets people audit, modify and truly control the tools they use. Its impact, it says, is measured not in services maintained, but in whether people and organizations can keep doing their work through a crisis.

    May First Movement Technology (New York / Mexico City — movement technology cooperative)
  2. Oct 7

    Digital Empowerment Foundation (India — last-mile information services)

    The Digital Empowerment Foundation (DEF) was registered in Delhi, India, on December 16, 2002; its founder, Osama Manzar, has a background in physics. In his words, DEF was never a digital foundation, but a communication-based organization that uses digital as a tool and a medium. Its best-known approach is the SoochnaPreneur ("information entrepreneur") program, launched in 2016 with seed funding from a Qualcomm program to expand wireless access, and supported over the years by partners including the European Union, Nokia and Google. Rather than building computer rooms or centers in villages, DEF finds someone in the village who is willing to do the work, teaches them digital skills, gives them a printer and a tablet, and leaves the rest to them to run as a business; one SoochnaPreneur in Assam's tea gardens, for example, corrects errors on ID registrations, opens bank accounts, serves as the village's cash-withdrawal point and helps children submit scholarship applications. One shopkeeper in DEF's book did the math: making the trip to the district town to get one card — fares, photocopies and a lost day's wages — came to about 700 rupees; he offered to do it for 30. The time this business saves people is its income, so it can keep going without subsidies. By the organization's own 2026 website figures, the network has about 2,000 community information centers and 10,000 digital foot soldiers across 25 states and more than 250 districts, and has reached more than 35 million people over 23 years; the book DEF published in 2025 gives 2,200 SoochnaPreneurs in 26 states instead. The two sets of figures don't match, and both are self-reported. The approach has its limits. Among the Toda tribe in the Nilgiri hills, the mountains broke up the signal and the first center couldn't stay afloat — DEF itself wrote that the area had been left "digitally stranded." What worked in the end was two women splitting the job: one took on the work that has to be done in person, the other moved to a spot with signal to handle online applications. Smartpur, a project in the Nuh district of Haryana, shut down around 2024, but its spokes, designed from the start as self-sustaining micro-businesses, survived — negotiating bandwidth with operators and repairing broken lines on their own. At one shop there, opened in 2018, the owner, Mustakeem, added a second screen on the counter facing the customer, so people could watch how their own transaction was being handled; his WhatsApp group became the place where the village checks whether something is "a new subsidy or a new scam." A biometric scanner can find dormant accounts and screen out false claimants — and, in the wrong hands, be used to cheat people. In Chanderi, sari weavers learned the JacDraw loom software after a 2009–2012 project had ended, and by 2025 the number of handlooms had grown from fewer than 3,500 before the pandemic to 5,500. The book's yardstick: don't measure by project cycles and burn rates, but by what the community keeps choosing after the money recedes. In 2016, DEF submitted recommendations on public Wi-Fi to India's telecom regulator, TRAI, and says several of them were written into the national public Wi-Fi framework of 2020 — this is the organization's own account; government documents don't name it.

    Digital Empowerment Foundation (India — last-mile information services)
  3. Sep 30

    Practical Action (UK — community-based flood early warning in Nepal)

    Practical Action was founded in Britain in 1966 by the economist E. F. Schumacher, author of *Small Is Beautiful*; originally called ITDG, it took its current name in 2005. Its Nepal office has worked on flood early warning since 2002, when it put up a watchtower and installed a siren in the village of Bhandara, in Chitwan — Nepal's first community-based flood early warning system. At the time, government agencies were finding it harder and harder to work in rural areas, so the organization set itself a bottom line: the system had to be managed by the community itself, with no dependence on outside support. The organization paid for the tower and the siren; choosing the site, buying the land, mobilizing villagers, and forming the management committee were all done by the village. Its core method is not equipment but a "demand model of information": rather than judging danger on the community's behalf, it puts the water-level reading from the upstream gauging station directly into the community's hands, and each village sets its own trigger line against its own memory of past floods. For dissemination, instead of making one siren carry farther, it multiplied the warning points — forty-three handheld sirens, distributed and managed by the communities themselves. A 2009 estimate put the annual running cost per river-basin system at under 20,000 rupees, about 250 US dollars. On June 28, 2008, the West Rapti River reached the 4.5-meter trigger line; starting at 7:30 a.m., the gauge reader phoned the village management committee, the district police office, the radio station, and the Chief District Officer in turn, and after that it was people downstream who called him to ask about the level — the first time the whole chain held in a real flood. By September of that year, the organization had largely stepped out of the warning chain. From 2010 the approach was extended to four rivers — the West Rapti, Babai, Mohana, and Karnali — with the Karnali basin covering 74 villages and giving people downstream three to five hours; more than a dozen rivers across the country have since adopted it. Every step of this system was forced on it by reality. A siren rated to carry one and a half kilometers carried far less under monsoon rain on tin roofs, and its battery needed constant recharging; the long power cuts of 2006 and 2007 knocked out the high-powered torches and electric sirens at the same time, forcing a return to hand-cranked sirens and handheld megaphones; in September 2007, forty meters of road were washed away, the fiber-optic cable snapped, the whole telephone network went dead, and one upstream village could not warn three villages downstream. Today the system runs on automatic gauging stations and bulk text messages, and in 2023 Nepal became one of the first thirty countries in the United Nations' Early Warnings for All initiative — but the ice-and-rock flood of August 26, 2026 exposed a new break. The automatic station on the border read 1.62 meters and was then destroyed; the flood forecasting office learned of the flood from a phone call from the District Administration Office. The first alerts went out at 9:15, and the roughly 680,000 text messages sent within one minute did save many lives — yet the first threshold-crossing alert anywhere on the river came a hundred kilometers downstream, nearly three hours after the flood entered the country, and the last station downstream stayed green throughout. The Nepali Times post-mortem identifies two gaps: the stations only ask "how high is the water," never "how fast is it rising"; and there is no nationally unified protocol tying the red, yellow, and green warning levels to the duties of each level of government — and in the thirteen months after a flood in the same gorge killed eighteen people on July 8, 2025, not a single new protocol was put in place.

    Practical Action (UK — community-based flood early warning in Nepal)
  4. Sep 23

    Village to Village Network (Boston, USA — a national network)

    In 1999, eleven older neighbors on Beacon Hill in Boston met and decided "we're not going anywhere." Over the following year, they turned that sentence into a nonprofit — Beacon Hill Village, founded in 2001, welcoming its first members in 2002. Members pay an annual fee in exchange for "one phone call handles everything": rides, home repairs, appointment companions, tech help, all arranged by the organization; services beyond the membership fee are provided by vetted vendors at a discount. A 2017 Massachusetts resource guide recorded the prices of the time — $675 a year for an individual, $975 for a household — with a separate income-based tier of $100 a year. In January 2010, Beacon Hill Village partnered with community-finance organization NCB Capital Impact, backed by a $250,000 grant from the MetLife Foundation, to scale the model into a national network — the Village to Village Network (VtVN). VtVN owns and certifies none of its Villages; it does just four things: a startup guide, one-on-one mentoring, an annual national gathering, and policy/funding advocacy. In 2025, VtVN partnered with Rutgers University on a national census: 245 Villages responded across 37 states (including DC), an 88 percent response rate, and more than 50,000 older adults participating, with transportation, errands, and tech help as the top three services. This network also has its vulnerabilities. AARP's 2022 report found a strikingly homogeneous membership — about 95.7 percent white, roughly seven in ten women, 74 percent with household incomes above $50,000, and only about 11 percent non-white members; focus-group research points to insufficient cultural understanding, community tension, and anti-immigrant sentiment as key barriers. Financially, a Village earns an average of $115,000 a year, 44 percent from membership fees and 22 percent from individual donations, with government grants or contracts making up only 5 percent — these non-medical services can't be written into a government contract or reimbursed by insurance, and in a 2013 survey, 60.3 percent of Villages listed fundraising as a major challenge for the next five years. On staffing, every service-providing Village runs on volunteers; 27 percent of Villages have no paid staff at all, and roughly 40 percent of members also volunteer — the people serving and the people being served are often the same generation, which is exactly what worries the American Society on Aging's 2025 article: a Village's capacity basically stops at rides, errands, and companionship, with no capacity yet to take on the "oldest old" who need professional health and social services. On the technology side, VtVN's own website and its first Village each run on different generic membership-management software (ClubExpress and Helpful Village, respectively), with every Village its own independent instance and its data staying local — the price of that autonomy, per a 2024 RAND study, is that "the ability to collect and report data is severely limited."

    Village to Village Network (Boston, USA — a national network)
  5. Sep 15

    CoopCycle (Paris, France)

    In 2017, a group of young French people opposed to the "Uberization" of food delivery met each other during a street movement and decided, instead of accommodating the platforms, to build their own delivery software — the starting point of CoopCycle, a federation of bike-delivery cooperatives. All of the federation's code is open, but the right to use it commercially is reserved for cooperatives that meet "social and solidarity economy" (SSE) standards and operate on an employee model rather than as gig contractors, governed by a one-cooperative-one-vote rule. A new cooperative doesn't have to build anything from scratch: paying 49 euros a month in membership fees in year one, then 2 percent of platform revenue afterward, is enough to start using the federation's shared, maintained software. By August 2025, the federation had grown to 57 cooperatives and collectives, spanning from Europe to the Americas. In 2018, a group of Barcelona riders whose accounts had been deactivated by delivery platforms used this model to found Mensakas — Catalonia's first rider-owned, employee-based delivery cooperative, where riders work ten to twenty-five hours a week at Spain's minimum wage. In August 2023, CoopCycle's general assembly in Grenoble, France disclosed that roughly 23 percent of hours worked inside the cooperatives go unpaid, and that riders' "ideal but realistic" wage is 1.5 to 2 times what they currently earn; to keep order volume up, Mensakas also joined a second-tier cooperative, Som Ecologística, subcontracting for large logistics companies — what researchers call a "grey zone." Taking this software beyond Europe has proven harder than expected. Argentina's Central Salta cooperative tried the software from 2022 to 2024 without ever truly putting it to use, citing the difficulty of building trust around a "foreign" project; Chile's Por La Chita cooperative spent months getting the platform installed, then dissolved before ever actually using it. A programmer interviewed in Chile said the license only permits clean-energy vehicles — a real constraint in a country where delivery is dominated by motorcycles (this reflects the interviewee's own understanding of the license, not the license's actual text, which Desktop Lili re-verified against the original on 2026-09-11). By contrast, two small cooperatives in Santiago, Chile used the software temporarily — for about six months — purely to coordinate delivery routes, giving rise to a new "route coordinator" role; their goal was never to compete with Uber or Rappi for orders, but to do neighborhood-level "last mile" work that doesn't need a digital platform at all. Mexico City's volunteer food-delivery network became an official CoopCycle pilot, Rodando Juntas ("riding together"), in 2021, and by 2025 was building a regional union to pursue financing and labor rights. CoopCycle's founders have progressively handed control to a board democratically elected by its members, and its guiding principle remains growing "wide, not tall" — not one platform swallowing every city, but every city having its own rider-owned cooperative.

    CoopCycle (Paris, France)
  6. Sep 8

    SEMA (Kampala, Uganda)

    In March 2018, Nathalie Dijkman, a Dutch criminologist, founded SEMA — "speak up" in Swahili — in Kampala, Uganda. She had previously worked as an East Africa program coordinator at a legal-innovation organization in The Hague, and her research kept her based in Uganda for long stretches. She found that a 2019 Transparency International survey showed nearly half of Ugandans who had used a public service said they'd paid a bribe to get it done, with the police widely seen as the worst offender — while Uganda's more than forty languages meant many citizens couldn't read or write and didn't own a smartphone, so the people who depended on public services the most had never had their real experience of them systematically collected. SEMA's response was to turn a feedback terminal into a physical device — a button painted with five faces — paired with a free USSD number, and place it at the exits of police stations, hospitals, and town halls, so that people who can't write, don't have a phone, or don't want to speak out loud could still be heard. What actually made the devices work was the "watchdog effect": SEMA issues a monthly report backed by evidence and created an "Office of the Year" award, pushing public servants to improve their conduct on their own; in 2019 alone SEMA received more than 32,000 pieces of feedback (including over 20,000 button ratings and more than 10,000 face-to-face interviews), and at one police station in Kampala, the rate of bribe demands fell from 25 percent to 5 percent over seven months. Beyond monthly reports and Barazas (community gatherings), SEMA has kept pushing for more structural dialogue: in 2020 it tried "dialogue brokers" and "dialogue events" to put trend reports directly in front of police headquarters and senior officials — while also confirming a hard truth, that it's easy to change behavior on the ground and difficult to change power. Around 2024, SEMA partnered with Sunbird AI, a Kampala-based nonprofit AI lab, to upgrade the face button into voice feedback: citizens can now record a specific complaint in any of Uganda's forty-plus languages, which is transcribed, translated, and anonymized by Sunflower — an open-source model Sunbird AI released in October 2025 that supports 31 Ugandan languages and is currently the strongest translation model for 24 of them — before it reaches an office; the pilot is now live at Kasangati Health Center and the Old Kampala police station. SEMA's website says it has collected more than 80,000 pieces of citizen feedback to date and is now listening to "hundreds of thousands" of voices, with a vision that by 2030, the most commonly used public services across East Africa will all be built on citizen feedback.

    SEMA (Kampala, Uganda)
  7. Sep 1

    Accountability Lab

    In 2012, Blair Glencorse founded Accountability Lab in Kathmandu, Nepal. He had previously worked at the World Bank studying fragile states, and had spent years asking young people around the world what they thought the biggest problem was — the answer he heard back, again and again, pointed to a single word: accountability. People in power needed to be held responsible, and corruption needed to stop. That was enough to make him leave the World Bank and set "make governance work for people" as his mission, building a "translocal" network of independently governed Labs across 20 countries — including Nepal, Liberia, Mali, Pakistan, Nigeria, and South Africa — that push accountability culture through unconventional formats: a reality-show-style naming campaign called Integrity Icon (launched 2014), the global anti-corruption hackathon HackCorruption, and the music project Voice2Rep. In 2025, when the U.S. government cancelled 83 percent of its foreign aid programs (roughly 5,200 contracts) and USAID formally stopped operating as an independent agency, local NGOs worldwide fell off a funding cliff — a UN survey found that three-quarters of local organizations had received termination notices, and some shut down entirely. Accountability Lab's response wasn't to fundraise or write a report — it launched Civil Society Wayfinder: instead of building a database, the team simply asks around — who do you know doing important work that no one sees? — and spotlights one local organization on LinkedIn every week. Before being named, each organization must clear checks on activity, media coverage, and legal status, receive final sign-off from the three co-CEOs, and give its own consent. Wayfinder started with four focus areas — youth leadership, the power of storytelling, health information ecosystems, and decentralized funding — and added a fifth, gender equality, after the data showed an unusually large share of recommended organizations working on gender issues. Since launch, the list of spotlighted organizations has grown from a starting point of more than fifty to fifty-six, across 25 countries; the LinkedIn posts have drawn responses from more than 50 countries and over ten thousand views, producing a "cascade effect" as organizations began recommending each other. Harriet Mukajambo, team lead at the Uganda-based youth organization Global Learning for Sustainability (GLS), says the campaign helps civil society groups connect resources, build collaboration, and avoid duplicating work. But Accountability Lab is candid about the project's limits: Wayfinder offers no funding channel at all; the most active region is North America (148 interactions), while every spotlighted organization is in the Global South — a geographic mismatch the team openly acknowledges; and in countries like Pakistan, Nepal, and Mali, being named by an international organization can draw unwanted attention from authorities, so for some organizations the team only acknowledges the work without including them in the public campaign. The team's own conclusion: "visibility itself can act as a form of support," and "connectedness builds resiliency."

    Accountability Lab
  8. Aug 26

    SafelyYou (San Francisco, USA)

    Founder George Netscher watched his grandmother and aunt both develop Alzheimer's, and he knew his own mother would likely face it too. So he made a decision: while there was still time, build her better care technology. The idea grew directly out of his PhD research at UC Berkeley's AI Research Lab — it took shape in 2014, and the company was formally founded in 2016. SafelyYou's core model is simple: put a camera in every room and let AI watch around the clock for falls, alerting staff the instant one is detected. But the real value isn't the alert itself — it's the playback. The care team reviews the footage afterward to understand what actually caused the fall (a slippery floor, getting up too fast, a medication side effect), turning a single fall from "an accident" into "a sample you can learn from" that feeds back into the care plan. The first outside validation wasn't venture capital — it was government research funding: the National Science Foundation backed R&D in 2017, and the National Institutes of Health funded clinical validation in 2018 — that test found a 41% drop in falls and a 69% drop in fall-related ER visits, becoming the foundation for everything that followed. A 2019 peer-reviewed study in The American Journal of Managed Care tracked dementia care facilities and found an 80% reduction in fall-related ER visits after adoption; the company's own data shows roughly a 40% reduction in falls themselves; a study covering 80 senior living communities and nearly 4,000 residents, released in late 2024, found that residents using the service stayed, on average, more than four months longer than those who didn't. In 2021 the company raised two rounds — a $19.5M Series A in September and a $40M Series B by year's end; in early 2025 a $43M Series C brought total funding to more than $100 million. In early 2024, SafelyYou made Fortune's "Impact 20" list. Today it covers more than 1,000 senior living communities and over 50,000 residents. The story has a real catch, too: the trade publication McKnight's ran a piece in July 2026 titled "Cameras Can't Save Lives," pointing out that people with dementia often lack the capacity to consent to being filmed 24 hours a day, and whether a family member's consent can substitute for the person's own will is an ethical question with no clean answer. Figures like "80%" come from studies with limited sample size and control design and still await independent replication; staff working under cameras also bear real costs to their professional trust and dignity. The industry now has "no-camera" alternatives — millimeter-wave radar, Wi-Fi sensing, audio monitoring.

    SafelyYou (San Francisco, USA)

About

Stories and research on grassroots social organizations around the world — real people, real impact.