FireChat

No mesh app has ever been more famous than FireChat was in the fall of 2014, and none has left less behind. Open Garden’s messenger ran on iPhone from March 2014, carrying traffic across Bluetooth, Wi-Fi, and Apple’s Multipeer Connectivity with no tower and no data plan in the loop, and for a stretch it was the closest thing the world had seen to infrastructure-free communication at scale. It is also the category’s founding cautionary tale. The servers were switched off quietly in February 2020, the company behind the decade’s most-photographed protest tool stopped operating, and nothing was ever published to mark either event. The gap between those two states, global fame and silent disappearance, is most of what there is to learn from it.

From earthquake relief to a protest tool nobody marketed for

FireChat was never designed for the use that made its name. Its lineage traces to disaster relief: SPAN, an open-source Wi-Fi meshing experiment that Josh Thomas and Jeff Robble wrote in 2010 after watching a flood of earthquake responders overwhelm what was left of Haiti’s cell networks. Micha Benoliel and his co-founders turned that instinct into Open Garden in 2011, on the premise that the phones already in everyone’s pockets could carry a network by themselves. When the consumer app shipped in 2014, first on iPhone in March and then on Android on April 3, the target was the opposite of a disaster: the dead zone created by too many phones packed into one venue, a Burning Man or a Coachella.

The market it captured was protest, and it got there ahead of the festivals. Taiwan’s Sunflower Movement adopted it within weeks of the iOS launch; that June, when Iraq throttled web access, something like 40,000 people in Baghdad installed it over one weekend, even as Open Garden’s own staff were warning publicly that the app was “not meant for secure or private communications.” Then came Hong Kong. A single call from Umbrella Movement student leader Joshua Wong in September 2014 produced the numbers the category still measures itself against: two million chat sessions across the four days from the 26th to the 29th, on top of half a million accounts registered inside a week and roughly 100,000 downloads in the first day alone. Lesser spikes in Russia, Ecuador, Catalonia, and India trailed over the following two years.

Why virality never became a business

By the time of the Hong Kong surge, Open Garden had already raised $12.8 million total: about $2 million in a 2012 seed round, plus a $10.8 million Series A led by August Capital that closed quietly in March 2014 and wasn’t announced until that December. (Numbers as high as $28.7 million show up in some funding databases, but no press coverage backs that figure up.) None of that money, and none of the worldwide press attention, solved the app’s real problem: anyone could read any chatroom, which made it genuinely useful in the middle of a crisis and pointless once the crisis ended, an uncomfortable fit given that the company itself had already warned users the app wasn’t secure. Open Garden kept trying to patch the gap: private messaging with claimed end-to-end encryption arrived in July 2015, a disaster-oriented FireChat Alerts feature followed in May 2016, and by February 2017 the company was trying to license the mesh technology to other developers as an SDK. Nothing caught on. App updates stopped altogether in 2018, and in February 2020 the servers went dark, with no announcement. By then Benoliel had already moved on to something else, co-founding Nodle in 2017 to apply the same Bluetooth-mesh concept to IoT devices paired with crypto incentives, a venture still running as of 2025 and 2026. The underlying idea kept going. The app built on top of it didn’t.

Where it stands against this project

FireChat’s mesh only existed for as long as phones stayed awake and in the foreground, so Apple’s background restrictions quietly killed it the moment a crowd dispersed between crises. Always-on Windows relay hubs with a dashboard solve that specific problem here: coverage doesn’t evaporate when fewer phones happen to be nearby at a given moment. FireChat also had no way to confirm a message arrived, it broadcast and hoped; this project treats acknowledgment and relay confirmation as a baseline, not an extra. Disaster response was something FireChat bolted on late, via the 2016 Alerts feature; here, last-known-location mapping is part of the core design rather than an add-on. And where FireChat sold the idea of a parallel, unstoppable internet and never came close to delivering on it, this project keeps its claims modest: ordinary phones, internet optional, nothing more.