Straight answers to the
questions people actually ask.

Tailscale, Nebula, and hand-rolled WireGuard meshes give you an encrypted IP fabric — essentially a flat private network over the public Internet. That's the first half of what FrogNet does, and FrogNet uses WireGuard as one of its transports. But the IP fabric isn't the product; it's the floor.

On top of it, FrogNet adds three things those meshes don't: a full per-node application stack (Apache, MySQL, DNS, DHCP, PHP) so every node is a working service host, not just a routable address; a semantic compression layer (BLDC-1) that learns the structure of real HTTP traffic and sends only what changed, cutting bandwidth by ~94% on repeated traffic; and a transient database that any node can read or write, so applications share state without every node needing direct connectivity.

Short version: Tailscale makes packets flow. FrogNet makes applications work — including over links Tailscale would consider broken.

Meshtastic carries short messages over LoRa. Reticulum is a framework for building delay-tolerant applications on top of RF links. AREDN is an 802.11-based ham radio IP mesh. cjdns and Yggdrasil are encrypted IPv6 overlays for the Internet.

None of them run a full web application over a narrowband link. FrogNet does — by treating the link as just another transport for semantically compressed HTTP. The same dashboard that runs on WireGuard over fiber runs on a 4800-baud-equivalent link, because the application stack is identical on every node and the wire is spoken in compressed frames, not literal bytes.

FrogNet and Meshtastic actively interoperate for sensor telemetry: a Meshtastic / LoRa RF bridge writes ESP32 readings into the transient database, and every FrogNet application treats them as native data.

No — by design. FCC Part 97.113(a)(4) prohibits "messages encoded for the purpose of obscuring their meaning" on the amateur bands. FrogNet's BLDC-1 semantic layer transmits deltas against template dictionaries that the receiver must already hold; to an RF monitor without the dictionary, the traffic is indistinguishable from a cipher. That's a conflict with both the letter and the spirit of Part 97, and we're not going to argue the edges of it.

The radio isn't the point anyway — it's just a transport. FrogNet is the application layer, and any IP-capable pipe works underneath it. The candidates we're actively looking at as replacements for the ham role are Meshtastic and MeshCore, both of which run on ISM-band RF where template-based compression is perfectly legal. Depending on how we integrate, FrogNet can sit on top of those networks (using them as the bearer for semantic traffic), alongside them (receiving their telemetry as a data endpoint into the transient database, which is what's happening today), or both. Commercial narrowband, satellite data services, and any Part 15 digital radio are also open to us.

Validated throughput at 4800 baud was measured over a shaped Ethernet link — rate-limited, lossy, and jittery to match a narrow RF channel — not over any particular radio.

The amateur-radio community, though, remains exactly the community FrogNet is built alongside: the operators who stand up comms when the grid is down. We collaborate with EmergencyHam and friends over WiFi / LoRa / WireGuard at EOCs — not over the amateur bands themselves. If you're a ham operator who wants to help, email me.

Signal is excellent at what it does — it protects the content of your messages from the people carrying them. FrogNet eliminates the carrier. There is nobody in the middle to trust, compromise, or compel. No servers to subpoena. No phone numbers. No accounts. No metadata. And it keeps working when the Internet does not.

A VPN is an encrypted tunnel to someone else's building — the cloud, Google, Apple, Meta. Your data still ends up on their servers. A FrogNet is not a tunnel. It is your own building. The services run on your hardware. There is no "other end" where a company has your data. There is no company at all.

No to all four. FrogNet has no token, no consensus, no ledger, no caching tier for someone else's content, and no onion routing. It is a private mesh that runs applications. You put your data on your nodes. Your nodes talk to each other. Nobody else is involved.

Short of running a node yourself (which I can set up — email me), the public evidence is: live dashboard screenshots showing nine peers online with real RTT, compression, and throughput numbers; the architecture papers documenting the wire protocol and semantic layer; the "How to Think Like a Frog" technical writeup; the book on Amazon; 18K+ LinkedIn impressions on the FrogNet article series; three issued U.S. patents in related domains; CAGE code 1A5Y5 with SAM registration; and an NSF SBIR Phase I invitation.

That's not proof — proof is reading the code. But it's not vaporware either.

Yes. If you've got a Raspberry Pi or a cheap VPS and want to join the live mesh, email me and I'll get you onboarded. You'll run a pre-release image, join the WireGuard broker, and appear on the dashboard as a peer. You'll see every other node, measure real RTT, and watch the semantic engine compress your traffic.

Pre-release operators get credit as early contributors once the repo goes live, and their deployment notes become the seed of the operator handbook.

TRL 6–7. Running in a relevant operational environment across Seattle, New York, and Amsterdam. 93.8% semantic compression measured in production. 44× throughput improvement on constrained links. Active development partner in New York City. Field deployment with the EmergencyHam community. Not production-hardened — that's part of what open-sourcing is for — but well past prototype.

Right now, Fawcett Innovations, LLC — a Washington State LLC with CAGE code 1A5Y5 and SAM registration. Donations are held for the specific purposes listed on the Sponsor page: four provisional patent filings (~$7,500) and 501(c)(6) Foundation formation (~$2,500). Once the Foundation is formed, it takes ownership of the patents and the codebase.

501(c)(6) is the structure used by the Linux Foundation, the Cloud Native Computing Foundation, the Eclipse Foundation, and most other software foundations that coordinate industry contributors around open source. It's a trade association — designed to support an industry or technology, hold IP on behalf of a community, and enable corporate membership. It fits the mission.

501(c)(3) is for charitable and educational organizations. FrogNet's work isn't primarily charitable — it's infrastructure — so (c)(3) would be the wrong fit.

One consequence: contributions to a 501(c)(6) are not tax-deductible as charitable donations for U.S. donors. Some contributions may be deductible as ordinary and necessary business expenses for businesses that benefit from the Foundation's work — ask your accountant.

Then I do what I can with what comes in, in priority order: file the most important provisional first, then the next, and so on. Each filing is roughly $1,800 all-in, so the first dollar still moves something. Any shortfall I cover from my own pocket as I can. I don't take the money and walk.

If a donor asks for a refund because the goal wasn't met by a specific date, they get one. PayPal has a process for this, and I will not contest it.

Apache 2.0 is the current plan. It's the de-facto standard for foundation-governed infrastructure projects (Kubernetes, Apache HTTP, Cassandra, Spark, Linux Foundation projects), it's patent-friendly, and it lets commercial users build on FrogNet without adding friction for anyone.

Apache 2.0 will be the default for the core networking substrate — proxy, daemon, discovery, transient database, wire protocol. The Foundation may choose a different license for specific components (CC-BY for documentation, for example). The final decision sits with the Foundation's first board, and I'll advocate loudly for Apache 2.0 on the core.

Apache 2.0 is irrevocable. Once a version is published under Apache 2.0, that version stays Apache 2.0 forever — anyone can fork it, redistribute it, or build on it. The licensing rug-pulls you saw at Elastic, HashiCorp, and MongoDB only applied to future versions of code those companies wholly owned. Existing Apache releases are still there on every mirror on Earth.

The architectural protection is ownership. The Foundation owns the code and the patents, not Fawcett Innovations. Fawcett Innovations is a customer-facing services company that runs on top of the Foundation's code, the same way Red Hat runs on top of Linux. It has no unilateral authority to change the license.

Community governance. The Foundation's bylaws will include a board elected by the membership, a technical steering committee elected by active contributors, and a requirement that any license change on existing code require a supermajority. The model here is Eclipse and Apache — both have survived decades without a governance crisis, and both have bylaws the Foundation can adopt wholesale as a starting point.

The last line of defense is Apache 2.0 itself: if the Foundation ever did something unconscionable, the community forks the last good release and moves on. That possibility, by itself, disciplines the governance.

Defense, not offense. The patents exist to prevent someone else from patenting the same ideas later and using those patents to attack FrogNet contributors or users. The Foundation will hold them; the Apache 2.0 patent grant will cover everyone downstream; and the Foundation will commit (as the Linux Foundation and OIN already do) not to assert them offensively. It is the same strategy used by Linux, Kubernetes, and most serious open source infrastructure to protect the community from patent trolls and predatory competitors.

Everything in the core networking substrate: the semantic proxy, the semantic daemon, the discovery system, the transient database, the FNW1 wire protocol, the BLDC-1 codec, the tunnel broker, and the node configuration tooling. This is the platform — the thing that makes FrogNet a FrogNet.

Applications built on top (specific dashboards, vertical apps, commercial support tooling, managed deployments) may be separately licensed by whoever builds them. Fawcett Innovations' consulting, training, and certification services are the commercial side of the house — open source core, commercial value-add. That's the Linux / Red Hat split, applied here.

The core Android framework and the libraries it depends on will be open. Some app-specific product decisions (pricing, branding, the managed broker service) are commercial. A fork built from the open code will have everything it needs to run a family network; a fork will not get the hosted broker or the commercial support.

Yes. Apache 2.0 permits commercial use without asking anyone. Build it, sell it, no royalty. The Foundation doesn't assert the patents against downstream users of the Apache release. What you can't do is claim your product is endorsed by the Foundation without going through Foundation membership — that's trademark, not copyright.

Consulting, training, certification, and deployment services. Companies that want to run FrogNet in production will need help configuring brokers, operating at scale, integrating with their existing infrastructure, and certifying their engineers. Standard open-source economics — the code is free, expertise is paid. It's how Red Hat, Canonical, SUSE, HashiCorp (originally), and dozens of others built sustainable businesses around open source.

Standard Linux. Current deployments run on Raspberry Pi (Zero 2W and up), small x86 boxes, a MacBook, and DigitalOcean droplets. A node needs Python 3, PHP, Apache (or nginx), MySQL / MariaDB, and WireGuard — everything available from the package manager on a recent Debian / Ubuntu / Raspberry Pi OS.

Minimum practical hardware: a Raspberry Pi Zero 2W with 512MB RAM will run a node. Anything more comfortable runs anywhere.

Clone the repo, run the install script, point it at a broker (the Foundation's or your own), and register. The node assigns itself a /24, joins the WireGuard tunnel mesh, and starts discovering peers. Once discovered, it appears on every other node's dashboard.

The install script is the same one I use to provision new nodes on the live mesh. That script will be part of the initial release.

No. The Foundation will run a public broker that anyone can use free of charge. Fawcett Innovations will offer a managed broker service with SLAs for commercial users. And you can run your own broker on any internet-reachable machine — it's the same code.

Architecture review (read the papers, push back on decisions you disagree with — save me from myself before the decisions are public). Documentation scoping (what does a new contributor need to know? what's missing from the writeups?). Deployment testing (run a pre-release node and tell me what broke). Application ideas (what would you build on this?). Every one of those is a contribution before line one of public code.

Email me — I'll scope something to your skills.

Python for the semantic proxy and daemon. PHP for the application-level API. Bash for the node configuration and discovery. JavaScript for the dashboards. MySQL / MariaDB for the transient database. WireGuard and basic Linux networking for the transport layer. Kotlin / Android for the mobile app. C for ESP32 sensor firmware. Nothing exotic — this is the boring, widely-known, long-supported stack on purpose.

John Fawcett — 67, writing code since 1975. Fifty-plus years building real-time networked systems at Boeing (Special Achievement Award), Sierra On-Line / Electronic Arts / Wizards of the Coast (multiplayer game servers in the early Internet era), Microsoft (SCCM Linux agent), and NanoString Technologies (random-access S3 driver for multi-gigabyte spatial-biology imaging). Three issued U.S. patents. Author of FrogNet: A Living Network. Over $500,000 of personal R&D invested. Living on social security. Full background on the Story page.

Two reasons. The shorter: Anthropic announced Claude Mythos — an AI that finds and chains software vulnerabilities faster than any human team can patch them. Anthropic judged it too dangerous to release. When something like Mythos gets out, every system with a live Internet attack surface becomes a target at machine speed. FrogNet's architecture addresses that structurally, by removing the attack surface rather than hardening it. Sitting on the technology would be wrong.

The longer: I've been building this for a decade with the intent of going commercial. The Mythos threat convinced me the technology belongs in the public domain — and the economics of a defensive patent filing plus Foundation formation are within reach of small-dollar sponsorship. So here we are.

FrogNet: A Living Network is the technical-origin story of building systems that refuse to lie. It explains where the FrogNet philosophy comes from — fifty years of networking under constraints that forced honesty — and it's the deepest available writeup of the "why" behind the architecture. Not required reading; it's the founder's narrative, not a spec. Free on this site as foreword + Chapter One; full book on Amazon.