Open Source Changed Software. Open Energy Changes Everything Else.

My co-worker used to work at WordPress. This was early — back when it was still a startup and "open source" was a weird ideology that made managers uncomfortable. I didn't get it. My boss really didn't get it. "Why would anyone write code and release it for free?" he asked, genuinely confused. He wasn't dumb. He just couldn't see it yet.

That was less than twenty years ago. Today, open-source software runs the infrastructure of human civilization. This is the story of how that happened — and why we think energy research is about to do the same thing.

The before times

If you weren't writing software in the 1990s, it's hard to convey how bad things were.

Software was expensive. Not "subscription expensive" — capital expenditure expensive. A database server license could cost $50,000. A compiler could cost $10,000. A web server — the thing that serves web pages — cost money. Microsoft's IIS was bundled with Windows Server (which cost money). The alternatives were proprietary Unix systems (which cost more money). Even basic developer tools were licensed and metered.

The result was predictable: software development was slow, fragile, and concentrated in the hands of companies that could afford the tools. Small teams couldn't compete. Individual developers couldn't experiment. Innovation happened inside large corporations on their schedule, at their pace, serving their priorities.

And the software was bad. Not because the engineers were bad — because the ecosystem was closed. Every company solved the same problems independently, behind closed doors, with no way to share solutions. A bug fix at IBM didn't help anyone at Sun. A clever algorithm at Microsoft stayed at Microsoft. The entire industry was doing redundant work, constantly, everywhere.

The turn

The shift didn't happen all at once. It happened in stages, and at each stage, the incumbents said it couldn't work.

1991 — Linux. Linus Torvalds, a 21-year-old Finnish student, posted the kernel of an operating system to a Usenet newsgroup. "I'm doing a (free) operating system," he wrote. "Just a hobby, won't be big and professional." Within a year, hundreds of developers were contributing. The professional OS vendors said it was a toy.

1995 — Apache. A group of webmasters created the Apache HTTP Server by sharing patches to an existing open-source web server. "A patchy server" — Apache. Within two years, it was the most popular web server on the internet. It still is. The commercial web server vendors said it couldn't scale.

1998 — The term "open source" is coined. Christine Peterson suggests the phrase at a strategy session after Netscape releases its browser source code. The movement gets a name. The suits start paying attention — and they're confused. "Why would anyone work for free?"

2001 — Wikipedia. Jimmy Wales launches a free encyclopedia that anyone can edit. Britannica says it'll be full of errors. Today Wikipedia has 60 million articles in 300 languages and is the fifth most visited website on earth.

2004 — WordPress. Matt Mullenweg forks b2/cafelog and releases WordPress as open-source software. My co-worker joins the team. His friends think he's nuts — working for a company that gives away its product? WordPress now powers 43% of all websites on the internet. Forty-three percent.

2005 — Git. Torvalds builds a version control system to manage Linux development. He releases it as open source. Git becomes the universal tool for collaborative software development. GitHub (built on Git) becomes the world's largest code repository. Microsoft buys it for $7.5 billion.

2008 — Android. Google releases Android as an open-source mobile operating system. Nokia and BlackBerry say smartphones need proprietary platforms. Android now runs on 3.6 billion devices.

2014 — Kubernetes. Google open-sources its internal container orchestration system. Every major cloud provider adopts it. It becomes the standard for running software at scale.

2023 — AI models. Meta releases LLaMA. Mistral releases Mixtral. Stability AI releases Stable Diffusion. The most powerful AI models become open source. OpenAI's closed approach starts looking like the exception, not the rule.

The graph

Here's what the adoption curve of open-source software looks like:

Open Source Software Adoption

100% ─┐
      │                                           ╱── 96% of servers
      │                                        ╱─╱    run open source
 75% ─┤                                     ╱─╱      (2026)
      │                                  ╱─╱
      │                               ╱─╱
 50% ─┤                           ╱──╱
      │                       ╱──╱
      │                    ╱─╱
 25% ─┤               ╱──╱
      │           ╱──╱
      │       ╱──╱
  0% ─┤──────╱
      └──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬
        '91'93'95'97'99'01'03'05'07'09'11'13'15'17'19'21'23'25
              ↑     ↑     ↑     ↑        ↑        ↑
           Linux  Apache  WP   Git    Android  Kubernetes

Three things to notice:

1. The early years were flat. From 1991 to about 2000, open source existed but adoption was marginal. It was "that weird thing" that idealists did. Managers didn't understand it. Investors were skeptical. The mainstream ignored it.

2. The curve inflected around 2003–2005. This is when WordPress, Git, and the broader ecosystem crossed a threshold where open-source tools were simply better than the proprietary alternatives. Not cheaper. Not philosophically purer. Better. Better because thousands of people were iterating on them simultaneously. Better because bugs got found and fixed by users worldwide, not by a single QA team. Better because the best ideas from everywhere got merged into the same codebase.

3. By 2015, the debate was over. Nobody argues about whether open source works anymore. 96% of the world's servers run Linux. 90%+ of cloud infrastructure is open source. Every major tech company both uses and contributes to open source. The question shifted from "why would you open-source it?" to "why wouldn't you?"

Now look at energy

Open Energy Research

100% ─┐
      │
      │
 75% ─┤
      │
      │
 50% ─┤
      │
      │
 25% ─┤
      │
      │
  0% ─┤────────────────────────╳ ← You are here
      └──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬──┬
        '91'93'95'97'99'01'03'05'07'09'11'13'15'17'19'21'23'25

We are in the flat part.

Energy research today looks like software in 1995. The knowledge exists — scattered across 768 patents, university labs, garage experiments, and obscure forums. But it's fragmented, proprietary, unreplicated, and largely invisible.

• Inventors patent their work and then can't afford to commercialize it
• University labs publish papers behind paywalls that cost $30 each to read
• Garage experimenters get interesting results and post them on YouTube with no structured way to replicate or verify
• The Invention Secrecy Act can sequester anything that gets too close to something someone doesn't want public
• Companies with energy interests have every incentive to acquire and shelve competing approaches

This is the same dysfunction that software had before open source. Redundant work. Fragmented knowledge. Concentrated control. Slow iteration. And a prevailing assumption that "serious" research requires institutional backing, expensive equipment, and proprietary IP protection.

Open source proved that assumption wrong for software. We think it's wrong for energy too.

The parallel

And the parallel for the solution:

What the inflection point looks like

In open-source software, the inflection came when the open tools became better than the closed ones. Not just cheaper. Not just more accessible. Better.

Apache didn't win because it was free. It won because hundreds of sysadmins contributed patches from their production environments — real-world hardening that no single vendor could match. Linux didn't win because Torvalds was a genius (though he is). It won because thousands of developers scratching their own itches produced an operating system that was more reliable, more flexible, and more secure than anything a single company could build.

For Open Energy, the inflection will come when the open experimental dataset becomes more trustworthy than any single lab's results. When 50 independent replications of a bifilar coil measurement — conducted by 50 different people with 50 different NanoVNAs in 50 different garages — produce a clearer picture of the coil's behavior than any single published paper.

That's not a fantasy. That's how citizen science works in astronomy (Galaxy Zoo), biology (eBird, iNaturalist), and climate monitoring (Weather Underground). The aggregation of many independent observations is more robust than any single observation, no matter how well-equipped the observer.

We just need enough observers. The experiments are defined. The platform is built. The map is waiting.

Why it'll happen faster this time

Open-source software took about 15 years to go from marginal to dominant (1991–2005). Open energy research has structural advantages that software didn't:

1. The tooling is already cheap. A NanoVNA costs $30. A function generator costs $50. A soldering iron costs $15. The "expensive lab equipment" barrier that slowed early open-source adoption (you needed a computer, which cost $2,000+ in 1991) barely exists for basic energy experiments.

2. The collaboration infrastructure exists. In 1991, sharing code meant mailing tarballs on Usenet. Today, the entire global collaboration stack — version control, web platforms, video hosting, real-time communication — is already built and mostly free.

3. The cultural precedent is set. Open source proved the model works. Wikipedia proved it scales. The argument "why would anyone do this for free?" has been answered so thoroughly that the next generation doesn't even ask it. They grew up on it.

4. The urgency is real. In 1991, proprietary software was annoying but not existential. Energy is existential. Climate change, energy poverty, geopolitical dependence on fossil fuels — these are problems that affect everyone, everywhere, right now. The motivation to contribute is stronger than "I want a better compiler."

Your boss still won't get it

Some version of "why would anyone do this for free?" will be asked about Open Energy, just as it was asked about open source. And the answer is the same:

Because the alternative is worse.

The alternative is every garage inventor working alone, patenting their work, and hoping a corporation licenses it. The alternative is knowledge locked behind paywalls and secrecy orders. The alternative is slow, redundant, institution-gated progress on the most important engineering problem of the century.

Or you can do what Torvalds did. Post it. Share the patches. Let anyone who wants to contribute, contribute. Let anyone who wants to verify, verify. Let the network do what networks do — route around failure, amplify signal, and make the whole greater than any single node.

My co-worker at WordPress understood this before I did. He saw that giving software away wasn't a sacrifice — it was a strategy. The most powerful strategy available: make the thing you care about impossible to suppress by making it belong to everyone.

That's what we're doing with energy research. The question isn't whether it'll work. Open source already proved the model. The question is how fast, and whether you'll be part of it.

In 1998, your boss couldn't understand why anyone would write code for free. By 2026, open source runs 96% of the world's servers.

The same curve is starting for energy. We're in the flat part. The experiments are ready. The platform is built. The map is waiting for you.

Build your Hum →