What actually decentralizes, and why the most sophisticated tools have decentralized the least
There are more than ten thousand decentralized autonomous organizations operating right now. Between them they hold treasuries worth around twenty two and a half billion dollars, and they count over three million registered voters.
Fewer than one in ten of those voters participate in any given decision.
In most large DAOs, a typical governance vote attracts a few hundred people. Uniswap has more than a million token holders. Its proposals routinely draw a few hundred votes.
Meanwhile, the top ten percent of token holders control roughly seventy six percent of all voting power, and in a number of organizations, less than one percent of holders control close to ninety.
I want to sit with those numbers before drawing any conclusion from them, because they describe the most technologically sophisticated attempt at collective governance that humans have ever built, and it is producing something that looks a great deal like what it was designed to replace.
The experiment was fair
This deserves to be said plainly, because the easy move is to sneer, and the easy move is wrong.
The premise was serious. If governance could be encoded, executed automatically, and made transparent by default, then the informal power that accumulates inside every organization would have nowhere to hide. No back rooms. No captured boards. No one able to override the group.
And the population running the experiment was ideal for it. Technologically literate, ideologically committed, financially invested, and self selected for exactly this belief.
If decentralized governance cannot work under those conditions, that tells us something about decentralized governance rather than about the people who tried.
The failures have been specific and documented. The 2016 attack on The DAO. The Beanstalk exploit. Losses across governance attacks exceeding two hundred forty million dollars. Compound facing a coordinated acquisition of voting power by a group operating in the open. Jupiter pausing its governance entirely, citing community division and negative feedback loops.
But the exploits are not the real finding. The real finding is quieter.
Almost nobody votes.
Not because voting is hard, though gas fees and technical proposals do not help. Because the expected impact of any single vote is close to zero, understanding a proposal takes real work, and the stakes feel abstract. It is the same rational apathy that empties municipal elections, and it hits harder here because the community is thinner.
Which produces the outcome nobody designed. When five percent of a body votes, the five percent who show up decide. And the people who reliably show up are the ones with enough at stake to justify the effort, which is to say the largest holders.
Plutocracy did not sneak in. It was the mathematically predictable result of open participation in a system where almost nobody participates.
What has actually worked
Now look at the other side of the ledger, because there is one, and it is older and quieter and almost entirely absent from the conversation.
There are roughly five thousand complementary currencies operating in the world.
The WIR in Switzerland has run since 1934, founded in the depths of the Depression by business owners who could not get credit. It became a nationwide mutual credit network among small and medium enterprises, and as of the most recent widely cited figures its bank held assets equivalent to over five and a half billion dollars.
Sardex, in Sardinia, began in 2009 as a mutual credit system for regional businesses that could not access conventional financing. It has since grown to the point of covering portions of salary payments.
The Chiemgauer in Germany includes a feature worth pausing on. The currency carries demurrage, which means it loses value over time on purpose. Holding it costs you. Which means the only sensible thing to do with it is spend it, which is precisely what a local economy needs money to do.
In Ghent, in one of the poorest neighborhoods in Belgium, the city created Torekes. Residents earned them by gardening, repairing bicycles, and helping at community events, and spent them on bus tickets and meals. The design started from the observation that people had time, skill and willingness, and no paid opportunity to use any of it.
Here is the part I did not expect when I started looking.
Almost none of these run on a blockchain.
The most durable examples of decentralized economic coordination in the world are largely paper, ledgers, cooperative banks and mutual credit agreements, some of them ninety years old.
The pattern underneath both
Put the two columns side by side and something becomes difficult to avoid.
The systems with the most sophisticated technology have decentralized the least. The systems with almost no technology have decentralized the most.
That is not an argument against the technology. It is an argument about what the technology can and cannot supply.
What DAOs built was a mechanism for executing collective decisions. What they could not build was a community that wanted to make them. Code can enforce a vote. It cannot manufacture the willingness to show up, and it turns out that willingness was the scarce resource all along.
What WIR and Sardex and the Chiemgauer had was the reverse. Weak tooling and a real constituency. Businesses that genuinely needed each other, in a defined territory, with a shared problem, and enough relationship that defaulting on the network carried a social cost.
Decentralization was never a technical property. It is a social achievement, and technology can support it or quietly undermine it.
And the undermining is worth naming, because it is not obvious. A system that makes participation frictionless also makes it weightless. When joining costs nothing, leaving costs nothing, and a membership that costs nothing to abandon does not generate the mutual obligation that made the older systems work.
What the tools are actually for
So what is left, once the promise is set aside?
Quite a lot, provided we are honest about which problem each thing solves.
A distributed ledger solves verification. It makes a record that no single party can quietly revise. That is genuinely useful, and it is not the same as governance.
Automated execution solves enforcement. Once a decision is made, it happens without depending on whether someone with authority feels like honoring it. Also useful. Also not governance.
Tokenization solves accounting for things that were previously uncountable. Care, maintenance, ecological restoration, the work that holds communities together and never appears on a balance sheet. That is the most interesting of the four, and the least developed.
And peer to peer exchange solves the intermediary problem, which matters most precisely where intermediaries have been extractive rather than useful.
Notice that not one of those is a governance solution. They are infrastructure. The equivalent of good roads. Roads do not decide where a society goes.
The 2026 direction reflects this, incidentally. The serious projects have moved toward hybrid models: off chain deliberation where humans actually talk, on chain execution where the record must hold, legal wrappers so the thing can sign a contract and protect its members. Messier than the original vision. Considerably more functional.
The test I would apply to my own work
There is a pilot in incubation in the fifth region of Chile called Rainbow Tribe Legion, aimed at integrating exactly these tools with regenerative social and ecological practice.
It is not a finished model, and I would rather say what would make it fail than what would make it inspiring.
It fails if the technology arrives before the constituency. If there is a governance structure before there is anyone who genuinely needs the others, the turnout problem arrives on schedule and the whole thing becomes ten thousand and one.
It fails if participation is free. Every durable system in that older column had a cost of entry and a cost of exit, and those costs are what produced the obligation.
It fails if it optimizes for scale. WIR did not scale for forty years. Sardex is regional by design. The boundary is not a limitation of those systems. It is the mechanism.
And it fails if it promises abundance rather than solving a specific, felt, local problem that the people involved could name without being prompted.
That is the standard, and it is the one I would want applied to anything I build.
The tools are real. Every one of them.
What they are not is a shortcut around the part that has always been hard, which is a group of people who are willing to keep showing up for something that costs them, when nobody is forcing them to, and when their individual vote will almost certainly not decide anything.
Ten thousand parliaments were built. Almost nobody came.
The building was never the difficult part.
Myriam V.
