← All posts
Jun 2026 · Vision · Netizen Labs

d/acc, built in a town

In late 2023, Vitalik Buterin proposed a frame for thinking about technology that he called d/acc. The "d" does a lot of work: defensive, decentralization, differential, and democratic acceleration. The argument is a third path between two loud camps. On one side, effective accelerationism, e/acc, says build everything as fast as possible and trust that it works out. On the other, the pause-and-doom camp says some technology is too dangerous and should be stopped. d/acc says neither: accelerate, but steer. Favor the technologies that make defense easier than offense, that spread power rather than concentrate it, and that keep humans in the loop.

A year on, he revisited it and kept the core: the goal is a world safe for a plurality of people and communities, reached by deliberately developing the protective technologies first. Nick Bostrom called this differential technological development, advancing the defensive faster than the offensive. d/acc is that idea turned into a design preference you can apply to what you actually build.

Most of the d/acc conversation is necessarily big and abstract: pandemic preparedness, AI alignment, resilient physical infrastructure, defending the information ecosystem. Important, and mostly out of reach for any one person. Röbel is the small, concrete version. Not a manifesto about civilization, but a credential, a secret ballot, and a coin, running defense-first in a real town. Here is how each letter of d/acc shows up as something already live.

Defensive

d/acc's first instinct is to ask which technologies help people defend themselves, their identity, and their discourse. The town's stack is built almost entirely out of those.

The foundation is proof of personhood: a soulbound citizen credential that two existing citizens have to vouch for, which makes the town resistant to the oldest attack on any online system, one person pretending to be a thousand. On top of it sits privacy by construction: votes are secret, the key that could decrypt them is split across a federation so no single party can read a ballot, and citizenship is granted without collecting documents. The data the town never holds is the data nobody can demand from it or steal from it. That is information defense and identity defense, exactly the d/acc category, built for a place small enough to actually use it.

Decentralization

d/acc is suspicious of single points of control, because a thing one party can run is a thing one party can capture. So the town is built so that no one party is load-bearing. The keys that decrypt votes and own the money are held three-of-five across citizens, not by an operator. The governance runs onchain through a governor and a timelock, not through an admin panel. We host infrastructure; we cannot read a vote or move the treasury. Decentralization here is not a slogan, it is which keys exist and who holds them.

Democratic

The point of all that defense and decentralization is to let a community actually decide things. So the town votes privately and provably, and a passing vote can release money from its own treasury through participatory budgeting. The credential that proves you are a real resident is the same credential that lets you vote and, now, lets you hold the town's own money. Democratic technology, in the d/acc sense, is technology that makes collective decision-making cheaper and harder to rig. That is the whole product.

Differential

The last "d" is the discipline: of all the things you could build, build the defensive ones first. It would have been easier to ship an extractive version of every layer, an identity system that harvests data, a currency designed to make a few people rich, a governance tool with a backdoor for the operator. We built the other one each time on purpose. The clearest example is the money.

The money as defense-favoring technology

Röbel Münzen, the town's currency, are a study in choosing the defensive design. They are citizen-gated, so they are hard to fake: only verified humans can mint them. They are demurraged, so they are hard to hoard: balances gently shrink, which means wealth has to flow rather than pile up, and a single mechanism becomes both a basic income and a wealth tax. And they are backed one-for-one, so they are hard to inflate: every coin is collateralised by a real person's own issuance. Hard to fake, hard to hoard, hard to inflate. A currency whose own rules resist capture is a defense-favoring technology in the most literal sense.

Defense-in-depth, on purpose

d/acc at the level of grand domains is a worldview. At the level of a town's money it becomes ordinary engineering, the discipline of making defense cheaper than attack at every layer. The currency's design notes spell it out, and it is worth reading as applied d/acc:

None of that is glamorous. That is the point. Differential defensive development is mostly unglamorous, a hundred small decisions to make the protective path the default one.

Plurality, one town at a time

d/acc shares a horizon with the idea of plurality that Audrey Tang and Glen Weyl write about, and with the network-state conversation: a world of many self-governing communities rather than one center. A town that can verify its own people, decide its own questions, and issue its own money is a node in exactly that kind of world. The next step, towns trusting each other so value can move between them while each keeps its own gate, is plurality made operational, decentralization at the scale of polities rather than servers.

The honest size of the claim

We are not claiming to address the largest items on the d/acc list. A town does not solve AI risk or pandemic defense. What it can do is take the civic slice, identity, voice, and money, and build it defense-first, decentralized, and democratic, then run it for real and publish every lesson. d/acc usually reads as an argument about the future of civilization. We wanted a version small enough to be true today. It turns out to be a credential, a secret ballot, and a coin in a real person's wallet.