Tezos Athens: How a Vote Became a Running Rule

One upgrade, two choices

In early 2019, Tezos participants were not simply asked whether they liked an upgrade. Nomadic Labs offered two versions. Athens A would lower a roll—the stake unit then used in the proof-of-stake system—from 10,000 to 8,000 tez. Athens B would leave it alone. Both would increase gas limits. A seemingly small parameter made the choice concrete: the community had to select which rules it wanted the network to execute.^[1][2]

The proposal was executable OCaml code, not just a statement of intent. Nomadic Labs published the changes as patches and urged people to check that the implementation matched its description. This gave the vote a specific object. Supporting a slogan such as “improve the network” and selecting this particular program were different commitments.^[1]

The ballot was only part of the transition

Athens A advanced. Bakers, the network’s block-producing delegates, participated in a stake-weighted process: proposal selection, an exploration vote, a testing period and a promotion vote. It was not one person, one ballot. Jacob Arluck’s contemporary TQ Tezos account also records discussions across forums and chat channels. Putting votes on a chain had not moved the entire job of deliberation into software.^[2]

The tests mattered because the new rules would inherit a running system. Nomadic Labs tested migration against mainnet state and measured its time and storage costs. A test chain could begin with existing accounts and contracts, giving participants a more realistic trial than an empty demonstration. Choosing code did not by itself establish that the transition would work.^[1][2]

What actually changed

Athens activated in May 2019. The historical protocol documentation records the smaller 8,000-tez roll and higher gas limits. It also preserves an easily missed qualification: the number of computation steps available in a block doubled, while permitted input/output operations did not. Gas accounting changed too. “Twice the capacity” would therefore be a misleading promise about every kind of transaction. These are the rules of that historical upgrade, not today’s baking requirements.^[3][4]

Nor did self-amendment mean that every connected application could ignore the change. Athens renamed fields in the interface used by other software. The documentation told wallet, explorer and tool developers how to support the transition. The protocol could adopt the approved rules through its amendment mechanism; the surrounding ecosystem still had work to do.^[3]

A repeatable decision, not an end to judgment

Babylon followed in October 2019. The first amendment had become a procedure the network could use again. Athens matters for that practical achievement: a choice between concrete programs reached activation. It did not prove that every future vote would be wise or every implementation safe. The durable result was a route from proposal to operating rule, with human scrutiny needed along the way.^[2][3][4]