This post answers the question I kept asking myself while studying Cardano: why are its upgrades so slow? The answer isn’t technology — it’s governance. Every change to the core has to pass a long on-chain process. And that’s exactly the peak of Cardano’s “slow” philosophy.

First, decide what chain it wants to be

Cardano has never hidden its positioning: a settlement layer. It doesn’t want to be the “do-everything platform.” It wants to be the foundation that carries large, long-term, trustworthy transfers of value.

That positioning dictates its behavior: better to have too little than too much. If a feature can’t be shown to strengthen the core without harming it, it doesn’t ship. Outsiders see “other chains ship a feature every week”; Cardano sees “every change is surgery on a foundation carrying real money.”

Voltaire: on-chain governance arrives

In 2022, CIP-1694 proposed a full on-chain governance system, which landed on mainnet with the Conway era in 2024 under the name Voltaire. At its center are three bodies:

flowchart TB
    A[Any ADA holder<br/>submits a governance action] --> B[Voting weighted by stake]
    B --> C[Constitutional committee<br/>rules on constitutionality]
    B --> D[DReps<br/>one Lovelace, one vote]
    B --> E[SPOs<br/>one Lovelace, one vote]
    C & D --> F[Most actions: two bodies suffice]
    C & D & E --> G[Hard forks: all three must approve]

• Constitutional Committee: a small group that judges whether a proposal is constitutional. Powerful, but replaceable via a no-confidence motion.
• DReps (Delegated Representatives): representatives. Holders who don’t want to vote delegate their voting power to one. One Lovelace, one vote — the more stake, the louder the voice.
• SPOs (Stake Pool Operators): the people who actually run the network. They get extra say on security-relevant changes.

The three bodies check each other: ordinary actions need two approvals, but a hard fork — the biggest kind of change — needs all three. Hijacking an upgrade requires capturing both the stake-weighted vote and the constitutional committee. The bar is very high.

How many steps a hard fork takes

A Cardano hard fork isn’t “the team posts an announcement and ships.” It goes through a complete on-chain process:

flowchart TB
    A[Proposal: bump protocol version] --> B[Developers implement and test first]
    B --> C[Three bodies vote on-chain]
    C --> D[Activated at an epoch boundary]
    D --> E[Network-wide node upgrade]

In June 2026, the van Rossem hard fork — the one paving the way for Leios — was submitted exactly this way: implemented and tested first, put through on-chain voting, then activated with the whole network upgrading in sync. Every step is auditable and traceable. No “quietly changed a config overnight.”

Why caution equals security

Treating “slow” as a principle buys a whole set of security benefits:

• Anti-hijacking: a minority can’t change the protocol; actions need stake thresholds and constitutional review.
• Time for the network: a hard fork activates at an epoch boundary, giving operators a clear window to upgrade — no “chain already forked, half the nodes left behind” chaos.
• Traceability: every action and vote is on-chain, so it’s possible to reconstruct who pushed what.

The price is obvious: slow, and decision-making is cumbersome. But it’s a price Cardano is willing to pay, because for a settlement layer meant to last decades, the loss from one successful hijack dwarfs “being slow.”

Slow, on purpose

“Slow” is the label I gave Cardano too — until I saw the full governance picture. Slow is the direct result of putting security, trust, and long-term viability first. A governance system that needs no top-down permission, that is triple-checked, and where every action is traceable, buys the confidence that the protocol won’t be hijacked by a minority for decades. It doesn’t promise overnight riches — it promises that what should stay solid, stays solid for the long run. A governance mechanism that can update itself is the part of “long-term stability” easiest to overlook and hardest to rebuild later.

Article Link: