What Const Said About Conviction in Yesterday’s Novelty Search

The conviction upgrade has caused mixed reactions across the community since its announcement. In yesterday’s Novelty Search, Const stepped in to clarify what conviction does, why it’s being rolled out now, and where it fits in the broader trajectory of Bittensor’s mechanism design.

Read more about the Conviction upgrade below:

The community reacted live on the call. Striker pushed back hard. Sammy weighed in. Below is what was actually said and how each concern was addressed.

Clearing the Air About Conviction

Two things got clarified up front:

Locked stake earns yield. This was the part the Taostats tweet got wrong, and Const made the point explicitly. Locking is not a penalty.

Conviction has a maturity period. You don’t acquire conviction instantly. Building enough to take over a subnet takes multiple months. A team that’s actively running its subnet has plenty of time to respond to any attempt to wrestle control away. As Const put it, it would be a mechanistic failure if a working team got displaced arbitrarily or too quickly, and the system is being designed specifically so that can’t happen.

The upgrade was actually written into the codebase at the launch of dTAO last year. It just wasn’t switched on, because the OTF team felt at the time that it was too much for subnet owners. Recent events with Covenant became the trigger, but the design itself is not a reaction to those events.

Get filled in on the Covenant crisis:

Striker’s Argument Against Conviction

Striker came up to defend a fairly direct position: conviction is anti-founder, won’t actually prevent the behavior it’s targeting, and risks driving builders out of the ecosystem.

His main points:

• Building a serious team is expensive. Frontier labs offer engineers $500K to several million in base plus equity. Bittensor is already squeezing founders, after dilution from funding, owners often retain only 9–12% of emissions.
• Forcing owners to lock tokens to retain control of their own subnet is a step too far. Founders should choose whether to lock, not be forced to.
• Locking is easy to circumvent. If shorting gets introduced, it’s trivially defeated. Even without shorting, an owner can hold the minimum threshold needed to defend the subnet, accumulate owner emissions, and slow-rug instead of fast-rug.
• It creates an unhealthy public dynamic. Every time a founder needs to take liquidity, they’d have to publicly signal an unlock 21 days in advance, which gets front-run, and the price gets dragged before they sell.
• This whole upgrade is a reaction to one person, Sam from Covenanat, exiting badly, and the network is becoming anti-founder over a single bad actor. The result will be lower-quality teams, less money flowing to builders, and founders raising outside Bittensor where they can keep 80% equity.
• He also raised the risk of coordinated coups: with conviction-based voting, someone who dislikes a subnet could bribe holders with alpha tokens to vote out the owner.

How Const Addressed Each Point

On founder economics and the 18%: The 18% subnet team supply isn’t changing. If the concern is that founders get too little, that’s a separate conversation about the 18% itself. Conviction doesn’t touch it.

On recent direction: The last several upgrades to the Bittensor network have all favored subnet owners. Conviction is one upgrade in the other direction, balancing investor protection against owner control. The trajectory isn’t anti-founder.

On circumvention: A 21-day unlock period genuinely changes the attack surface. A flash takeover isn’t possible. The duration of lock required to acquire enough conviction, and the duration of unlock, are the levers that determine whether this works, and those are tunable mechanism-design parameters, not loopholes.

On OTC sales at a discount: Const pushed back firmly here. Most investors he’s spoken to are willing to take locked stake and lock it themselves. Anyone planning to dump 100% of an OTC purchase within 21 days is, by definition, a bad-faith counterparty; that’s not a normal investor behaviour. A 21-day lock period shouldn’t meaningfully change the price of a real OTC trade with a real investor.

On the comparison to public markets: The frontier labs and traditional companies Striker compared Bittensor to all have investor lockups. If you raise $100M or take a million-dollar equity package, you don’t get it all on day one; it vests on a schedule. Locking is a standard already implemented in traditional and regulated settings.

On who actually rugs: Const made the point pretty bluntly. The teams that have sold 100% of their supply and are still running are essentially the teams that rugged their investors, taking Nodexo, Kinitro for instance. Conviction is targeting exactly that pattern. And it would have helped projects like Templar, where investors lost everything when the team rugged.

On voting and coups: The intent is for governance to eventually rest with token holders who have actually invested in a subnet over time. That is the point. Founders should be able to sleep easy knowing arbitrary takeovers can’t happen, while investors should have real recourse if a founder rugs. The maturity period and unlock period are designed to make hostile takeovers slow enough that healthy teams always have time to respond.

On the practical day-to-day for founders: For a team that sells a small portion every month, which is most teams, nothing meaningfully changes. Instead of one transaction (unstake and sell), it becomes two (unlock, then unstake and sell). Owners can also transfer locked tokens to employees as compensation, which creates a useful new tool rather than a constraint.

Sammy’s Contribution

Sammy disagreed with Striker’s claim that conviction wouldn’t have protected anyone in the recent Covenant situation. If Covenant-Sam’s stake had been locked, the on-chain extrinsic to begin unlocking would have been visible 21 days before any sale could happen. Investors would have had a clear, public red flag, exactly the kind of warning that hyperliquid holders get when they see Jeff’s supply approaching unlock and start pricing it in.

His broader point: capital isn’t flowing into crypto right now, and one of the reasons is that crypto lacks the basic protections that exist in traditional equity. Settling for “we can’t prevent rugs” is the wrong instinct. Conviction forces teams into proper investor relations and makes selling more programmatic. That’s a net benefit to all stakeholders, even if it’s a minor inconvenience for founders in the short term.

How It’s Being Rolled Out

Const closed by emphasizing that mainnet will get a “muted” version of the upgrade first. The hope is that the community sees it as a positive signal for both teams and investors and that the fear-inducing framing fades once the mechanism is live. If it doesn’t play out that way, the conversation reopens.

Striker’s closing position was that he still disagrees, that the issue itself isn’t that important relative to other things that need attention, and that the whole thing has felt reactive, a PR patch after a difficult week rather than a deliberate piece of mechanism design. He hopes the rollout includes proper thinking about the lock weights and parameters before it goes live.

Both views are now on record. The upgrade goes out soon, muted first. The community will see how it plays out.

Watch the full Novelty Search episode: