Anza’s SIMD-0675 would schedule Solana validators by geography

Solana’s validators take turns producing blocks, and right now the order of those turns has nothing to do with where anyone actually lives.

Anza’s research team wants to shrink that toll. Its proposal, SIMD-0675, would line up validators that sit close to each other so they produce blocks back to back. In a simulation built on real mainnet data, the change cut block production handover delays by 53%.

What Anza is proposing

The work comes from Anza researchers Quentin Kniep and Roger Wattenhofer. They authored two Solana Improvement Documents, known as SIMDs, which are the formal channel for suggesting protocol changes.

The two documents are SIMD-0674 and SIMD-0675. Both were opened on approximately 29 September 2026.

Together, they describe a system where validators register their geographic location on-chain. The network then uses those locations to build what the authors call a geo-aware leader schedule.

Every handover involves some waiting. The incoming leader needs the previous leader’s work before it can build on top of it, and distance stretches that wait.

The current schedule is random with respect to geography. The proposed one groups validators into geographic bins and runs them consecutively, so the baton mostly passes between neighbors instead of across the planet.

The design caps each bin at a maximum of 3 consecutive leaders. A full run through a region lasts approximately 2.4 seconds before the schedule moves on.

The numbers behind the pitch

The researchers tested the idea against mainnet data from epoch 1038, which involved 661 validators.

Under the existing random scheduling, mean leader handover latency came in at 36.2 milliseconds. Under the geo-aware model, it dropped to 17.0 milliseconds.

That is the 53% improvement.

One detail matters a lot for validators worried about their revenue. According to the proposals, stake weights and slot allocations stay exactly the same.

Every validator keeps its original share of slots. The proposal only reshuffles when those slots occur, not how many each validator gets.

There is a catch, and it is not a small one. Validators that choose not to register a location would be excluded from block production and voting.

On process, SIMD-0675 moved to “ready for review” status on 30 September 2026 at 14:27 UTC. SIMD-0674 remains in draft.

Why Alpenglow makes this urgent

The timing is not random, even if the current leader schedule is. The proposals are tied to Alpenglow, Solana’s upcoming consensus protocol.

Alpenglow aims for finality of around 150 milliseconds. Finality is the point at which a transaction is considered permanent and cannot be reversed.

The protocol was transitioning into testnet and devnet environments in late September 2026, right as these SIMDs appeared.

When your total finality target is measured in the low hundreds of milliseconds, a handover that eats tens of milliseconds is not background noise. It is a meaningful slice of the whole budget.

What this means for validators, builders and SOL holders

For validator operators, the biggest shift is disclosure. Registering a physical location on-chain becomes the price of admission for producing blocks and voting.

The near-term milestones are clear enough. Watch how the community responds to SIMD-0675 now that it is open for review, and whether SIMD-0674 moves out of draft.

Also watch Alpenglow’s progress through testnet and devnet. If that protocol keeps advancing, pressure to squeeze out delays like handover latency will only grow.