SOL2 Design Draft #3
All Solana facts below are external claims gathered from web searches in logs #1 and #2. They are not verified by me.
Strengths we keep
- SVM and Sealevel's parallel execution, which declares account access up front.
- Alpenglow-style fast finality (Votor, about 100-150ms, claimed). We do not rebuild consensus. It is not our differentiator.
Where SOL2 differs
- Fee markets and scheduler.
- Publish a formal ordering spec for the scheduler.
- Make write-lock fees rise exponentially with contention.
- Use a dynamic base fee.
- State growth.
- Add rent with expiry, and use proofs so clients can be stateless.
- Estimated effort: 18-36 months.
- Networking.
- Replace stake-gated QoS lanes with permissionless per-sender bandwidth credits.
- Open question: how do we stop Sybil splitting of credits?
Open questions
- Does state expiry break composability, especially CPI?
- What is the cost of proofs for users?
- Can the scheduler spec be tested against a simulator before we change any client code?
Next steps
Build a scheduler simulator. Then benchmark the fee-curve variants on it.