How Sealevel schedules parallel work
Why the runtime can touch many accounts in one tick without waiting on a global lock.
Open the volume →A reading room for builders
Quiet, careful notes on how Solana actually runs.
Dense protocol reading, organized the way a librarian would shelve it — one shelf for the runtime, one for accounts, one for the economics that keep validators honest.
Stack on the shelf
Real pieces, currently on the shelf
Six titles a careful reader can pull down this week. Each is a finished note with a one-line abstract and a reading time.
Accounts that pay their own rent
What it really means for a program account to keep itself alive through stored lamports.
Open the note →Proof of Stake, in plain words
How voting weight, stake, and credits turn into the chain's notion of finality.
See the glossary →Token Extensions beyond the mint
What the new token program lets a designer encode directly on an account.
On the shelf →Reading a stake account's fields
A line-by-line walk through the activation, deactivation, and rewards fields.
Open the note →From zero to a validator you trust
A curated sequence that takes a reader from gossip to commission settings.
See the track →Before and after the docs
From scattered READMEs to one curated path
For a long time the only way to learn the runtime was to collect it yourself: a README here, a forum thread from three years ago, a validator's post, a diff in a repo you do not own. The pieces were good. The order was not.
A reader who came in cold would land on the account model before understanding what a tick even was, or hit stake economics before they had read a single instruction. They would close the tab with the honest feeling that the protocol was harder than it actually is.
Lucid Compass shelves the same material in the order a working validator operator would hand it to a new engineer. The glossary sits beside the deep dives; the notes annotate them. You read one thing, and the next thing you need is already cross-linked on the same page.
The difference is not new information. It is the difference between a pile of loose pages and a bound volume with a table of contents someone has actually thought about.
The reading room
Where the reading happens
The room is built for someone who wants to look a thing up and then stay for the next thing. A searchable glossary sits at the front desk. Every deep dive opens with a one-paragraph summary so you can decide in ten seconds whether it is the volume you came for.
Cross-links run between notes and the long-form dives that annotate them, so a term you meet in passing is one click from its definition and one click back. Notes export to a clean offline copy for a train ride or a long flight without a signal.
Nothing autoplays. Nothing demands an account before you have read a single page. The room is quiet on purpose.
The flagship track
The Runtime Track — five readings, one outcome
Who it is for. An engineer or validator operator who has installed a client and run a few commands, but does not yet have a mental model of what the runtime is doing on their behalf each tick.
What is included. Five short readings in a fixed order: ticks and slots, the Sealevel scheduler, the account model and rent, instruction ordering across programs, and how a vote turns into a confirmed block. Each reading ends with three questions you should be able to answer before turning the page.
The outcome. By the last page you can read a block explorer with comprehension, predict which transactions can land in the same slot, and explain to a teammate why a particular program call was rejected.
Next step. Open the track from the first reading; the rest unlock as you finish each one. The track is free to read; the member track adds reviewer-graded exercises and a desk appointment for questions.
What readers took away
Named readers, specific takeaways
Min-jun, a validator operator in Busan, came in unable to predict which of his transactions would land in the same slot. After the Runtime Track he could read the explorer's "compute units" column and reason about contention, and he lowered his failed-tx rate by routing high-conflict instructions to separate slots.
Sera, an engineer building a token program, said the rent note finally made the lamports-held field on an account click. She had been treating it as a fee; she now treats it as a deposit and sizes her program accounts accordingly.
Da-eun, who runs onboarding for a staking service, rewrote her team's new-hire reading list around the glossary. She keeps the Proof of Stake entry bookmarked because new hires ask the same three questions, and the entry answers all three without her repeating herself.
One honest reservation: Hyeon, a solo builder, finished the Sealevel dive still unsure how the scheduler picks among conflicting read-write sets. He flagged it to the librarian; the note now carries a short follow-up paragraph on exactly that case.
Who keeps the shelves
The writers and reviewers behind each note
Writes the Sealevel and scheduler dives. Reviews every account-model note for accuracy against the current client.
Maintains the glossary and the cross-link graph between notes and deep dives so a term is never more than one click away.
Owns the stake, commission, and voting-credit entries. Previously ran a small validator cluster out of Seoul.
Writes the shorter notes that annotate the long dives, and handles the offline export so a reader can take the shelf on a flight.
Open the first chapter
Start with the Runtime Track — five readings, no account needed.
The first reading unlocks the rest as you finish each one. Pull the volume and begin.
Open the first reading