Notes on the concepts and risks behind crypto, checked against official sources.
What an exchange's proof of reserves shows about assets and leaves out about liabilities, pledged collateral, and snapshot-day borrowing, and what Merkle balance checks cover, with a hypothetical solvency table.
Why a wallet's transaction simulation can differ from what actually executes, how minimum-output settings and failed-transaction fees work, and what simulations miss about approvals and malicious contracts.
Who can change code in upgradeable smart contracts, how much reaction time multisigs and timelocks give users, and when emergency powers or slow withdrawals erase that time, with a comparison table and checks.
What zero-knowledge proofs guarantee in rollups (the computation is correct) versus what data availability guarantees (anyone can rebuild the state), and how rollups and validiums differ when operators stop.
How wallets get blockchain data through full nodes, light clients, or RPC providers, what each verifies and what each must trust, and what can go wrong when the data is false, with a comparison table.
How Bitcoin's UTXO model and Ethereum's account model record balances differently, how UTXO selection and change outputs work, and why Bitcoin fees depend on transaction size, with a worked transaction.
How EIP-4844 blobs make rollup data cheaper to post, how they differ from calldata, what the roughly 18-day retention window means, and who must keep data afterward, with a hypothetical cost comparison.
Why each Bitcoin confirmation makes reversal less likely, calculated by attacker hash share, how to choose confirmations by amount, and how exchange deposit policies compare, with a probability table.
How the source of an oracle's price determines how expensive it is to manipulate, what it costs to move a shallow pool, how feed update rules work, and how distorted prices cascade into liquidations.
Why a liquid staking token's market price can differ from its exchange rate, what causes discounts, and how a widening discount can trigger liquidations when the token is used as collateral.
How data availability sampling lets light nodes check that data was published, how detection probability rises with the number of samples, and which network assumptions the guarantee depends on.
How account abstraction changes what a wallet can do, the roles of validation logic, bundlers, and paymasters, and the trust and upgrade risks behind social recovery, with a hypothetical recovery flow.
