1/ Practically all L2s so far have been designed with payments as the use case and most of them are UTXO based. UTXO based L2 are highly unsuitable for trading. Some explanation here: ethresear.ch/t/a-dex-on-pla…
2/ When we started there was no account based L2. Recently, MATIC (similar to Gluon) has proven to be a good account based L2. OVM is another solution like MATIC that tries to create an entire EVM on L2. There are some impedance mismatches using an EVM based DEX.
3/ An #EVM based #DEX needs to write most of its logic in solidity. Upgrading this is not easy, since the upgraded logic could do anything with the funds. @dydxdigital for example has this issue. The admin key can arbitrarily upgrade contracts.
4/ One solution for this is to have a governance process. Making minor fixes is impractical and major upgrades are a burden. In Gluon, we can upgrade the contracts WITHOUT IMPACTING trading.
5/ The burden of writing the key financial product logic in solidity. In Gluon, an app is simply a ledger definition and the actual implementation can change often (we push upgrades almost every day) without impacting the published contracts.
6/ Bitcoin's transaction verification is guaranteed to terminate but is not expressive, while Ethereum is highly expressive but needs a gas cost to prevent runaway code. EVM based L2 flaw is that they will degenerate to Eth performance once expensive operations become cheap.
7/ he Gluon approach is to take the middle path. Define the ledger, which is deterministic and lightweight but expressive enough to describe financial applications without the need for EVM/gas/etc.
8/ Some other L2s are just fancy math added on top. zkrollups is basically a more efficient L1, zkstarks are not easily changed (a minor order type change requires recreating all the polynomial equations, pushing a new contract, etc).
9/ Many of these may get better with time (as they have) but the important issue is that they are unnecessary. A merkle proof takes 800 bytes for a block with million transactions and takes kilobytes in a stark proof. It is unknown if any of these can have a block explorer.
10/10 Gluon aims to be simple, transparent and is designed for rapid upgrades. Learn more t.me/leverj
• • •
Missing some Tweet in this thread? You can try to
force a refresh