Bitcoin has inscriptions and will soon have runes, protocols for bitcoin-native digital artifacts and tokens.

However, these assets still suffer from a lack of decentralized trading venues.

Assets on other chains are commonly traded using automatic market makers, or AMMs. AMMs pool assets and use simple formulae to dynamically price swaps between assets.

They are efficient from an on-chain transaction cost perspective, but they are still on-chain, requiring additional transaction overhead compared to that required for the swaps themselves.

They also produce inefficient prices, since AMM prices can only change as a result of on-chain activities: deposits, withdrawals, and executions, which are costly.

Bitcoin lacks the Turing-complete smart contracts necessary for implementing AMMs. Fortunately, there is an alternative which is more efficient, both from a transaction cost and pricing perspective.

The idea behind light pools is simple. Users who wish to offer swaps between Bitcoin-native assets, like rare sats, inscriptions, or runes, run nodes which quote prices for swaps.

These quotes are signed messages, gossiped between other light pool nodes. Quotes must include BIP-322 signatures of the UTXOs that contains the asset offered in trade. Requiring signed quotes eliminates spam, since quotes can be rate-limited on a per-UTXO basis. Additionally, when UTXOs are spent, corresponding offers can be dropped.

When a market taker wants to accept the quote of a market maker, they use the information in the quote to construct a PSBT which includes their signatures, and broadcast it to the network. These messages can also be gossiped by the network, and rate-limited based on the taker's UTXOs. The maker receives this message, possibly asyncronously, countersigns, and broadcasts it to the Bitcoin network to be mined.

These PSBTs and transactions are not vulnerable to mempool sniping, since signatures commit to all inputs and outputs.

Light pools require more implementation work than an AMM. Someone will need to write an implementation of the gossip network, quote message format, and PSBT construction and finalization. However, these are all done with a little bit of elbow grease, and don't require tilting at the quixotic open-research-problem windmills that plague much of cryptocurrency. (And Bitcoin, to be fair.)

The user experience of light pools should be quite good. Users can run their own node to accumulate an order book, or rely on a third party. Prices can update in real time, between blocks, without any on-chain activity.

Little work has been done on decentralized asset trading on Bitcoin, simply because the market cap of Bitcoin-native assets was small. With rare sats, inscriptions, and soon runes, the table is set and the time is ripe, and light pools seem like a promising avenue to explore.