It seems Robinhood Chain Infra is poised to enter the next phase. Recently, I've seen two DEXs, one by @metricxyz and the other by @notesystems. What they have in common is that they use external oracles for pricing, unlike the traditional AMMs like Uni's x*y=k. This suddenly reminded me of @BreederDodo, who invented PMM during the DeFi Summer. Those who didn't experience the DeFi Summer might be unfamiliar with this, so let me briefly introduce PMM - Proactive Market Maker. PMM - By introducing oracles to obtain external market reference prices, it actively gathers a large amount of liquidity around the current market price. When a certain asset in the pool is bought up and decreases, PMM automatically raises the price of that asset, thereby attracting arbitrageurs to replenish the missing asset through arbitrage. This design can significantly improve capital utilization, reduce transaction slippage, and mitigate the risk of impermanent loss. Dodo's complete defeat to Uniswap back then was likely due to two reasons: First, mainstream coin prices were on centralized exchanges (CEXs) like Binance. Oracles were slower than CEXs at the time, allowing arbitrageurs to exploit this window to profit from limited-access coin (LP) transactions. Second, the main battleground for DeFi during the summer was long-tail coins, many of which lacked both CEXs and oracles, relying solely on AMMs for price discovery. PMMs naturally couldn't capitalize on this. However— RWA's situation is different now! Let's talk about the previous version—we invented Meme-to-Stock pairings and created blockbuster products like AMC, harboring a dream of "making on-chain transactions influence the visible world." See how this month, with the market cooling down, hasn't that dream been completely forgotten? Because it was always just a dream. Even if it were realized, it would only affect a few obscure, unheard-of small-cap stocks; the 7 Sisters (Alibaba's seven most popular cryptocurrencies) were simply impossible. What might the next version be? AAPL's price was originally discovered on Nasdaq, and on-chain DEXs are essentially price takers. Letting the pool discover stock prices on its own is a first-principles error. On-chain AMM prices are simply noise plus arbitrage costs. The DODO design was somewhat mismatched for crypto assets at the time, but it suddenly feels right for RWA. Finally, let's discuss the characteristics of the two Dexterities, many of which differ from DODO's design. 1. Metric - Liquidity Follows Price Separating "pricing" and "liquidity": Pricing is handled by oracles (Chainlink, Pyth), while the pool only provides liquidity. LP liquidity isn't tied to a fixed price range, but rather to a position "deviating x% from the reference price." When the reference price moves, the entire liquidity pool shifts accordingly, never falling outside the range. Those familiar with Uni V3 or Meteora should understand; you no longer need to manually ebalance every day. Details like Robinhood's stock token dividend multiplier changes are handled by dedicated oracles; otherwise, the price would fluctuate wildly after each dividend or stock split. In short: DODO "gathers" liquidity around the oracle price, while Metric makes liquidity "follow" the oracle price. 2. NoteSystem's Swap - What happens when prices are outdated or the market is closed? The engine is called S-DAMM, which I feel is a complete version of DODO PMM: Reference price + inventory adjustment + one-way LP—DODO has all three. It adds a time dimension: if the price hasn't been updated for approximately 5 seconds, the confidence level is halved, and the quote automatically widens while the depth decreases. DODO was preyed on by arbitrageurs because it didn't know how outdated its prices were. Specifically designed for US stock market closures: Regular inventory is withdrawn 60 minutes before closing; a dedicated Continuous inventory is used only at night and on weekends, offering wider price arbitrage; Discovery mode is activated when no new prices are available; at the open, prices are revalued based on the true value before accepting new orders. LP tiering: Senior USDG vault, plus a junior backstop to absorb losses. DODO's LPs were essentially unsecured. In short: DODO only knew the price, while Note knew how outdated that price was and whether the US stock market was open. Metric is already running on its mainnet with decent data, attracting many international KOLs to farm. Note is likely still in the testnet phase, but it's worth keeping an eye on (by the way, StoneBroker's options, which I was closely watching, haven't been released yet; it seems like it's going to fail...).