Vitalik's new idea is to use "options" instead of the traditional "borrowing+liquidation" approach in DeFi for "index tracking assets" (such as synthetic stablecoins, assets tracking US dollars or CPI), which is very interesting. (Vitalik probably plays Polymarket quite a bit, haha) To understand his meaning, first take a look at the traditional gameplay: For example, if a user pledges ETH and lends stablecoins (such as DAI), and the ETH price drops sharply, the system will force liquidation (liquidation) and sell the collateral ETH to repay the debt; This will lead to a market drop and everyone bursting together, adding insult to injury. This is also an important reason why ETH has always plummeted during black swan events in history. Moreover, the system must rely on real-time oracle machines (which instantly tell you how much ETH is currently), otherwise the clearing will be chaotic. However, the problem is that real-time oracle machines are extremely vulnerable to attacks or manipulation, and once there is a problem, there is no time to save them. New gameplay with options: Split 1 ETH into two types of "option tokens" (called P and N, positive token and negative token), and use a slow oracle (the type that predicts the market in real-time, just confirm the price slowly over a few days) to decide how to divide it on the day of expiration. Key point: P+N can always redeem 1 ETH When the price fluctuates greatly, your position does not suddenly return to zero or explode, but slowly deviates from the target like a quadratic curve (smooth drift), giving you buffer time. If it's still complicated, let's use a simpler analogy: Traditional gameplay: "Borrowing money to buy a house, once the price drops, the bank will directly take away your house"; Option gameplay: "If you buy an insurance contract, your losses will gradually increase when the price fluctuates, not reset at once, but can be adjusted slowly. The benefits are: • No need for real-time oracle, use slow ones (the prediction market has proven that slow oracle is safer and has time for humans to intervene and check). Algorithmic stablecoins are much safer, and Vitalik said he feels more at ease when he holds them himself, not afraid of the oracle being instantly deceived. There are also obvious drawbacks: Regular rebalancing is required: Before expiration, you need to manually or have the agreement help you change positions, otherwise the exposure will gradually shift. There may be slippage (transaction costs) during rebalancing, and how to design robots/MEVs that do not deceive people is still under research. At present, the Prism/Spectrum project is also using Uniswap V4 hooks for option style index tracking, which is basically based on this idea. In summary, DeFi should not be a "faster version of traditional finance". Instead, we should fundamentally change our approach: using "options+slow oracle" to bring smoother and more attack resistant risk design. Although there are challenges in rebalancing, it is worth trying.