Point of no return: May 9, when the mint-burn arbitrage mechanism began hyperinflating LUNA supply. Once LUNA's market cap fell below UST's outstanding supply, the peg mechanism became mathematically impossible to restore.

Primary Mechanism Design Flaw

The UST/LUNA mint-burn mechanism was reflexive: UST could be redeemed for $1 worth of newly minted LUNA, and LUNA could be burned to mint UST. This created a circular dependency where UST's stability depended on LUNA's market cap, and LUNA's value derived from demand for UST. When confidence broke, the mechanism became a positive feedback loop accelerating collapse rather than restoring equilibrium.

Incentive Misalignment

Individual rationality led to collective harm through two channels:

• Arbitrageurs: Each arbitrageur rationally redeemed UST for $1 of LUNA and immediately sold LUNA. This was individually profitable but collectively diluted LUNA supply to worthlessness, destroying the collateral backing UST.
• Anchor depositors: The 19.5% APY attracted capital that was not organically generated -- it was subsidized by reserves. When subsidies depleted, rational withdrawal triggered the initial depeg.

Nash Equilibrium Analysis

The collapse was a Nash equilibrium: once UST depegged, no individual actor had incentive to hold UST or LUNA. Selling was a dominant strategy for every holder. Even if all holders coordinated to hold, any individual defector profits by selling first -- a classic prisoner's dilemma at scale.

Trust Assumption Violation

The mechanism assumed that LUNA would always maintain sufficient market capitalization to absorb UST redemptions. This assumption failed under stress because LUNA's value was itself derived from UST demand -- a circular dependency with no external anchor.

Point of no return: November 7, when FTX halted withdrawals. This confirmed that customer funds were not available, converting fear into certainty of insolvency. Once a bank run reaches withdrawal halts, recovery is essentially impossible.

Primary Mechanism Design Flaw

FTX operated as a centralized custodian with full discretion over customer deposits. There was no on-chain transparency, no proof-of-reserves mechanism, and no technical constraint preventing the misuse of funds. This created an extreme principal-agent problem: customers (principals) trusted FTX (agent) to custody funds, but FTX had both the ability and incentive to misappropriate them.

Incentive Misalignment

• Moral hazard: SBF and Alameda could take excessive risks with customer funds because downside risk was borne by depositors, not management. The absence of monitoring created a textbook moral hazard.
• FTT token circularity: FTX used its own FTT token as collateral for Alameda loans -- a reflexive arrangement where the collateral's value depended on the entity that issued it. When confidence broke, the collateral became worthless.
• Bank run dynamics: Once CoinDesk revealed the balance sheet, each depositor's dominant strategy was to withdraw immediately. The collective rush made insolvency certain even if the initial shortfall was recoverable.

Nash Equilibrium Analysis

Post-revelation, the bank run was a Nash equilibrium: each depositor does best by withdrawing regardless of what others do (withdrawing is a dominant strategy). Even depositors who "trusted FTX" had no rational reason to leave funds on the exchange once information asymmetry was resolved.

Trust Assumption Violation

Users assumed FTX was solvent and holding deposits 1:1. The mechanism (centralized exchange with no proof-of-reserves) provided no way to verify this assumption. The violation was not a mechanism design failure per se, but rather the absence of any mechanism -- trust was based on reputation alone, with no cryptographic or economic enforcement.

Point of no return: June 17, when the exploit occurred. Although funds were recovered via hard fork, the event permanently demonstrated that smart contract vulnerabilities can be catastrophic, and that "code is law" has practical limits.

Primary Mechanism Design Flaw

The DAO had two compounding flaws:

1. Smart contract vulnerability (reentrancy): The splitDAO function sent ETH before updating state, enabling recursive withdrawal. This is a technical bug, but it reflects a mechanism design failure: the system did not account for adversarial execution ordering.
2. Governance inadequacy: Despite public security warnings 3 weeks before the attack, the DAO had no emergency mechanism to patch the vulnerability or pause operations. The governance mechanism was too slow for security-critical decisions.

Incentive Misalignment

• Speed vs. security: The DAO launched quickly to capture momentum, but this meant insufficient security review. Developers were incentivized to ship fast (first-mover advantage) rather than ship safely.
• Attacker incentives: The vulnerability created a dominant strategy for any sufficiently skilled actor: exploit now, before it is fixed. The 28-day delay on splits created a window, but the attacker bet (incorrectly, as it turned out) that the community would not coordinate a response.
• "Code is law" ideology: The DAO community assumed that code execution was the social contract. This created a moral hazard for developers (if the code is always right, there's less incentive to audit thoroughly) and left no recourse for participants when the code behaved in unintended ways.

Nash Equilibrium Analysis

Pre-fork, the attacker's strategy was a best response: exploit a known vulnerability for $60M with (perceived) impunity. The community's response -- coordinating a hard fork -- was not a Nash equilibrium in the traditional sense because it required unprecedented coordination. The fork succeeded because the Ethereum community was small and cohesive enough to coordinate (a Schelling point formed around "return the stolen funds").

Trust Assumption Violation

The DAO assumed its smart contract code was correct ("code is law"). This assumption was violated when the reentrancy bug was exploited. More broadly, the assumption that governance could operate at the speed of code execution proved false.