# Mechanism Design ### Key Points from ["Counterspeculation, Auctions, and Competitive Sealed Tenders" by William Vickrey](https://www.cs.columbia.edu/coms6998-3/lecture1-auctions.pdf): 1. In a homogeneous auction, the progressive auction and Dutch auction are both Pareto-optimal and have the same expected price, but the Dutch has lower variance. 2. In a nonhomogeneous auction, the progressive is Pareto-optimal but the Dutch is manipulable and isn't always Pareto-optimal. 3. In both cases, the expected price of the progressive auction is the second-highest bid, and is Pareto-optimal so the winner (the highest bidder) is the highest valuer. 4. Why not directly jump to this optimal result through a sealed-bid auction where the price is determined as the second-highest bid? 5. **A sealed-bid second-price auction makes the optimal strategy for maximizing earnings be: Bid exactly the value of the object to oneself.** 6. As long as the highest bidder and 2nd-highest price are published transparently, and the seller and top bidder and 2nd-highest bidder don't collude to set the price to the 3rd-highest bid, **the optimal allocation of resources is guaranteed**. 7. Moreover, the optimal allocation of resources is achieved without costly (both time- and resource-wise) market appraisal and manipulation. ### Key Takeaways for Application to Funds Allocation: 1. Organization members (**funders, workers**) must determine how organization funds are allocated to projects (**projects**). 2. Project members (**workers**) must determine how project funds are allocated to project members (**subprojects**). This can be applied recursively to **projects/subprojects/.../milestones/tasks/subtasks/contributors**. **Q: How do we guarantee the optimal allocation of resources in this case?** **A:** [Continuous Voting](https://hackmd.io/@themathematicianisin/SyLhKHCWS)