Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Token Design as Optimization Design

2,324 views

Published on

The core feature of tokenized ecosystems, aka public blockchains, is getting people to do stuff. In this talk, I give more structure to this idea using a framing from optimization literature, and more precisely, evolutionary algorithms (EAs). I give examples of this approach using Bitcoin and Ocean Protocol as examples.
Link to video: https://www.youtube.com/watch?v=Sm8j0u5NuGQ

Published in: Technology

Token Design as Optimization Design

  1. 1. Token Design as Optimization Design Trent McConaghy @trentmc0
  2. 2. Outline • Blockchains as trust machines • Blockchains as incentive machines • Case study: Bitcoin • Towards a token design practice • Case Study: Ocean Protocol • Getting objective function right really matters
  3. 3. Blockchains as Trust Machines
  4. 4. Blockchain data structure = chain of blocks -Block = list of transactions, where tx = “create asset” or “transfer asset” action, digitally signed -Chain = linked list, where links are hashes Header Tx1 Tx2 Tx3 .. Header Tx1 Tx2 Tx3 .. Header Tx1 Tx2 Tx3 ..
  5. 5. Blockchains as Distributed DBs, with 3 new characteristics • Decentralized: via Byzantine fault tolerant (BFT) consensus • Immutable: do undo a tx, need to undo each block in chain • Assets: digital signature on every transaction. Create, transfer. Alice Mongo DB
  6. 6. From Permissioned  Permissionless Blockchain • Permissioned: to be a server node, need to be on an approved list • A classical BFT setting • “1 public key = 1 vote” • Permissionless: anyone can join as server node • Need BFT and Sybil tolerance (to “attack of the clones”) • E.g. via “1 electron = 1 vote”
  7. 7. “Trust machine” because it minimizes trust needed to operate. It’s more socially scalable. (Ref Szabos)
  8. 8. Blockchains as Incentive Machines
  9. 9. What’s the amazing thing about blockchains? •Decentralized? •Immutability? •Assets •One more…
  10. 10. Blockchain Superpower: Get people to do stuff By rewarding with tokens
  11. 11. “I think I've been in the top 5% of my age cohort all my life in understanding the power of incentives, and all my life I've underestimated it. Never a year passes that I don't get some surprise that pushes my limit a little farther.” -Charlie Munger
  12. 12. Economic Incentive for Bitcoin Objective: Maximize security of network • Where “security” = compute power • Therefore, super expensive to roll back changes to the transaction log
  13. 13. Economic Incentive for Bitcoin Objective: Maximize security of network • Where “security” = compute power • Therefore, super expensive to roll back changes to the transaction log E(Ri) α Hi * T E() = expected value # tokens (BTC) dispensed each block block rewards hash power of actor = contribution to “security”
  14. 14. E(Ri) α Hi * T Bitcoin Token Release Schedule F(H, t) = 1 - (0.5t/10) = % tokens released after t years • Schedule is fixed in advance • 4 years for 50% of tokens released = half-life “the halvening”
  15. 15. Result of Bitcoin maximizing security? Maximizing energy usage! > USA by mid 2019!
  16. 16. Convergence of Bitcoin (Hash rate vs time)
  17. 17. Towards a Token Design Practice
  18. 18. Formulation of an optimization problem
  19. 19. Canonical formulation of optimization problem 0 or more objectives, inequality constraints, and equality constraints
  20. 20. Apply or design an optimization algorithm that’s appropriate to the optimization problem
  21. 21. Convergence of the optimization algorithm against the objective function (and constraints) If it doesn’t converge or converge well enough: try new algorithm
  22. 22. Design of Tokenized ecosystem as Design of EAs (Evolutionary Algorithms) What Tokenized ecosystem Evolutionary Algorithm Goals Block reward function E.g. “Maximize hash rate” Objective function E.g. “Minimize error” Measurement & test Proof E.g. “Proof of Work” Evaluate fitness E.g. “Simulate circuit” System agents Miners & token holders (humans) In a network Individuals (computer agents) In a population System clock Block reward interval Generation Incentives & Disincentives You can’t control human, Just reward: give tokens And punish: slash stake You can’t control individual, Just reward: reproduce And punish: kill
  23. 23. Design of Tokenized Ecosystems = Mechanism Design Analysis: Synthesis: Game theory Mechanism Design Optimization Design Practical constraints
  24. 24. Other labels for design of tokenized ecosystems: Mechanism Design Tokenomics Crypto-economics Financial cryptography Token engineering Incentive engineering Economics Finance Electrical Engineering Control systems Cybernetics Computer science Distributed systems Behavioral psychology Game theory Optimization Related fields: Complex systems AI
  25. 25. Agent-based Systems for Token Simulation • Q: How do we design computer chips? • A: Simulator + CAD tools • Q: How are we currently designing tokenized ecosystems? • A: By the seat of our pants! What we (desperately) need: 1. Simulators: agent-based systems 2. CAD tools: for token design (Alas, we must be patient…)
  26. 26. Application to A Future Tokenized Ecosystem
  27. 27. 1000x more data The Unreasonable Effectiveness of Data
  28. 28. Mo’ data (and mo’ compute) Mo’ accuracy Mo’ $
  29. 29. Here’s your personal data
  30. 30. A new data economy Have lotsa AI (1000 AI startups) Have lotsa data (1000 enterprises) DM DM DM DM DM DM DM DM Ocean
  31. 31. Ocean goal: maximize supply of relevant data Token rewards if: supply data, and curate it
  32. 32. Economic Incentive for Ocean Objective: Maximize supply of relevant data • This means: reward curating data + making it available • Where “curating” = betting on data. Reward taste-making. E(Rij) α log10(Sij) * log10(Dj) * T *Ri Expected reward for user i on dataset j Dj = proofed popularity = # times made dataset available Sij = predicted popularity = user’s curation market stake in dataset j # tokens during interval
  33. 33. From AI data to AI services Motivations: • Privacy, so compute on-premise or decentralized • Data is heavy, so compute on-premise • Link in emerging decentralized AI compute Objective function: Maximize supply of relevant services =reward curating services + proving that it was delivered E(Rij) α log10(Sij) * log10(Dj) * T *Ri proofed popularity of service predicted popularity of service
  34. 34. Ocean is a network of curated services. An AI services pipeline. Availability Consumption Privacy GovernanceProduction commons Inter- Operability Discovery *Note: logos shown are examples and do not imply partnerships or integrations
  35. 35. Self-driving cars: fewer accidents, more mobility
  36. 36. >100x more data for health care research
  37. 37. Erode the data silos
  38. 38. Erode the data silos
  39. 39. Erode the data silos
  40. 40. Why Getting The Objective Function Right Really Matters
  41. 41. The Paperclip Maximizer (Nick Bostrom, 2003) Suppose we have an AI whose only goal is to make as many paper clips as possible. The AI will realize quickly that it would be much better if there were no humans because humans might decide to switch it off. Because if humans do so, there would be fewer paper clips. Also, human bodies contain a lot of atoms that could be made into paper clips. The future that the AI would be trying to gear towards would be one in which there were a lot of paper clips but no humans.
  42. 42. The Paperclip Maximizer (Nick Bostrom, 2003)
  43. 43. Blockchains as Life (Merkle) -Ralph Merkle, “DAOs, Democracy and Governance”, May 2016
  44. 44. Blockchains as Life (Merkle) -Ralph Merkle, “DAOs, Democracy and Governance”, May 2016
  45. 45. Result of Bitcoin maximizing security? Maximizing energy usage! > USA by mid 2019! A life form optimizing maniacally, but for energy (i.e. the thing we fight wars over)
  46. 46. Conclusion
  47. 47. Trent McConaghy @trentmc0 • Blockchains as trust machines • Blockchains as incentive machines: you can get people to do stuff! • Case study: Bitcoin • Towards a token design practice, using optimization & more • Case Study: Ocean Protocol • Getting objective function right really matters Conclusion

×