SafetyScore

Transcript

1. SafetyScore protecting lives, livelihoods, and liberty
2. • As a result of globalization, new pandemics have the potential to spread like wildfire around our planet • The last decade has seen an alarming increase in the number of new infectious diseases cross over from animals to people • Until we address the root causes (food supply chain, urbanization, biodiversity loss, etc.), we need to be prepared for new pandemics Threat of Pandemics
3. • The economic impact of pandemics can be devastating. Millions unemployed and many more facing an uncertain future • Unless we make major breakthroughs in medical science, it will always take time to create effective treatments and vaccinations • Many businesses will simply not be able to survive another wave of lockdowns in the face of a new pandemic Economic Challenges
4. SafetyScore SafetyScore is a health passport designed to address these challenges while maintaining privacy. It can keep people safe and free while keeping the economy and critical industries open.
5. SafetyScore 11 63 85 67 21 99 71 74 73 82 60 68
6. • SafetyScore calculates a score for everyone based on the encounters they've had with potentially infected people • Authorities can verify an individual's SafetyScore by simply scanning a QR code on the user's app • Being able to assess people's risk levels gives countries a powerful tool to contain epidemics, save lives, and keep economies alive SafetyScore Health Passport Scanning ... 14:18 Verified. Entry Granted 14:18 92
7. • Places like universities, offices, restaurants, and bars can operate safely by simply checking people's SafetyScore before letting them in • This helps to keep premises safe. Governments can regulate the required Safety Levels on a regional basis depending on the premise type • SafetyScore empowers communities to self- manage their risk so that local economies can function while keeping people safe Keeping Businesses Open SafetyScore 11 63 85 67 21 99 71 74 73 82 60 68
8. Traffic Light System 26 13:27 Nov 22 13:27 45 18:11 Nov 9 18:11 SAFE ON ALERT SELF-ISOLATE Unrestricted Access Restricted Access at Large Events Regulated Access 10:52 88 10:52 Oct 5
9. SafetyScore Air Bridges • SafetyScore enables people to visit other Air Bridge countries without the need for quarantine • Governments can limit their exposure to foreign infection risk by checking people's SafetyScore before letting them in • Governments specify the Air Bridge countries that they trust, and this will be factored into the verification process
10. How It Works
11. SafetyScore SafetyScore takes a radically different approach to first-generation contact tracing apps built on the Apple/Google framework
12. • SafetyScore uses Bluetooth on mobile phones to detect others that come in close proximity to a user • Unlike first-gen apps which broadcast random IDs over Bluetooth, SafetyScore broadcasts cryptographic public keys • This allows us to establish secure communication channels between users that will remain private from others Proximity Detection 10:42 11:15 13:28
13. SafetyScore distributes risk Patient confirms diagnosis Health official checks patient First Degree Contacts Second Degree Contacts 82 -10 71 -20 11 -9 15 -6 85 -2 09 -17 • Health officials certify when someone is infected. This propagates risk to the contact chain of the infected person, updating SafetyScores • Unlike first-gens apps that only notify direct contacts, SafetyScore also propagates risk to indirect contacts, improving accuracy • If a person's score drops below a threshold, governments can use this to prioritize access to testing, treatment, and benefits Risk Propagation
14. FIRST-GEN APPS Exposure notification to direct contacts only !
15. 99 71 68 82 71 11 15 09 23 23 -21 -1 -2 -5 -10 -1 -14 16 ! 85 -4 -7 -17 -6 -2 SAFETYSCORE Fine-grain risk propagation to both direct and indirect contacts
16. • A range of factors are brought together to make an individual's SafetyScore as accurate as possible while maintaining privacy • The score includes risk exposure from direct and indirect contacts, recent test results, and the Geo-Regions the individual has visited • To minimize cheating, the score factors in user behavior, e.g. that they haven't turned the phone off SafetyScore Calculation SafetyScore 1st, 2nd Degree Risk Exposure Geo-Region Activity Profile Risk Profile Test Results (Immunity, Negative Tests)
17. • Most contact tracing apps have seen adoption rates of less than 20% — resulting in less than 4% of contacts being traced • In contrast, the SafetyScore health passport is able to work effectively at the local community level, and can thus grow to scale organically • The ability to safeguard spaces aligns with the operational and liability concerns of universities and businesses, further accelerating adoption Designed for Growth
18. • SafetyScore has been designed from the ground up to maximize a user's privacy and to protect them from being tracked • User activity is obscured through multiple cryptographic layers and constantly changes so as to limit the potential for mass surveillance • SafetyScore builds on established primitives such as cryptographic blinding to obscure the social interactions between people Privacy First
19. • SafetyScore nodes act as an API server for clients like mobile apps. Collectively, these nodes form a peer-to-peer network • Since they are not controlled by a single entity, it protects the system from malicious parties such as foreign agents • While we expect governments to run nodes to serve their population, anyone will be able to run one to further improve network security Decentralized Protocol Distribution Commitments Risk Account Day 7 Notary 1 Notary 2 Notary 3 A 305 B 305 B 305 B 306 C 306 D 306 E 306 B 307 1 2 3 4 7 6 5
20. • SafetyScore is built around a distributed ledger which serves as a high-integrity, shared database without a single point of control • Our distributed consensus algorithm is able to scale to meet the high throughput needs of SafetyScore • To protect against resource exhaustion attacks and spam, nodes will only gain the right to write to the ledger by acting reliably over time Distributed Ledger Signature for Block 5 Signature for Block 6 Block Reference Block Reference Block Reference Block Reference Block Reference Entry Entry Entry Entry Entry Entry Entry Entry Entry Entry Entry Entry Entry Entry Entry Entry Entry
21. • When health officials certify infections, they make public some anonymized data, e.g. type of disease, test method, location, etc. • Epidemiologists can use this to track disease spread and can act as an early warning system for new infectious diseases • Health authorities can use this to prioritize resource allocation and co-ordinate responses both locally and globally Epidemic Response Hub
22. • SafetyScore is being developed as an open source project. This helps us maximize transparency and build a global community • The SafetyScore protocol will be an open standard so that others can build on it, e.g. hardware devices, software services, etc. • We will also release an extensible mobile SDK so that others can integrate SafetyScore into their existing apps and services Open Source
23. • SafetyScore’s protocol is flexible enough to address all contagious diseases, whether it's the seasonal flu or a new pandemic • It can be continuously updated to address new scientific data, e.g. new strains, new diseases, our understanding of immunity, etc. • SafetyScore can self-learn and adapt the score calculation to changes in disease transmission and social behavior Future Proof
24. • SafetyScore primarily accounts for transmission via people and is limited in its ability to track transmission via surfaces • While smartphone adoption is high enough in urban areas, where most transmission happens, it might be an issue in certain countries • SafetyScore will use about 30MB of mobile data every day. This may prove to be costly for users in some countries Limitations
25. • The targeted approach of health passports might feel unfair, but we believe this is better than blanket lockdowns • Governments will be able to help ensure fairness by making special allowances for key workers, vulnerable citizens, and essential services • SafetyScore will make visible to users the various factors that are affecting their score so as to maximize transparency and trust Fairness
26. SafetyScore Health Passport & Tracing Apple/Google Contact Tracing Decentralized & Privacy-Preserving ✓ ✓ Exposure Notification ✓ ✓ Verifiable Risk Score ✓ ✗ Second Degree + Contacts ✓ ✗ Designed for Growth ✓ ✗ Air Bridges ✓ ✗ Comparison
27. International Team
28. Assembled from World-Leading Experts ✓ Public Health ✓ Epidemiology ✓ Privacy ✓ Cybersecurity ✓ Distributed Systems ✓ Open Source ✓ Human-Centered Design
29. Tav Siva Social entrepreneur, distributed systems engineer, open source developer. Chainspace, UK Home Office, WikiHouse, Freenet Tom Salfield Social-tech entrepreneur, collaborative systems designer. Wikifactory, Impact Hub, Opencoin, Imperial, LSE Dr. Luke Robinson Deep tech entrepreneur, quantum physicist. Post Urban Ventures, Hazy, Oxford, Cambridge Dr. Oliver Zahn Ex-Head of Data Science at Google. Tech entrepreneur. SpaceX, Impossible Foods, UC Berkeley, Harvard Dr. Laura Kahn Co-founder at One Health Initiative. Public health expert, physician, science and global security scholar. Princeton Dr. David Spergel Director at Simons Foundation. Computational astrophysicist. Breakthrough prize winner. Princeton, NASA, Harvard Imtiaz Shams AI entrepreneur, non-profit builder, fundraising. Post Urban Ventures, Charities: Think & Faith to Faithless, LSE Alice Fung Director at Health Foundry. Top 100 Women in Social Enterprise. Architecture 00, Mayor's Design Advocate, Impact Hub Hinesh Mandalia Senior full-stack engineer, fintech. G-Research, Lloyds Banking Group, Logica, Imperial William Pritchard Senior QA engineer, automation tester, data security, AI. Genomics England, UK Cabinet Office, Capgemini Krishna Kotecha Ex-iOS team lead at Babylon Health. Award-winning mobile developer. Hailo, Deutsche Bank, Symbian Ramesh Nair Full-stack polyglot engineer, blockchain enthusiast. Nayms, TokenCard, Ethereum Foundation, Imperial Alex McKinlay Senior software engineer, distributed systems, fintech, smart contracts. Refinitiv, TokenCard, Pantheon Ventures Huy Dinh Experienced CTO, scalable systems architect, app developer. NTT Docomo. Guardian News, BBC Worldwide Dr. Nina Fefferman Infectious diseases modeling expert. Professor in Ecology, Evolution & Maths at UT Knoxville. Rutgers, Tufts, Princeton Anthony Leung Senior finance, treasury, and strategy executive. Tesco, Credit Suisse, Morgan Stanley, LSE Seyi Ogunyemi Senior full-stack engineer, graphic designer, UX specialist, fintech. Civic, Espians, Coca-Cola, Diageo
30. How It Compares
31. We built an agent-based simulation tool to test our assumptions and see how SafetyScore compares
32. The baseline, without any apps or other interventions, results in 80% of the population infected for the median case
33. For an Apple/Google-based app, even assuming a high adoption rate of 60%, we see 73% of the population still get infected
34. In contrast, with SafetyScore, only 18% of the population get infected and 1% require self-isolation
35. This is almost as effective as lockdowns, where 7% get infected, but without the economic impact of isolating 83% of the population
36. SafetyScore These simulations show that SafetyScore can keep people healthy and economies open at the same time
37. SafetyScore is 3x better than Apple/Google-based apps for preventing infections and 48x better than lockdowns in terms of time spent in isolation
38. Appendix
39. Simulation Tool Try out our simulation tool: https://simulation.safetyscore.app
40. Read our whitepaper: https://safetyscore.app/whitepaper Whitepaper
41. Establishing Contact Keys Save Contact Key from X Save Contact Key from Y Broadcast Key Broadcast Key X Y Derive Shared Secret Derive Temporary Contact Key for Y Derive Temporary Contact Key for X Distributed Ledger Post Contact Proof to Ledger Post Contact Proof to Ledger Post Contact Key for Y Post Contact Key for X Aggregate proofs from X corresponding to the same Activity Period Aggregate proofs from Y corresponding to the same Activity Period
42. Layered Identity A 1 A 2 A ... A 48 R 1 R 2 K 1 K 2 K 3 K 4 K 5 K ... K ... K 118 K 119 A 49 A ... K 120 K 122 K 123 K ... K ... K ... Broadcast Keys Activity Keys Daily Risk Accounts
43. Distributed Ledger Node C Node B Node A Tick 1 Block A1 Block B1 Block C1 Block A2 References A1 Block C2 References A1, B1, C1 Block A3 References A2, B1, C1, C2 Block B2 References A1, B1, C1, C2 Block B3 References A2, A3, B2 Block C3 References A2, A3, B2, C2 Block A4 References A3, B2 Tick 2 Tick 3 Tick 4
44. Node Relays Legend: Path 1 Path 2 Node Relay Path Relay Start Relay Termination & Publication
45. Risk Distribution Flow A 5 A 6 A 7 A 8 A 9 A 10 B 5 B 6 B 7 B 8 B 9 B 10 E 5 E 6 E 7 E 8 E 9 E 10
46. Risk Distribution Tree A 6 B 6 C 6 D 6 F 10 F 12 G 12 H 10 I 12 J 12 Distribution Implicit Distribution Activity Key Legend: C 11 B 8 B 9 D 12 F 11 F 9 E 8 J 7 Connection
47. Links 1. Bidirectional Contact Tracing is Required for Reliable COVID-19 Control 2. The UK's Contact Tracing Plan Has Two Fatal Flaws 3. Effective Configurations of a Digital Contact Tracing App: A report to NHSX 4. Contact-Tracing Apps Aren’t the Technology Panacea People Hoped 5. Exposure Notifications: Using Technology to Help Public Health Authorities Fight COVID-19 6. A Socio-Technical Framework for Digital Contact Tracing 7. Longitudinal Evaluation and Decline of Antibody Responses in SARS-CoV-2 Infection
48. SafetyScore connect@safetyscore.app