12 KiB
12 KiB
Privacy scoring modelling > Web3privacy now analytical platform
Approach
| Phase | Description |
|---|---|
| 1. Expert take | Outreach privacy experts behind core privacy services - aggregate their professional opinions on how to analyze if a service is truly private or not.) |
| 2. Scoring model prototyping | Create an open & flexible scoring model for a communal feedback loop - share with the privacy community, evaluate. |
| 3. Scoring model MVP release | Deliver balanced model for privacy services assessment - powered by pros & general public opinions. |
Current status (02/04/2023)
0. 350+ privacy solutions in 1 database - delivered
1. On-going community research (survey) within the privacy experts.
I asked experts behind privacy-services or contributors to the privacy-centric communities to share their visions on analysing whether a service is private. Answers were collected via chats & survey form
Criteria:
- min 50 different experts
- a broad range of positions: tech, ops, marketing, devrel, strategy
- a broad range of services: from privacy coins to mixnets
- different geographies: from the USA to Russia
Questions from the privacy experts
Take 1
- does it has traceability? (ie addresses is hidden from the public ledger)
- does it has unlinkability? (ie transactions can't be linked to each other)
- does the amount of transfer is hidden?
- does IP addresses of participants hidden?
- is it decentralized and based on open-source technology?
Take 2
| Question | Observation |
|---|---|
| much the users in control of their data disclosure? | (Scale 1-10) |
| how well community feedback and evaluation is built into product dev? | the less -» the more centralized it is -» the smaller the % of it staying private without collective intelligence. This is like the web3privacynow - platform part, actually, for sales, but also I found this really relevant. |
| is there a community bug/security bounty program/platform? | yes, no |
| how much transparent disclosure is available on the tech and company | like smart contract audits, security audits, source of financing? |
| how private the tech stack it uses on all layers. from hardware to l3/l4 etc. | how well it is disclosed what they built on and where they host stuff, or if the tech is decentralized like nym - is there available dashboard data about this? |
| product roadmap and release flexibility - this is a harder one, and I'm not sure it makes sense. What I mean is it's also important to have a clear vision while reacting to current needs/bugs /fixing vulnerabilities. | maybe its redundant with no2 and no2b |
Take 3
- What are the trust assumptions the user has by using the platform?
- What and how is user information stored and transmitted?
- How much PII is stored/collected?
- How is information collected + processed + disseminated
- How completely can you participate with total privacy?
Take 4
| Direction | Observation |
|---|---|
| network privacy | how do you connect to the chain? Can you do it via Tor? |
| blockchain privacy | do the resulting on-chain transactions offer the user any on-chain obfuscation? |
| censorship resistance | how resistant is the project to external pressure? Will the project censor you? |
| permission | do I need to create an account to access the thing, or is the thing open access? |
| custody of funds | is the user out of control of their keys at any point? |
Answers from the privacy experts
Additional lenses: is it accessible to a non-web3 person & is it accessible to a non-tech web3 person?
Contents
- General
- Docs
- Third-party analysis
- VCs
- Team
- Privacy policy
- Token
- Infrastructure
- Storage
- Data aggregation
- Traction
- Governance
- Privacy execution
- Product-centric
- Testing
- Other
General
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| Immutability | - | - |
| Decentralised throughout, including hosting | - | - |
| Permissionless & accessible to all | - | - |
| Open-source | + | + |
Docs
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| read the documentation | - | - |
| Good and comprehensive documentation | - | - |
Third-party analysis
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| ask about its weaknesses from competitors | + | + |
| Number of peer-reviewed articles at conferences and journals of team members | + | + |
| Where's the code? Has it been audited? | + | + |
| Validation by trusted and respected independent scientists and researchers | + | + |
VCs
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| Who are the VCs | - | - |
| Not funded by big US VCs like a16z | - | + |
Team
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| ideological team | - | + |
| Reputation of the team | - | + |
| is it purely marketing oriented, or it seems created by researchers/developers, are the developers anons? | + | + |
Privacy policy
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| Privacy Policy content | + | + |
| Non-vague and non-intrusive privacy policy | + | + |
| #privacy protection policies | + | + |
Token
| Scoring | Non-web3 person assesment | Web3, but non-tech assesment |
|---|---|---|
| is there a token since the beginning? | - | + |
| if the token since beginning - weird | - | + |
Infrastructure
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| How much to run a node | - | + |
| Where are the nodes | - | + |
| Number of nodes/servers/ -> the larger the footprint the best privacy | - | + |
Storage
| Scoring | Non-web3 person assesment | Web3, but non-tech assesment |
|---|---|---|
| e2e encrypted LOCAL storage | - | + |
| What user information is stored? (username, IP address, last connection, wallets associate, etc) | - | + |
| Where is it stored? (centralized server, certain jurisdictions, on-chain, in browser/local cache) | - | + |
Data aggregation
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| no email or tel nr for signup | + | + |
| control over personal data | - | - |
| does not implement KYC or AML | + | + |
| Metadata privacy / Minimal to no metadata capture | - | - |
Traction
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| Amount of transactions | + | + |
| number of people using it | + | + |
| is it famous | + | + |
| Latency | - | - |
| Time of test and battle-tested code - (e.g. how BSC had passed the stress time of withdrawals with FTX drama or crypto schemes such as ECDSA with more than 2-3 decades alive) | - | - |
| Cost | - | + |
Governance
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| DAO structure (if applied) | - | + |
Privacy execution
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| How is it being transmitted? (encrypted, unencrypted, offuscated, etc) | - | - |
| Combined those encryption methods effectively (holistic solution) | - | - |
| Confidentiality of transactions | - | - |
| the ability to hide transactional data from the public | - | - |
| strong encryption algorithms | - | - |
| If the speed in connection is too fast, there most probably no privacy there and rather a direct channel between user - app | - | - |
| p2p / no central server | - | - |
| Trustless - No ID required (this is where ZKs are useful) | - | + |
| Usage of ZK | - | - |
Product-centric
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| Onboarding steps | + | + |
| Usability - for end users or in the developer experience if it is a B2B project. | + | - |
Testing
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| Ability to run part of the service and verify for myself | - | - |
| try to trace a transaction | - | - |
| There is a way to verify the code I think is running, really is running e.g. attestation service | - | - |
| Other tooling to verify e.g. block explorers | - | + |
Other
| Scoring | Non-web3 person assesment | Non-tech assesment |
|---|---|---|
| Entropy (non-trivial to estimate, different measurements for type of service). Some examples: https://arxiv.org/abs/2211.04259 or https://blog.nymtech.net/an-empirical-study-of-privacy-scalability-and-latency-of-nym-mixnet-ff05320fb62d | - | - |
| Censorship-resistant (how hard it's for a powerful party to block/censor a given service) | - | - |
| Precise description of the concrete privacy properties. Privacy is complicated, so if they don't say exactly what they protect, then its likely vapour | - | - |
| Doesn’t sell your data | - | - |
| protects against a global passive adversary | - | - |
| strong secure anonymity tech | - | - |
| Credibly neutral | + | + |
| ISO/IEC 29190:2015: https://www.iso.org/standard/45269.html | - | - |
| Anonymity Assessment – A Universal Tool for Measuring Anonymity of Data Sets Under the GDPR with a Special Focus on Smart Robotics: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3971139 | - | - |
Huge thanks everyone who contributed! I make it anon now, but will thank everyone (who would liked to be credited) once a scoring model will be published on GitHub for community evaluation.
2. My personal notes on privacy scoring (they were made before communal survey)
Sketches what could be put inside privacy-solutions scoring model (note: think of these as questions to experts for a workshop on scoring ideation).
Key observations
| Topic | Observation |
|---|---|
| Broad range of different takes on privacy assesment | Privacy experts have around 50+ tips |
| Tech-centricity of assesment | Majority of the expert takes are hard to execute by non-tech people (they need info-help!) |
| Privacy assessment takes enormous time | Time-To privacy-fit - potential for analytical service |
| Privacy literacy isn't enough | The scoring model demand both "decentralisation", "open-source" & "privacy" topics understanding |
| Mix of objective & subjective takes | Scoring criteria are different from objective (example: transaction traceability) & subjective (example: backed by a16z crypto) takes |
Open-source transparency
- GitHub repos: # of commits, # stars, date of repo creation.
Third-party validation
- Security audits: yes, no; type of audit; ammount of audits.
Community validation
- Existing bugs
- White hackers assessment (like Secret Network TEE bug)
- Negative Discord, Twitter, other public feedback (product & founder-centric)
Team
- Market validation
- GitHub contribution
- Track record (incl. red flag projects)
Financials
- Investments
- TVL (like Aztec's L2)
- Donation-based
- Public treasury
Liveliness
- How active is GitHub activity
- How active is the community
- Is there public product traction?
Product-readiness
- State of product-readiness
- MVP-readiness
- Protocol (test-net/main-net)
- dApp (release timing, third-party validation like AppStore/Play Store)
- network-reliability (the state of privacy in Ethereum, Solana, Avalanche etc)
Cross-checked data leakage
- Complementing privacy stack data leakage (example: phone + dApp; wallet + RPC etc)
- Third-party data leakage (from the hackers to state agents (think of Iran or North Korean govs))
Data aggregation policies
Reference: https://tosdr.org
Centralisation level (incl KYC)
Reference: https://kycnot.me/about#scores