Email, Messaging, and SSI/DID

discussion slides @ IIWXXX

Transcript

1. Email, Messaging, and SSI/DID Ryo Kajiwara @ IIWXXX, 2020/04/30
2. Note • This is a presentation of a preliminary idea, which means: • This is NOT a demonstration of a product in development • This only outlines ideas for discussion • There may be flaws in the logic / assumptions that I am making • There just might be right solutions out there! • It's my first IIW, and I heard there has been lots of discussion on Email before • Language issues may/will happen
4. TL;DR We want better Email, or an alternative to Email Can SSI/DID help...?
5. Topics to discuss • Why Email is still relevant, Why we need messaging • Features we want for messaging, and how current-day solutions are lacking them • End-to-End encryption • Encrypted group communication • Control of data • Control of your identity
7. Q: Why use Email when there's WhatsApp / Facebook Messenger / Signal ...
8. Q: Why use Email when there's WhatsApp / Facebook Messenger / Signal ... • SMTP is an archaic protocol without proper encryption and authentication • Email does not have End-to-End encryption and encrypted group communication • Email has spam
9. Q: Why use Email when there's WhatsApp / Facebook Messenger / Signal ... A: You can receive email from people without pre-established trust
10. Q: Why use Email when there's WhatsApp / Facebook Messenger / Signal ... This can be achieved by some messaging services, but under an assumption that both parties already have an ID on the same messaging service. • Some people use Facebook for personal use only. Some don't trust Facebook at all... • LinkedIn is popular among business people but may not be popular among academics
11. But Email has spam!
12. Email has spam because of its inherent anonymity
13. Anonymity of Email The same properties (no need for pre-established trust) applies to telephone networks, but email lacks an effective anti-abuse mechanism built into the protocol. This is due to email's anonymity. If you abuse: • the telephone network: You may be caught due to reverse detection • email: There are many easy ways to spoof your identity, making the other side hard to catch you
14. Email abusers (spammers) use email's inherent anonymity to their advantage
15. Do email receivers really want anonymous email? Anonymous email have a high chance of being spam
16. Okay, enforce S/MIME then ...?
17. Problems with S/MIME • Cost of issuance • Yes, money cost • Bound to a single context • One certificate might prove you belong to a certain organization • But you might not want to use that hat all the time • Multiple certs? Go back to top
18. Initial idea: Always trust email with signatures from government-issued IDs1 1 We (kind of) have something like this in Japan (ެతݸਓೝূ)
19. Nobody spams with a government-issued ID, right...?
20. I assume everyone here is aware of the problems of centralized IDs...
22. Email and its "Self- Sovereign-ness" Email(SMTP/POP/IMAP protocol) is designed to be self-sovereign (you can self-issue your ID, you have control of your data), as long as you can set up your own server Nobody do that these days because ... • SMTP: Authentication is difficult, single misconfiguration results in sending of spam • IMAP: Multi-device access, Storage and backups
23. Email and its "Self- Sovereign-ness" As such, we are giving up control of personal messages to Email service providers (mostly Gmail) This also worsenes the spam problem; they have a spam filter, but its inner workings are not transparent enough that many innocent emails get caught in them
25. Potential Solutions
26. Use VCs/DIDs, Selective Disclosure What if you can select representations of your identity on each transaction (=each separate email in this context)? minimal/selective disclosure of your identity representation Spam filters will check the legitimacy and trustworthiness of the DID associated with the email
27. What would this enable? • Senders: Less mail caught by spam filters (as long as your email is legitimate) • Also, you don't need to expose your full official identity all the time • Receivers: Less spam, More real mail getting into your inbox • Can coexist with current SMTP protocol/infrastructure (with the right extension)
28. Messaging Layer Security https://messaginglayersecurity.rocks/ IETF Working Group that builds secure group messaging protocol, designed to be interoperable with systems that share this protocol End-to-End encrypted, has encrypted group communication, but still needs an ID on a certain platform
29. DIDComm?
31. Insights from yesterday's sessions
32. JSON Web Messaging (Session by Kyle Den Hartog, 12-E) https://github.com/mattrglobal/jwm Standardized format for secure messaging through extending JOSE family of specifications Designed to be used in combination with other delivery mechanisms such as HTTP(S), MLS, DIDComm, ...
33. Nōtif (from a garden talk with Jim Fenton) https://www.slideshare.net/jim_fenton/notifs-2018 Migrating some use cases, specifically notification to a separate protocol • Opt-in only • Sender is authenticated • Pairwise address (different address for sender-recipient pair)
34. Principles of User Sovereignty / Fundamental Problems of Distributed Systems (Session by Dave Huseby, 9-C, 10-F, 11-I) "When a distributed system fails to address any of the fundamental problems, it opens itself up to corporate capture." Email is a great example of this! Email is designed to be a decentralized system, but opened itself up to corporate centralization from failing to address the fundamental problems.