Evolving communications paradigms and Security

1. Evolving communications paradigms and Security Karen Sollins MIT CSAIL January 23, 2007

2. Overview: pulling on several threads • Evolving communications paradigms • Evolving social model • Evolving security challenge

3. Communications: E2E • Point-to-point • Letters/email • Telephones • TCP connections • Broadcast/multicast • Print media - underneath 1:1 • Radio/TV • IP multicast • From source to destination: some direct, some store-and-forward (e.g. intermediate servers)

4. Client/server: mostly E2E • Remote invocation of specific server • Distribution of “server”: clusters, load balancing, even some P2P systems (collaborating servers) • P2P systems: each element can be both client and server

5. Intermediated communication: losing E2E • Middle boxes • Forwarding (e.g. home for mobiles) • Firewall • Caching • Rendezvous (e.g. for multimedia conferencing) • Beginning to break direct, realtime communication

6. It’s the content • WWW and URLs • Time and space separation • Not a question of when (realtime, etc.) • Not a question of where • Question of what • Identification • Search • Pub/sub • Specification of what something is • Specification of interest or subscription • Current examples: social networking, news subscription services, …

7. Key components • Information • (Set of) Publishers • (Set of) Subscribers • Attributes: how to publish or subscribe • Policies: (publisher, {attributes}) or (subscriber, {attributes}) • Trust model Note: Can be simplified to achieve any of the other models, subsumes them.

8. The evolving social model: Trust and security • Letter-writing: recognize handwriting • Telephone: recognize voice • Email: recognize email address • TCP: recognize IP address • Trust based on • Confidence in unmodified delivery • Confidence in correctness of source

9. And along came…(in the Internet) • Forgeable email addresses • Forgeable IP addresses • The Morris worm • Viruses and other malware • Business opportunities • Enterprise and other organizational controls • ISPs • … Note: not all “bad”, just competing objectives

10. Tussle: competing concerns • Question: why do we care? • Sharing • Cooperation • Exposed contention • Question: can we design for it? • Question: is it monolithic? • Economics • Security • Social status • … • Question: where are the control points? • Regulation • Specification • Design/implementation • Operation

11. Security challenge: Trust model • Not universal: regional, topical,…  context (e.g. Nissembaum, social networks) • Not binary or pairwise: scalable, commutative, …  value-based, community-based • Not immutable  evaluatable, assignable Consider: if assignable must have ability to assign “to something”. Therefore require appropriately defined identities.

12. Advertisement (disclaimer here) The Security and Privacy Working Group: current agenda To explore the nature of identity required in an information-based communications paradigm, as a basis for examining the nature and capabilities required for trust and security