The Web is broken. HTTP is inefficient and expensive, especially for large files. Webpages are being deleted constantly, with the average lifespan of a web being 100 days. The Web's centralization limits opportunity and innovation. And it causes problems in the developing world, with natural disasters or faulty connections. We can do better. In this talk, I'll explain IPFS, a project intended to replace HTTP and build a better web. IPFS is a peer-to-peer hypermedia protocol to make the web faster, safer, and more open. In addition, IPFS will use Filecoin as a reward mechanism. Filecoin aims to provide a decentralized network for digital storage through which users can effectively rent out their spare capacity, receiving filecoins as payment. Filecoin raised 200M$ last month, breaking all records in blockchain ICOs to date.
14. Server <> ClientWeb
Server provides the right service
Server uses data responsibly
Server is secure
Server is always online
Server is single point of reference
15. Server <> ClientWeb
Server provides the right service
Server uses data responsibly
Server is secure
Server is always online
Server is single point of reference
Server <> ServerP2P
Altruistic Network
Peers are equal
16. Server <> ClientWeb
Server provides the right service
Server uses data responsibly
Server is secure
Server is always online
Server is single point of reference
Server <> ServerP2P
Altruistic Network
Peers are equal
Network <> ClientIncentivized
Protocols
Clients delegate their service to the network
The network is “paid” to do so
17. 5,000,000,000+ files
• Video distribution & streaming
• Legal documents
• 3D Models (they’re big!)
• Games
• Scientific data & papers
• Blogs & websites
• Within blockchains
• Totally distributed web apps
ex. forums, chat, messaging,
cms, blogs, github, ...
IPFS
20. Content Addressing
Where is the data?
Who has the data?
?
http://nicolagreco.com/slides.pdf
QmeomffUNfmQy76CQGy9N…
[RFC6920]
[BITTORRENT]
[IPFS]
[BITCOIN]
21. Content-addressable Web where:
- data links work across application
- links are cryptographic hashes
- anyone can distribute data
28. SLIDE: SIZE OF THE
COMMUNITY
Large Open Source Project
Over 2000+ Contributors
Over 150+ contribute Weekly
29.
30. Server <> ClientWeb
Server provides the right service
Server uses data responsibly
Server is secure
Server is always online
Server is single point of reference
Server <> ServerP2P
Altruistic Network
Peers are equal
Network <> ClientIncentivized
Protocols
Clients delegate their service to the network
The network is “paid” to do so
65. Naming
Local Naming
./nicola.jpg
Global Naming
Trusted Party
[RFC1035]
Blockchain
[ENS]
[BLOCKSTACK]
[NAMECOIN]
Untrusted Party
[RFC6962]
[CW2009]
[CONIKS]
hash( )
Qas13jdsjw
Cryptographic
Hash
[RFC6920]
[BITTORRENT]
[IPFS]
66. Naming
Local Naming
./nicola.jpg
Global Naming
Trusted Party Untrusted Party
[RFC1035] [RFC6962]
[CW2009]
Blockchain
[ENS]
[BLOCKSTACK]
[NAMECOIN][CONIKS]
hash( )
Qas13jdsjw
Cryptographic
Hash
[RFC6920]
[BITTORRENT]
[IPFS]
Trusted
Majority
Trusted
Party
Cryptography
67. Fair Exchange of
Services
Nicola Samer
Fairness
Timeliness
Either both parties receive their inputs or none
Exchange either happens or does terminate
Completeness If seller is honest, both parties receive their inputs
Soundness
90. Proofs of Storage
Complete
Sound
Public Verifiable
Transparent
Useful
everyone with storage will generate valid proofs
no adversary can generate fake proofs
everyone can verify proofs
there is no secret information that can generate proofs info
proofs are about useful storage
117. Open Questions
Are there ways to do efficient proofs of replication?
Are there ways to overcome front-running attacks?
Are there ways to avoid miners posting proofs on chain?
Can build a storage-based consensus?
What are other incentives beyond economic reward?
153. Question?
What are other trust models? What is any-trust?
What are other types of Verifiable Markets?
Would this be enough for decentralizing the web?
Can DDoS attacks be possible on Filecoin?
How do you know Samer?