Future-proofing talk -- gathering feedback.

[jbenet]

Talk materials:

• video: https://www.youtube.com/watch?v=soUG72j7kB0
• slides: https://cyber.stanford.edu/sites/default/files/juanbenet.pdf

I recently gave a talk about Future-Proofing at BPASE (Blockchain Protocol Analysis & Security Engineering -- whole proceedings of the conference here: https://cyber.stanford.edu/blockchainconf ). The video was released yesterday. This is the first stab I give to this sort of talk, so it's a super raw draft. I'd like to develop this into a good talk about Future-proofing, including the learnings we've acquired, and in particular how multiformats, libp2p, ipld, and other parts of the IPFS ecosystem can help. Please give me any feedback on content here. I may not have the bandwidth to respond, but I will read all these before the next iteration i give of this talk. And, i'm not looking for feedback on delivery-- that's a whole separate thing that will get better as i give this talk a few times and re-structure it to have the right flow.

Some example notes already:

• "never going to change" considered harmful needs an image. and a cleaner name. (the "future bites" principle?)
• explaining ipld and libp2p was shoddy (need to compress it and make it succinct)
• @lgierth it would be good to write a proper paper in the coming months about multiformats and the "Never going to change" considered harmful stuff.

[ianopolous]

This might be too detailed, but length prefixing is pretty prevalent all through this stuff, from multihash, to cbor, and, crucially, using varints.

[taoeffect]

This is an excellent subject. The feedback I would give is for the Open Problems & Challenges and Future Proofing Blockchains sections.

The truth is that we have yet to figure out how to future-proof them, and this should be explicitly stated in the slides somewhere. Concrete examples of why that is true should be given.

Let's take Bitcoin as an example. Here are some current, real problems with future-proofing Bitcoin:

• Bitcoin is constantly under attack by large and very powerful interests. It's possible that what we know as "Bitcoin" today will no longer be Bitcoin in 5 years. If that were to happen, we would no longer be able to assume any of these things:
  • That we have access to the blockchain history
  • That the system is censorship-resistant (i.e. it could become centralized)

In practice that means that timestamp proofs could disappear, and we would be back to trusting untrustworthy third-parties, which effectively destroys the utility of Bitcoin as a timestamping service.

This is not a solved problem, but simultaneously it would be incorrect to call it a totally unsolved problem. Blockchains are in fact—by far—the best solution we have to this today.

[taoeffect]

In practice that means that timestamp proofs could disappear, and we would be back to trusting untrustworthy third-parties, which effectively destroys the utility of Bitcoin as a timestamping service.

Let me expand upon that a bit. How could timestamp proofs disappear?

To understand, you have to look really far out (decades or more maybe), to where Bitcoin has been redefined to a different blockchain.

The original blockchain (the "true" blockchain), is now viewed as a forgery by most people. Even if someone demonstrates a timestamp that is validated by it, it would by denounced as a forgery, or simply news of it would be censored, just as 9/11 Truth is denounced and censored today. It doesn't matter if it's true, if most people trust untrustworthy (and wrong) third parties. Ultimately social consensus is what rules.