It seems like there's an even simpler solution that doesn't require bootstrapping a totally new network.
To me, it seems like Twitter serves two closely related but ultimately distinct core purposes. One is publishing. It basically hosts content on a globally accessible network. Two is identity. A person's Twitter handle is an authoritative record of their identity. A person knows that any tweet on the @wikileaks handle came from someone Wikileaks authorized to use the account (barring some sort of hack).
Twitter's censorship and enforcement ability mostly relies on the identity side. It bans you from continuing to publish under your previously, globally known Twitter handle. Creating a new sock puppet account is trivially easy. Even the best of IP bans is clunky and easy to work around. We all know if Wikileaks wanted to open a new handle, there's really nothing Twitter can do to preemptively stop them. They can only react with a game of whack-a-mole.
The problem is that this poses a coordination problem between publisher and consumer. Wikileaks' readers have to somehow discover and verify Wikileaks' new handle/identity. But think about how this dynamic changes if the identity function is off-loaded from the publishing function.
Imagine an unauthorized client that overlays on top of vanilla Twitter. Instead of subscribing to Twitter handles, you subscribe to cryptographic identities. Participating Twitter publishers cryptographically sign their tweets, proving their identity. The overlay client regularly scans the entire site's feed to discover any new handles using a known signature. If Twitter Inc. bans your handle, just fire-up one of your sock-puppets and all your overlay subscribers seamlessly re-point to the new handle.
The best part is this approach is backwards compatible with vanilla Twitter. You can keep using your pre-existing Twitter handle, and vanilla Twitter subscribers don't see any difference. But if you're afraid of potential censorship, you can encourage readers to gradually adopt the overlay system.
What stops Twitter from installing the client and using it to auto-ban any new account that has the same cryptographic key?
Also, if you have an overlay network to distribute mappings from keys to Twitter handles, why don't you just add the ability to distribute tweets and cut Twitter out of the picture?
You are trying to solve a social and legal problem with technology.
We all know if Wikileaks wanted to open a new handle, there's really nothing Twitter can do to preemptively stop them. They can only react with a game of whack-a-mole.
And that's what they do. Avoiding a ban by creating a new account is already against the terms of service.
Imagine an unauthorized client that overlays on top of vanilla Twitter. Instead of subscribing to Twitter handles, you subscribe to cryptographic identities. Participating Twitter publishers cryptographically sign their tweets, proving their identity. The overlay client regularly scans the entire site's feed to discover any new handles using a known signature. If Twitter Inc. bans your handle, just fire-up one of your sock-puppets and all your overlay subscribers seamlessly re-point to the new handle.
So then Twitter (or a motivated investigator) runs the overlay network as well and gets the new account automatically identified.
To me, it seems like Twitter serves two closely related but ultimately distinct core purposes. One is publishing. It basically hosts content on a globally accessible network. Two is identity. A person's Twitter handle is an authoritative record of their identity. A person knows that any tweet on the @wikileaks handle came from someone Wikileaks authorized to use the account (barring some sort of hack).
Twitter's censorship and enforcement ability mostly relies on the identity side. It bans you from continuing to publish under your previously, globally known Twitter handle. Creating a new sock puppet account is trivially easy. Even the best of IP bans is clunky and easy to work around. We all know if Wikileaks wanted to open a new handle, there's really nothing Twitter can do to preemptively stop them. They can only react with a game of whack-a-mole.
The problem is that this poses a coordination problem between publisher and consumer. Wikileaks' readers have to somehow discover and verify Wikileaks' new handle/identity. But think about how this dynamic changes if the identity function is off-loaded from the publishing function.
Imagine an unauthorized client that overlays on top of vanilla Twitter. Instead of subscribing to Twitter handles, you subscribe to cryptographic identities. Participating Twitter publishers cryptographically sign their tweets, proving their identity. The overlay client regularly scans the entire site's feed to discover any new handles using a known signature. If Twitter Inc. bans your handle, just fire-up one of your sock-puppets and all your overlay subscribers seamlessly re-point to the new handle.
The best part is this approach is backwards compatible with vanilla Twitter. You can keep using your pre-existing Twitter handle, and vanilla Twitter subscribers don't see any difference. But if you're afraid of potential censorship, you can encourage readers to gradually adopt the overlay system.
reply