The recent news that Meta’s Threads platform will eventually support/integrate the Fediverse has prompted the crypto economics side of my brain to do some soul searching.
Since the advent of Bitcoin, crypto economics has proven effective in incentivizing decentralized infrastructure providers to permissionlessly show up and work, and ensuring these providers find it in their best interest to follow protocol. The fediverse, however, is a permissionless network that doesn’t consider crypto economics.
Open, permissionless networks are made up of decentralized operators. These operators do nothing more than maintain software on connected servers across the globe. Anyone can run a server. Closed, permissioned networks operate similarly, in that there is a global network of connected servers running the same software with one slight difference, which is that all network operators are permissioned by a company or allowlist group, such as Meta’s Instagram or Threads.
Open, permissionless networks have two problems: (1) they need operators to show up and work (i.e., run the server, follow protocol, and communicate with other servers), and (2) they are vulnerable to spam, and may be assumed to be under constant threat of attack by adversaries (including those running the servers). These networks need money to pay workers and have controls in place that provide some degree of assurance that spam or attacks will not bring down the system.
Enter the token incentive. Ether, the network token of Ethereum, exists to properly coordinate network actors, which just means incentivizing them with rewards when they follow the protocol, and taking away their Ether when they break protocol. Ether also exists to facilitate an auction for network resources, which makes it difficult for some random entity (malicious or not) to overload the network with messages (i.e., spam).
Tokens, in short, exist to properly coordinate network actors.
The term federated network is used when describing networks of servers that are free to do what they want, and may communicate with other servers as they please. My early understanding of federated networks, like ActivityPub or the Matrix Specification, is that they provide a communication protocol between clients-servers and servers-servers, actors do not need permission to communicate in the network, and no token exists to coordinate actors.
Open, permissionless networks operating without a token are not my strength (I’m a protocol economist).
The key difference between an open, permissionless network with a token and one without seems to be around expectations for actors to show up and work, and guarantees for protecting against network attacks and spam. In federated networks, actors are assumed to show up and work for some implicit reward—either goodwill or some indirect revenue source—and implementing spam control seems to be on the network actors.
I have many open questions at this point.
Can guarantees for network availability, censorship resistance, and security be made in permissionless networks sans token? These guarantees are commonly available in crypto economically secured permissionless networks. Is a reputation mechanism, without a token, powerful enough to provide some of these guarantees?
How are network attacks mitigated?
How can we be sure professional grade infrastructure will join the network? Will it be sufficiently decentralized?
Maybe the guarantees provided by these networks are only meant for transacting low value information, such as social media (as opposed to money services, such as payments)?