How can open social protocols fail us?

I put together an analysis comparing several specific failure modes between Farcaster, Bluesky, and Mastodon.

Please let me know if any of the protocol descriptions could be presented more fairly!

How can open social protocols fail us?

I put together an analysis comparing several specific failure modes between Farcaster, Bluesky, and Mastodon.

Please let me know if any of the protocol descriptions could be presented more fairly!

https://shazow.net/posts/open-social-2025/

nice write up

Is the PLC sovereign right now? I thought it was placeholder / controlled by the core team.

A doodler and computerer. I like permissive/permissionless open source, smart contracts, p2p systems, room-scale VR, and NixOS.

shazow.net

Currently: WhatsABI

Working on Farcaster and Warpcast. Longer thoughts at https://dwr.email

Seems much more centralized than Farcaster is today (our smart contracts don't have admin keys that can change the data)

for the example you site, the same is true for PLC. You have to have the keypair that created the DID:plc to update the DID:plc.

why does that matter for the example you sited?

Once you have a DID it's cryptography that defines how valid it is.

Can I inspect the PLC code? Is it possible for a human to change it?  How would I tell if they did? Can they change anything or are they constrained? What happens if the web service is unavailable? How redundant is it? Who pays the bill to keep it running?

A smart contract seems better on all of these dimensions.

I'm eth newbie..just learning
(working on @farcaster)

terminally.online

(this is true of our centralized services like the fname registry too)

Sorta, I tried to avoid talking about "decentralization" and talk about specific failure modes.

The PLC is a signed ledger, sequencing is effectively centralized. Is Snapchain's validator set decided by JSON signatures on Github much better? Maybe, but social coordination is required either way and it wouldn't be a huge lift to improve either one in a black swan event.

Btw the final paragraph is my exercise for you. :)

The output of the directory is "self verifying" state transitions. Think of it like a centralized DA and sequencing. The infra that consumes it checks that signatures match up for the state transitions (eg key A signs over permission to key B). The main failure mode is same as DA/sequencer disappearing, would need social coordination to move to another DA/sequencer.

The output of the directory is "self verifying" state transitions. Think of it like a centralized DA and sequencing. The infra that consumes it checks that signatures match up for the state transitions (eg key A signs over permission to key B). The main failure mode is same as DA/sequencer disappearing, would need social coordination to move to another DA/sequencer.

https://warpcast.com/shazow.eth/0xfeba2af0

PLC server code is here, it's hilariously simple. But you can't verify it's really what they're running.
https://github.com/did-method-plc/did-method-plc

You can also easily run a mirror, but you can't verify that your data is complete. (They propose to fix this with a transparency log like https://sunlight.dev/ )