DrewF

I haven't dug into it deeply, but I think every PDS opens a websocket to relays that they want to send updates to, and push them into the relay (mainly the bsky one).

If a relay sees msgs that point at a PDS it doesn't have, it can call out to that PDS to directly fetch and/or open a socket.

04.11.2024 15:21 — 👍 2    🔁 0    💬 1    📌 0

Actually, blockchains are good.

Even worse, cryptocurrency is good.

Sorry for the inconvenience.

02.11.2024 12:55 — 👍 4    🔁 0    💬 0    📌 0

Fore either, it's because it enables you to sell services that less sophisticated actors can't.

For Ethereum, there's a single commitment (the current block) that enables anyone to verify accuracy from any archive node.

For AT, there's one commitment per PDS, so verifying archives is harder.

02.11.2024 11:31 — 👍 1    🔁 0    💬 0    📌 0

That's a full node, not an archive node in ethereum terminology. Even so, the slashing example isn't useful to the role of the really in AT as I understand it, the relay doesn't require consensus, it can change its mind about PDS history, right?

02.11.2024 02:24 — 👍 0    🔁 0    💬 1    📌 0

FYI, slashing is very narrow, and only applies to voting for 2 conflicting views of the chain tip. There is no financial incentive to store past state (an archive node) or even txn history (a full node), though not serving history may cause you to lose peers.

02.11.2024 02:13 — 👍 0    🔁 0    💬 1    📌 0

Do you know if the upcoming content-addressed mempool for celestia is ipfs/ipld? Would be super cool if it was just pluggable, and you could anchor timestamped PDS with blobstream.

02.11.2024 02:01 — 👍 0    🔁 0    💬 0    📌 0

Video player with loading spinner and large negative timestamps, not playing.

Maybe it does?

01.11.2024 18:46 — 👍 0    🔁 0    💬 0    📌 0