boscolo.eth on Warpcast

boscolo.eth pfp

Bluesky has more users than Farcaster and is also sufficiently decentralized. What is the best argument for why devs should build on Farcaster instead of Bluesky?

6 replies

3 recasts

25 reactions

Dan Romero pfp

What % of users use the primary server?

5 replies

1 recast

18 reactions

sean 👤 pfp

now that threads has joined the fediverse, do you think that in 5-10 years the % of fediverse users using bluesky's primary server (assuming AT protocol fully bridged to fediverse) will be greater than the % of farcaster users using warpcast as their primary client? this is a serious question.

1 reply

0 recast

6 reactions

Dan Romero pfp

Threads does not fully interoperate with the fediverse, i.e. the graph is not available. Bluesky is not in the fediverse? Any bridging they do could happen for Farcaster? I'm generally bearish on a single decentralized protocol as winner take all. Think there will be a plurality.

1 reply

0 recast

9 reactions

sean 👤 pfp

i believe the plan is to bridge all of them. if the long-term goal is to bridge hubs with fediverse-style servers, why go through the hassle of a distributed network that isn't turing complete if it has a very dominant client anyway? https://techcrunch.com/2024/06/05/bluesky-and-mastodon-users-can-now-talk-to-each-other-with-bridgy-fed/

1 reply

0 recast

5 reactions

Dan Romero pfp

I guess worth stepping back. Our bet is user-controlled identity will live on public blockchains, not consortium / non-profit servers. Our bet is also a tightly integrated identity to cryptocurrency will offer more interesting apps and services over time. If that doesn't happen, I expect a federated protocol to be the big one. I'm skeptical (based on the prevailing political views) that current federated systems will be open to cryptocurrency.

2 replies

0 recast

20 reactions

sean 👤 pfp

but because farcaster hubs aren't a turing complete consensus network (smart contract blockchain), it sounds like you're saying both identity and crypto will be on other networks anyway, so i'm not understanding how hubs is superior on these points. why not just build a federated protocol that is crypto-friendly? you can still store and share signed messages for activity verification

1 reply

0 recast

7 reactions

Dan Romero pfp

Federated is worse developer experience and will become oligopoly over time (see email)

1 reply

0 recast

6 reactions

sean 👤 pfp

sure, and i'm asking how farcaster hubs avoid that since all signs point to it being the same

1 reply

0 recast

3 reactions

Dan Romero pfp

There are 1000+ hubs that all have a full copy of global state? Quite different. Federated systems require "peering agreements" at scale to get the full state of the network.

1 reply

0 recast

4 reactions

sean 👤 pfp

as you are constantly reminding people, hubs don't get rewards, but i think many current hubs are being hosted with that inaccurate goal. what clients or services other than neynar are running their own hubs at any significant scale? in terms of peering agreements, i believe sharing limitations on historical network state is due to efficiency, and the entire history could be retrieved from those who store it when necessary

1 reply

0 recast

3 reactions

Dan Romero pfp

1. The Hubs do exist and work with full copies, so the reason doesn't matter? In the long run, if the network grows, more companies will run Hubs. 2. Peering agreements are about deciding who gets access to your data. It's by definition permissioned / not permissionless. Farcaster design is closer to Ethereum in that regard.

1 reply

0 recast

6 reactions

sean 👤 pfp

1. you mentioned federated becoming an oligopoly long-term in all likelihood, which is an argument based on the likely outcome of the network despite its abilities. i am merely making the same prediction about this network, despite its abilities 2. agreed on the superior openness of farcaster. however, since federated protocol aspires to be open as well, it sounds like you are arguing from an ideological standpoint that federated hosts should not be trusted. is that ideological standpoint worth the extra hassle and cost of a distributed network when it doesn't have consensus anyway (and therefore doesn't handle transactions or other time-sensitive data)

2 replies

0 recast

3 reactions

Dan Romero pfp

What at scale federated system has not ended up in an oligopoly that makes it hard for small developers to break through?

2 replies

0 recast

2 reactions

sean 👤 pfp

i agree with you on the oligopolistic endpoint likelihood. all i am saying is that at this point with warpcast, oligopoly already appears to be the outcome of farcaster as well

1 reply

0 recast

3 reactions

Dan Romero pfp

you can simultaneously use other clients, just like ethereum not possible in federated system (your client is just a front end on a single backend)

1 reply

0 recast

2 reactions

sean 👤 pfp

i'm not following how federated can't use multiple backends. a frontend could use multiple servers for data anyway. in terms of consensus, because transaction strictness is not needed here, you could simply use signer nonce + an external time-based third party (like blockchain block numbers) to verify timestamp correctness and have pretty accurate network consensus for message order

1 reply

0 recast

1 reaction

Dan Romero pfp

All of your data is stored on a single provider -- whether email or federated social.

3 replies

0 recast

0 reaction

boscolo.eth pfp

This is sort of true, but doesn't capture some important nuance. With both Bluesky and email, I can exit and find a new provider without disrupting the flow of communications.

0 reply

0 recast

1 reaction