dagelf on Farcaster

Dan Romero pfp

In the next few months, the Farcaster protocol will have the most meaningful increase in DAU decentralization since we started. Base will be adding a Farcaster-powered feed to Coinbase Wallet—an app that has millions of existing users and wallets with balances. They will have a paths for both existing users and new users to onboard to Farcaster, i.e. creating FIDs. That's a meaningful, tangible step toward increasing decentralization on the protocol. So if you're claiming that we're making changes to limit client competition, you are, in fact, factually incorrect (both in what we are working on at the protocol level as well the likely outcome after CB's launch).

26 replies

94 recasts

444 reactions

Dan Romero pfp

Our original thesis was incorrect: a decentralized permissionless protocol would allow other clients to grow users for the protocol. This evolved into product-led protocol development (see blog post). And even that has fallen short on the most important thing developers want: daily active users and user growth. Updated version of this philosophy: 1. Make the underlying data and APIs as open as possible 2. Ensure that accounts / identities are portable and sovereign 3. Build an opinionated, high-quality client that is relentlessly focused on user growth. There will be reasonable criticism early on "how can I trust you won't stop building the protocol", but time and consistent shipping on protocol promises will earn trust (and assuming 1 and 2 actually work). 4. At some sufficient scale—not sooner than 10M+ DAU at the protocol—work on further decentralizing the protocol / hardening it from capture. And because you have a big client, you can do that. https://danromero.org/product-led-protocols.html

14 replies

30 recasts

195 reactions

Jonny Mack pfp

#4 is delusional. at that scale organizational incentives are too powerful to justify working on anything other than maintaining centralized control and capturing value. no high growth enterprise will risk losing the goose that lays golden eggs

1 reply

1 recast

29 reactions

Dan Romero pfp

Varun and I have board voting control. Buck stops with us. You either trust that we are building a protocol or you don't.

5 replies

0 recast

25 reactions

shazow pfp

I trust you to change your mind about #4 before 10M+ DAU :D

2 replies

0 recast

2 reactions

Dan Romero pfp

I'm telling you I won't. I'm intellectually interested in building a successful protocol. The economic argument is a straw-man if I'm telling you it's not going to change anything. Otherwise, you are accusing me of lying. :)

1 reply

0 recast

3 reactions

shazow pfp

I believe your current position, and I respect you changing your mind when you felt the original thesis was disproven. In a couple years from now, what would have to be true to convince you that #4 is the wrong approach and that it's worth decentralizing earlier?

1 reply

0 recast

2 reactions

Dan Romero pfp

If we decentralize earlier, we'll move slower. Moving slower likely hurts user growth. If you don't have 10M+ DAU (with some amount of quality measurement!), then the protocol is unlikely to survive in the long run anyways. Get the protocol to state where it's likely to survive, then decentralize. 8M vs. 10M is roughly the same order of magnitude. 1M is not.

2 replies

0 recast

1 reaction

dagelf pfp

What do you base this on? If you decentralize later and you stuff it up, you will be destroying whatever value you managed to build up, assuming you can get there in the first place. 1M to 10M is not the jump you think it is. The leap happens much much sooner and at smaller numbers.

0 reply

0 recast

1 reaction