vrypan |--o--| on Farcaster

vrypan |--o--| pfp

The Farcaster ecosystem largely revolves around two main providers: Merkle Manufactory (MM) and Neynar. Both offer services that span the full stack, from low-level infrastructure (like Snapchain nodes) to end-user clients and miniapp tooling. I consider node improvements, specs, and documentation part of their offering in this context. This creates a resource allocation dilemma. As a startup with limited time and engineers, you naturally focus where ROI is highest. Right now, that seems to be at the top of the stack: ROI = end user growth. But this focus narrows innovation. While low-level infra allows many creative paths, the higher up you go, the fewer options you have. MM and Neynar appear to prioritize miniapps, wallets, and small UX features (like profile banners or max number of cast embeds). This is also reflected in the uneven documentation quality (ex., there is a dedicated website for miniapp development, 99% of the documentation is about using js/ts, the migration from hubble to snapchain is hardly reflected in the docs, and so on). The result: it’s increasingly difficult to build anything truly new. If you’ve tried using the low-level stack, you’ve likely hit countless friction points: a small detail here, a small detail there, a change that has not reached the docs yet, FIPs that would have made your life much easier but are of low priority, design decisions optimized for quick user growth and not DX. I know the answer: demand is king. I generally agree. But, maybe user growth is in places not on the path already laid out. And my concern is that diverging from this path, is getting harder and harder.

9 replies

6 recasts

53 reactions

Dan Romero pfp

What is something you want to build that you can’t today because of documentation?

1 reply

0 recast

2 reactions

Harris pfp

recently I was looking to take inspiration in a lot of the sentiment that has been echoed by some people re: farcaster the protocol has a lot more than warpcast / social media the product, but the more I tried to think of a way to build upon that the more I kept finding things preventing it through these lower level primitives. Experiment: imagine how you might create a new client that treats casts as blog posts with .md or something. I kept ending up with the only solution being to disregard text and use links like you have with lemon3, but then you have timestamps that may or may not be related to publishing, no edits, you can only link 2 things through a cast one of which would probably have to be where the actual content is hosted etc, on top of a lot of wasted data types like mentions and reactions(maybe), anything custom on top needing global protocol discussions and improvements or be rejected from the network.. 🤷‍♂️ maybe I'll play around again at a later date to see what can be done :)

1 reply

0 recast

2 reactions

Tony D’Addeo pfp

you are right that we are focused on the app layer and user growth but otherwise the picture is much more nuanced than the one you paint there has been huge investment in the protocol itself - those “small UX” features were part of an initiative to bring in 1.2m in independent funding for the protocol - snapchain is a huge huge investment and piece of infrastructure to make the protocol work, fine to nitpick about sequencers and docs but without this stride we wouldn't have a scalable decentralized network to nitpick - auth addresses are a huge step to making great UX across multiple clients achievable - weve grown the only decentralized mini app ecosystem of its kind (yes those dreaded manifest files, would’ve been a lot easier to just let people submit that data to our database) which has generated significant outside demand for protocol data and farcaster identities (~100k SIWF sessions per day) - every docs site we’ve touched this year has significantly improved from our baseline not saying there aren’t an innumerable things to improve but I reject the characterization that we've just phoned it on the protocol side of things

1 reply

0 recast

0 reaction

Stephan pfp

Fully echo your sentiment It almost feels they’ve taken an infinite canvas and shrunk it down to a button that opens a webview inside of an app Which is ironic because you absolutely don’t need all the decentralization tech for that

1 reply

0 recast

13 reactions

justin.ahn.eth pfp

pretty fair summary! not an indictment of mm at all as we get it, but a reality check for builders nonetheless

0 reply

0 recast

7 reactions

Colin Charles pfp

You want better low level developer experience. The docs - do you think we could help improve them?

1 reply

0 recast

1 reaction

MetaEnd🎩 pfp

The thing is: the DAU is even in decline. So what are they really doing?

1 reply

0 recast

1 reaction

Trigs pfp

Most things that last start out relatively centralized. We definitely need a second Neynar team that is focused on building out a different direction and draws in a new kind of builder that attracts a whole new userbase. Would be great to see another influence pushing MM to expand their focus or perhaps even start the process of outsourcing more protocol level work. But who's going to take on that responsibility and risk? It's a huge next step. One I hope to see someone take, and soon. But I understand why nobody has yet, because I am not ready to myself even tho I spent 3 years building a decentralized infrastructure project running validators in Cosmos.

0 reply

0 recast

0 reaction

barabazs pfp

>uneven documentation quality This is a major pain point if you want to build something truly novel. Good luck finding it in the unindexed notion pages, casts from the core team etc

0 reply

0 recast

1 reaction

👏 pfp

@vrypan.eth, you just received 1,776 claps from ＠project7 on this content! Want to join the fun? Explore content, swipe right to clap, and earn $HUNT based on your own clapping activity. Your daily clap allowance: 35 👏 Install the Clap Mini App to get +50 free daily allowance 👇

0 reply

0 recast

0 reaction