One of my favorite places to hang out in Warpcast is the chess channel. There are a lot of people talking about current events in chess and posting tactics for others to solve. People will respond to the tactics by writing out their answers move by move in a new cast. Usually, the first person who responds correctly will be awarded $degen for their efforts. It's a lot of fun.
I thought it would be even more fun if you could interact with tactics in a Farcaster feed. Frames are the perfect way to experiment with making this happen. I think of Frames as interactive applications that can be embedded into a Farcaster feed and usable in any FC client. The first Frames have been relatively simple, but it doesn't require a stretch of the imagination to see how these can evolve into full-blown games and social interactive experiences over time. They are very reminiscent of Zynga's Farmville infiltrating Facebook years ago.
I wanted to create a Frame where you could solve tactics within it instead of having to post a picture in a cast, reply with answers, and wait to hear if you were correct. These types of web applications exist (chess.com has a great tactic application), but why not make it an experience inside the chess channel feed? I created a bounty on bountycaster and a caster responded to it with a quick and dirty version of a Frame. We worked together over a couple of days to refine the experience, and ultimately got to a good place.
While it's not perfect, my goal was to make something that was fun for people who like chess on FC to use, and to hopefully inspire people to push the envelope of what Frames can do. Can we build real interactive games inside them? Can they be social or multi-player experiences? When can I play chess against another FC user in a separate client but an audience could follow along in real time in a Frame? Are these ideas even possible?
This was a fun experience for me because it helped push my thinking around two areas. The first is the idea of Frames as interactive applications, or "programmable anythings" as I call them in my head. We are really at the very beginning of what is possible here. I haven't seen many instantiations of Frames that have actual inputs other than pressing a button. The Frame we created through this bounty has a text prompt, but there will inevitably be other modes of interaction within them over time. The skeuomorphic expression of this is Farmville in the FC feed, but something weird and native to the onchain experience will emerge and be very special, and it will happen after a lot more collective experimentation.
The second area is around the future of headless applications. When I created a bounty on bountycaster, I did not use a bountycaster client. Everything was done entirely within the Warpcast feed. I simply crafted an initial post, someone responded, and a job spec was fulfilled.
This experience reminded me of a post I wrote about agent native applications. The premise is that in the near future, agents will act on our behalf and existing marketplaces or services that require your eyeballs to visit their website or specific client will not be compatible with them for a whole host of reasons, but headless ones will and they'll emerge to fill this need. These headless applications are already emerging onchain. Sometimes they are referred to as headless marketplaces. Using them is pretty magical and the implications are profound when we think about the future architecture of internet services and marketplaces.
Bountycaster is an example of a headless application that has emerged in the FC ecosystem. You can think of it like Fiverr or Upwork, but instead of having to do everything through a specific website, I can initiate or fulfill a job request from anywhere within Farcaster. This means that as a job-doer my reputation is not tied to Fiverr or Upwork. It's portable across any onchain application and I can bring it with me. And when I want some work to be done, in this example the creation of my chess tactics Frame, I can post it anywhere onchain and it can be distributed and responded-to through any client built on FC. Being headless enables a job post to proliferate and be accessed everywhere onchain, and any job-doer now has a portable and self-owned profile, work history, and reputation. This is an architecture that is also compatible with an AI agent-centric future. I invested in an iteration of this Bounty concept years ago with a project called Ahoy, but bountycaster pushes this several steps further into the future.
There are some obvious holes I encountered in the bountycaster experience, but those will all be patched (eg escrowing payment at the outset of a job and having a network of arbiters that can subjectively determine when a bounty has been fulfilled to spec). The whole process felt familiar but entirely novel. It was a taste of things to come in the future and yet another reason to be excited about the pace of rapid experimentation onchain that's creating fun and useful consumer experiences in entirely new ways.