RetroPGF 3 Retro


2023 marked the conclusion of RetroPGF Round 3, one of the boldest experiments in governance. It was incentive design, coordination, and public goods at truly a unprecedented scale. At Agora, we're incredibly fortunate to have built one of the software platforms that supported the latest iteration of this program. As it ended, we did a retro, and wanted to write out a few of our takeaways:

  • Unprecedented Scale: RPGF 3 generated unprecedented volumes of traffic, peaking at 13 million requests in its final week, underscoring the immense engagement it garnered.
  • The emergence of community tools: many developers built popular micro apps on top of our API to help projects and voters engage with RPGF3 in ways we didn't expect.
  • Changing the funding game: with its size and consistency, RPGF is slowly changing how projects think about funding their work. IMO, it could pave the path for a new kind of business model.

RetroPGF3 by the numbers

Traffic and uptime


Agora sees on average ~100k monthly users, so we're no stranger to traffic. However, we did not anticipate the level of interest the community showed in RPGF3. Over the course of voting period, Agora processed over 13M requests to and our API, averaging between 300-500K requests per day. In the final week leading up to the vote, our servers handled over 5M requests. During peak times we were serving nearly 30 requests per second. The most surprising thing was that we had traffic coming in from multiple clients beyond West and Agora – ones that we were not expecting.

uptime While Agora was up 99.65% of the time, we had severely degraded performance for 5-6 hours on our API during the last day of the voting period. This combined, with the CloudFlare outage on November 6th, means that our our true non-degraded uptime was likely under 99% at times. Luckily, our downtime was observed only by a small portion of citizens who voted on the last day. Regardless, we take our uptime very seriously and have major improvements on our roadmap to make sure this doesn't happen again.

Ballot Behavior

Another big surprise for us were the ballots, we expected Citizens to put 30-50 projects into their ballot as we assumed (wrongly) that doubling the amount of OP would only double the project and ballot size. It quickly became clear that we were quite off in this assumption. In reality, over 70 lists were created, and some of these lists contained over 100 projects. We suspect that when Citizens saw the number of projects, they decided to be more generous with their distributions:

  • Over 30 badgeholders had more than 200 projects in their ballot
  • One badgeholder had over 500 ‼️

For the next round, we're going to do a lot more scenario modeling to stress out different voting behaviors and situations. From a feature perspective, we'll also be thinking through support for:

  • Offline building of ballots
  • Power user features like ability to bulk-add projects (perhaps via CSV import)
  • Improved admin dashboard / support features (maybe even help better monitor conflicts of interest or whitelists, etc.)

Permissionless RetroPGF Clients

Opening up the API for all

We had originally scoped to build the RetroPGF to serve two clients: Agora and West. By the end of RPGF3, we counted a total of 8 unique clients that were using our API, not including the two official clients, with being the biggest. Unfortunately, we were not expecting this kind of behavior and didn’t instrument our API with the best metrics for proper accounting, but we will be doing that in the future. We love this energy by the Optimism ecosystem and fully intend to provide better support for builders by building a robust and accessible API for the whole community. Clearly there is an appetite for it and we can use it to help expand the ecosystem and encourage more developers to jump in building tools for the Optimism Collective.

Request for Feedback

The above is just to name a few changes we are publicaly committing to improving. However, we’re already seeing a whole spectrum of feedbacks just like the last round.

If you’re a builder in the ecosystem or a badgeholder, we'd love to hear your feedback or things you want to see improved for next round, please don’t hesitate to reach out and ping us on Discord anytime! We’re excited to hear from you!

Why we’re Optimistic on RetroPGF

All in all, I love that Optimism is in the arena, experimenting and constantly building. As fellow builders, we will always be supporting those who are trying to build things better – and RetroPGF is certainly one of the boldest initiatives in the area of funding public goods. If it's successful, there will be impactful 2nd order effects that will be incredibly exciting:

  • Public goods will for the first time have a potentially reliable and dependable source of revenue. One that’s potentially bankable.
  • I don’t say the word “bankable” as a meme. I built a $2B financing business in my prior life, and the key to any funding is in the ability to underwrite a future potential. RPGF makes that future underwriteable.
  • This means that VCs or even banks could potentially fund public goods and businesses who simply points at RPGF as their business model. Similarly, you can start to spend and hire with RetroPGF as a collateral of future earnings.

Ultimately, this is not exciting because of VCs or more coins... but because liquidity is the lifeblood of a strong market, and a healthy market is what will 10x an ecosystem to have more builders with more ambitious projects in the future. This is the type of virtuous cycles that makes me want to wake up in the morning. Here's to more public goods in 2024!