As a conference focused on prediction markets, great start seeing @robinhanson get to the messy details of futarchy in and putting prices against decision One issue was CEOs not wanting to give up decision making power to markets, so they GTM with multi stakeholder decisions

The best person to follow and ask questions to, IMHO, is @aquietinvestor, who also happens to be my boss at @ShieldedLabs.

in my final post on the iceland residency, i'm gonna go out on a limb and list my 5 favorite projects to emerge from it research is a tricky one to evaluate since ideas are so experimental that judges can be completely wrong in any case, here's my top 5; 1. The most high potential project in my view is @clesaege doc on redesigning blockchains so that block reward funding can be allocated to support the ecosystem instead of just burned or given as dividends, as is currently the case - the most important idea here is NOT enshrining any particular actor or mechanism as fund recipient (for eg in arbitrum its tokenvoting dao is recipient of sequencer & mev fees). instead, we have a meta mechanism that lets block producers or validators vote on amount to be allocated between different recipient contracts such as guild mechanisms, QF, deep funding, tokenvoting etc The "mandatory indirection" aspect of block proposers choosing mechanisms rather than recipients is important for keeping the mechanism not easily captured - the second most important idea is adaptability, where block producers can increase or decrease funding to the ecosystem this adds accountability, where if there aren't any good mechanism they set the amount to 0 which is basically the status quo - the final piece of the puzzle is having a condorcet style voting system instead of proportional so that its not gameable in the way of producers simply allocating the share to themselves the full proposal with his simulations is worth checking out in full! 2. in second place is a reimagination of @deep_funding to become more market based instead of one-off competitions (i worked on this piece so CoI) the core idea is a market for each edge in a dependency graph that is traded based on its value if it were evaluated by a jury while only some edges actually get spot checked (based on which traders win or lose money), we can still use the weights given by the market for all edges to allocate funding apart from mathematizing the new formulation of deep funding, i like how it solved the sybil issue (where builders submit multiple models in hope one is the winner), let participants specialize on only a few edges in the graph instead of scores for all and does away with manual intervention by making it recurring, which also keeps the weights fresh and relevant 3. in 3rd we have @dwddao work on simocracy, which draws upon the idea of a "digital twin" they created an LLM for 5 individuals in the retreat that have served on grant committees they then ask the judges "digital twin" to deliberate between themselves on funding to projects while RPGF rounds have a low cadence due to the high overhead on executive time, simocracy can allow for some portion to be given every month by the badgeholders digital twin & another kept aside for the regular rounds in my opinion, the 3 things to do for this idea to prove it can work are; - Create a reinforcement learning loop (either geminis high context window or vector embedding) so that scores from jury rounds are fed to the judges simulation, increasing accuracy over time - Compare the exact scores a judge has assigned to projects in each category with the ones its twin generates. Measure the variance as a target to reduce over time - Even if the variance of scores might be high, its more important to compare the overall rankings for projects. so if the judge and its twin both have ~same ordering of projects, that still makes it worthwhile 4. I loved the simplicity of @CS_Synthesist presentation: a comparison of how results differed from actual RPGF rounds vs what they would have been if we had applied condorcet voting to the picks by badgeholders The best part is we dont require any change for the judges; we can literally take their existing scores, use condorcet instead of whatever algo was used to get results, and compare the 2 to see which seems better condorcet is one of those rare ideas that all academics agree is better than most voting systems but is still rarely implemented in the wild, im hoping to see more use here 5. finally we have @iammadab work on lean, a programming language allowing for formal verification during the writing of software instead of a step after it lets us evaluate contributions much easier since it enables mathematical verification he used this to create dependency graphs, so i can see deep funding type mechanisms that scale human evaluation more easily applied in the lean environment please don't shoot me, other residents, this is a personal list based on my prior knowledge in general, these types of retreats are expected to follow a power law for impact, where the highest impact idea is orders of magnitude more than the next one however we can't easily identify which ideas are likely to be power law winners, overall i'm really happy with how the retreat went with every participant submitting some concrete output that can be taken forward by the community link to the proceedings page in the next post

Honestly the end of the residency was a big relief, it took as much effort to organize as an Indian wedding 🥵 I'm very proud of our end outputs: 22 submissions from participants! I think we've shown a new way to do these sorts of gatherings; 1. Book a coworking 2. Book airbnbs around coworking 3. Put people into airbnbs with room mates depending on who you think they would vibe with 4. Have cool outings on the weekend 5. during the week people work with clear expectations on what they should be producing as an output for the end Overall we can put these residencies or popups into 2 categories. The first involves  people paying (large) amounts to attend. They arrive with a more consumer mindset so it's essential to have good activities and vibes that give them a bang for the buck. @joinedgecity comes to mind as the best in this category . I've seen many relationships take place during their popups 🫥 The second are builder residencies that bring people together for making something l in a thematic area. Since these are (usually) paid for , the vibe is  more of a producer mindset where participants want to show they were worth the investment to bring and feed them. It's important to create feelings of specialness in such gatherings. Our iceland retreat & @FundingCommons stand as examples in this category Gonna take a break now and write up some of my favorite outputs in the residency