Can you give me some goals where you'd think this was really successful?
OPTIC is running in-person, undergraduate forecasting competitions in Boston, SF, DC, and London. Think olympiad/hackathon/debate tournament, but for forecasting: students compete for prizes by probabilistically predicting future events.
On competition day, teams of 4-5 forecast on a given set of questions in a timed period through Metaculus. Questions cover diverse topics (politics, economics, technology, pop culture, etc), and span from weeks to months, though we’re experimenting with ways to balance this with more immediate competitive payoff.
Competitions take place in-person at university venues, and include a speaker, meals, merch, etc, alongside the technical forecasting period.
In April 2023, we ran a successful pilot competition in Boston and wrote a detailed postmortem.
Our theory of change is that building the forecasting community through undergraduate students will:
Create a larger, more diverse pool of emerging forecasters and superforecasters, which increases the quality of individual & aggregated forecasts.
Normalize the field of forecasting to the general public, strengthening public support, knowledge, and awareness of the practice of rigorous forecasting
Improve future institutional decision making to reduce the scale and likelihood of future catastrophes.
Our goal is to add an intellectually energetic field of college students to the existing forecasting community through engaging and enjoyable competitions. Competitors will refine their forecasting skills, improving their ability to contribute to the forecasting community and to the future decision-making of key institutions. Currently students say they compete for their school’s debate team or MUN team: we want them to say they compete for their school’s forecasting team.
Success, for us, looks like a turnout of 50-200 (or more) competitors at each region, where most competitors continue to forecast both in their daily lives and at future OPTIC competitions. It also includes the emergence of forecasting clubs at various universities, which OPTIC is separately supporting.
Our actionable items for achieving success will follow these logistics:
Each competition lasts a full day, opening with a short intro and speech by leading individuals in forecasting and adjacent communities. Competitors split into teams of 3-6, and forecast on the first of two rounds of given questions. After a lunch break, where competitors socialize and reflect within and among the teams, they forecast the second round of questions. The competitions close with merch, info about future competitions, sponsorships, and other items of interest during a short concluding speech from the organizers.
Boston Oct 7 DC Oct 14 San Francisco Nov 4 London Nov 18
NOTE: This is meant to give a general impression of what a competition might look like. Each competition has a different precise schedule.
10:00-10:30 Intro & Welcome 10:30-11:30 Speaker 11:30-1:00 Forecasting Period 1 1:00-2:00 Lunch 1:30-3:00 Forecasting Period 2 3:00-4:00 Closing Talks
We’ll be hiring a regional coordinator (contract) in each participating city to geographically extend our organizing capacity. Working closely with the organizers, they’ll be responsible for booking venues & catering, as well as coordinating with local team captains and venues to ensure that outreach and forecasting club creations go smoothly.
During the event, teams make probabilistic predictions on topics ranging from Russia/Ukraine and China/US, to AI development, to global economics and indices, to political elections, as well as topics like sports, games, art, popular news and culture, public-figure tweets, etc. The timeframe of these questions is on the scale of weeks to months, all resolving within a few months. We're currently hosting our questions on private Metaculus tournaments.
To see our past questions, see here.
For examples of other potential questions, see Metaculus.
When all questions have been resolved, the main body of prize money will be split between the three teams that have scored most accurately using log scoring. The remaining prize money will be awarded after an optional, post-tournament follow-up in which competitors are encouraged to individually update on their predictions in the following weeks.
We’re also experimenting with non-monetary prizes for the coming semester.
OPTIC competitions host speakers from leading individuals in forecasting and adjacent spaces. The spring competition’s speaker was Seth Blumberg, a behavioral economist at Google who runs their internal prediction market. He spoke in an individual capacity on his opinions regarding the future of forecasting — you can watch his speech here.
We’ll distribute OPTIC and sponsored merch at the end of the competitions. In the spring, we handed out five versions of OPTIC stickers — see here for some pictures!
See our full, official budget here.
Our budget is split between logistics (venue, prize money, food, merchandise, transportation, etc), and wages (core organizers and regional organizers). Many of these costs begin at 0, as (a) an MVP competition could technically run without them, and (b) we’ve already received some funding from Metaculus and SIG through partnerships.
Our mainline request, at ~$39k, runs 3 competitions: Boston, San Francisco, and one of DC or the UK. It includes lunch, a significant prize pool (about $3k), organizer travel costs, and salaries for core and regional organizers. The breakdown for each competition is: prize money (about 30%), regional coordinator wages (about 25%), organizer travel (about 15%), and other, smaller expenses, like food and merch, as well as a buffer of 20%.
Our minimum request, which includes funding we’ve already received, is $0. That includes 2 small competitions in Boston and San Francisco, with a small prize pool (~$1k), and no regional organizers, catered food, or merch.
Our ideal request, at ~$114k, includes an additional competition (the remaining of DC or UK, or potentially also NYC) and higher quality logistics at every level. For example, the prize pool is bigger, there's more money dedicated to speaker honorariums, the regional organizer salaries are higher, etc.
The core teams consists of Tom Shlomi, Saul Munn, and Jingyi Wang. We ran a successful pilot competition in April, 2023. For the comprehensive postmortem of the pilot, see here; for a quick summary, see below.
114 people from 5 countries and 13 US states originally registered interest, though many of them couldn’t attend due to logistical issues (we encouraged people to register interest even if they knew they couldn’t come, so we could have their contact for future competitions). At the pilot, 31 competitors attended, representing 8 universities from Boston and nearby. In teams of 3-5, students forecasted on 30 future events for a total prize pool of $3000 USD split between 1st place ($1500), 2nd place ($800), and 3rd place ($400), with an additional prize ($300) devoted to post-competition forecasting. In August, we’ll resolve all the questions and award our winners! Based on this existing network and conversations we’ve had with people who registered interest but couldn’t attend, especially those outside of the Boston area, we’re confident that future OPTIC competitions will also be well-attended and impactful.
The pilot had a total of $6000 in funding, split between prize money, travel funding, food, merch, and misc:
$4000 from the ACX Forecasting Mini-Grants on Manifund
$2000 from the Long Term Future Fund
Tom organizes for Harvard EA, and ran a workshop series at Harvard focused on teaching practical forecasting skills. He is also a former summer research fellow at the Center on Long-term Risk and long-time forecaster on Metaculus and Manifold.
He organizes for Brandeis EA (BEA): he coordinates the reading group and helps facilitate the introductory fellowship. He’s taking this semester off school to focus on OPTIC and Manifest, returning to BEA in the spring.
In the past (at his synagogue and as co-president for his high school Jewish Student Union), he organized several 50-200+ person events, and many 15-50+ person events, as well as debate tournaments with similar structure to OPTIC.
Jingyi is the Brandeis EA treasurer. Aside from treasury, next semester she'll lead the reading group, co-facilitate the intro fellowship, and co-organize the intro event. In the past, she was a special projects assistant for the Providence General Foundation, helping coordinate funds for an end-of-year auction.
Our premortem for Fall 2023 is split into systemic and local failures, where a systemic failure affects OPTIC as a whole (eg, organization no longer exists), and a local failure affects a singular part of OPTIC (eg, only the DC competition fails).
We don't receive sufficient funding to run an MVP this semester.
We're likely going to receive some minimum sponsorship money, so this probably won't happen.
Organizers are too busy, burnout, quit, etc.
Because the organizers are doing OPTIC in addition to other commitments (Saul is organizing Manifest through September, Tom is working at a start-up, Jingyi is at school), this is possible, though not likely.
Saul is taking the semester off to organize for OPTIC; for him, this failure mode could look like a different project taking up a lot of his time, or some other important committment.
One competition goes so badly that this affects the whole of OPTIC
This would have to be really bad, either something behaviorally abhorrent (eg, assault), or the total failure of a local competition (eg, cancelled last minute).
We don't receive sufficient funding to ensure the event is engaging, fun, and organized (eg, small prize fund, speaker requests an honorarium but we can't pay it, can't afford food)
Participation is extremely low
This would be caused by poor pubbing/outreach, or simply the lack of interest in a particular area. This is highly unlikely for the Boston competition, given that we saw sufficient interest in the pilot, and have that existing network of competitors to work off of. It's comparatively more likely for the other regions.
Or, local organizing is weak: one important path to recruiting is through local groups (eg, university EA groups).
Logistically, the competition is very messy (a venue cancels, food is promised but not delivered, etc)
We fail to hire a good regional organizer. This is less likely to result in failure in Boston and SF because at least one core organizer will be in those cities to pick up slack, and is a bigger problem in DC and London.
Poor communication between core organizers and regional organizers
Though less relevant to the specifics of Fall 2023, there are also a few ways OPTIC could fail in the long term.
It's possible that OPTIC is successful on the surface (consistent/sufficient funding, high participation, good reputation), but actually fails to create positive impact — or creates negative impact.
This would be especially bad, since we’d likely eat up even more funding/resources without being impactful
OPTIC ends up not producing impact. This could be at any point of our theory of impact:
We could become a funnel toward something unimpactful
We could become a funnel to forecasting in an unimpactful way
E.g. they only forecast on their personal lives
Forecasting as a whole could be unimpactful
Not enough longterm organizers
The primary constraint on OPTIC is regional and core organizing capacity, and we expect that to remain true.
OPTIC has received funding from Manifund (through the ACX Forecasting Mini-Grants round) and the Long-Term Future Fund for the pilot competition. For the coming semester, our applications to Open Philanthropy and the Long-Term Future Fund applications are pending. Our application to Lightspeed was recently rejected. We will likely receive some funding (5-15k total) from sponsorships.