Above the Fold: Invasion Edition
News from Manifold Markets: Russia/Ukraine, EA grant, X-Risk, Shoutouts
Russian Invasion
The invasion of Ukraine took many by surprise this week—but not the users of Manifold Markets! As of the 18th, our most liquid markets were predicting a 72% chance of invasion this year and a 59% chance of invasion before the end of the month.
Events like these provide further evidence of the power of prediction markets to cut through the thicket of emotionally and ideologically charged opinion to deliver solid forecasting.
There is still considerable uncertainty of the Russian invasion will be resolved. Our traders now give a 98% chance Russian troops will enter Kiev before the end of March, a 60% chance they control Lviv before April, and a 20% chance they invade Moldova before May.
You can follow these markets and more from our Russia Ukraine community.
EA Grant / X-Risk Tournament
We’re proud to announce that we been awarded a $200,000 grant from EA’s Long-Term Future Fund. A hearty thank you to the people of EA Funds, especially Jonas Vollmer and Linchuan Zhang, for their generous support! We look forward to working with the EA community on many topics, especially on forecasting existential and longer term risks.
To that end, we are considering hosting a tournament (with real cash prizes!) on x-risk related topics. As you might imagine, forecasting long-term risks is a good deal more complicated than creating a market on “Will AI/bioterrorism destroy the world by 2050? Yes/No”.
What we are looking for is to partner with an individual or institution with a deep understanding of AI/ML or epidemiology/pandemics to create a series of short-term markets on proxy questions directly relevant to forecasting longer-term existential risks (e.g. How well will X large language model work on Y benchmark? Will biosafety standards across Z global labs improve according to W metrics?)
If this sounds like something you’d be interested in, please reach out to me personally at stephen@manifold.markets!
Shoutouts
Scott Alexander, the blogger behind ACX (and our first backer!), has written about us again on the benefits of our “zero-sum, relative-accuracy play money system”.
Separately, Scott has created a series of conditional markets on the performance of potential blog posts as measured by likes on Substack. This is an interesting low-key experiment in futarchy, but runs the risk of incentivizing users to artificially pump up his likes!
Other shoutouts this week:
Thanks for all the support!
Keep on predicting,
This is a weird market to hold up as a triumph. Yes it eventually galvanized, but significant information came to light to update the market weeks earlier, and yet every time someone predicted yes, money mysteriously flooded in to hold the chances at exactly 50% until it was a foregone conclusion. (In fact, due to volatility, there were moments you could pick days before the invasion where the market was only 36%.) The consistent refusal of the market to update in light of new information was so mysterious, people even speculated it was the work of Russian manipulation. "Look at how great this market worked" is a very weird takeaway from this particular example. I would love to hear more analysis from manifold as to why this market was so slow to respond to outside information, and if it does anything to monitor for coordinated manipulation.
EDIT: I want to clarify, I don't have a strong assumption this market was "manipulated," there were many popular experts out there also predicting "no" until the last moment. Maybe the simpler answer is just a lot of people legitimately botched this one and I'm confused as to why.