Above the Fold - Reflecting on 2022

Discussing how Manifold was founded and what we learnt in our first year!

Jan 6, 2023

Wow, 2022, and Manifold’s first full year of operating, comes to an end. And what a year it was!

For most people, New Year resolutions involve setting themselves goals which are quickly relinquished. At Manifold, it just means spending ages resolving all our end of year markets. 😅

Manifold Markets @ManifoldMarkets

Watching my phone for the new year countdown... The clock strikes midnight... the fireworks starts... and the market resolved push notifications start rolling in. Y'all are too addicted 😂

In this newsletter, we will reflect on what we learnt, and what we are still trying to figure out when it comes to creating a social network revolving around prediction markets.

Then as usual we will end with some further reading material recommendations and a summary of updates to our site. Enjoy!

What are some things we’ve learnt?

User-created markets works!

One day, at the end of 2021, our founder James was sky-diving out of a helicopter over the grand canyon. As he took in the magnificent view he thought to himself, why are there no forecasting sites that allow users to create and resolve their own prediction markets?

Hastily, he deployed his parachute, and before his feet even touched the ground he was messaging Stephen and Austin informing them of his ambitious plan. Austin then flew to Austin and with that they won the Scott Alexander grant and thus Manifold Markets was born.

Okay, that’s not exactly how it went down, although Austin did fly to Austin!! But it was crazy how quickly the product came together over Christmas in time to win the grant. And amidst this victory, there was still skepticism it could possibly work. Even YCombinator didn’t quite understand how users being responsible for resolving their own play-money markets could possibly work or be valuable.

The biggest problem with creator-resolved markets is that there are sometimes delays with the resolution if the creator hasn’t been on the site for a while. Other than that we are very happy with how user markets have developed. Fraud has happened, but to a very minimal extent, and almost entirely confined to markets which were created to be fun rather than provide serious information.

Users flock to prediction markets amidst uncertainty

It was quite surprising how much our DAU would spike when the media was trending about something that inflicts fear. We had a huge number of visitors during the days when Russia-Ukraine escalated and nuclear war was feared. Another big driver of traffic to our site was when FTX imploded and people were concerned about their investments, SBF, and EA’s reputation.

It’s great to know that we provided a little bit of certainty during these anxious times. Having a number users can quickly refer to can help ground reality.

People really care about play-money

Although we suspected this was the case, we didn’t quite realise how invested people would get in their profits. Many users who originally came to use our site to learn about forecasting and seeking information ended up staying because of how much they loved seeing their profits go up! This is a win-win as not only can it be more fun than traditional forecasting, but it also ensures markets are well-calibrated when people really care.

Manifold Markets @ManifoldMarkets

How well calibrated is Manifold Markets? Here's a plot which compares the probability of markets at the middle of their lifespan to how frequently they actually resolve to yes (the empirical probability). Pretty accurate!

Community experiments

Big shoutout to the Manifold community for all the cool experiments and data analytics done!

Data analytics on market accuracy
Self-resolving markets
Turns out having the foresight to create robust resolution criteria can be really hard! There was great discussion over the pros vs cons of intentionally leaving some ambiguity in order to ensure the resolution remained true to the spirit of the question while being as fair as possible.
“Interest rates”, ie markets don't go above/below x % due to opportunity loss with locking up capital

What are some things we are still figuring out

Product Market Fit

This is the biggest one, as at the end of the day it’s all about making a great product that provides value to as many people as possible. Although we have had good success in niche communities, we still haven’t quite found the product that reaches wider-spread adoption.

Hopefully, with the official launch of Maniswipe, Stonks, and our other plans for 2023, we can reach a large audience whilst still staying true to our existing users. Although we have continued to gamify and make Manifold more “casual” friendly, our main commitment continues to be ensuring accurate forecasting and its widespread adoption.

Content Curation & Discovery

Working out the best way to curate content so users see markets that are most relevant to them has become a big challenge, especially as new communities join that have less overlap with our original users. As it’s an interesting challenge, I thought I’d share some of the discussion and actions that have taken place behind the scenes.

The first thing we have to consider is the new user experience. Even if we can produce great personalised feeds for people who have used the site, it is important to make sure new users aren’t bombarded with an overwhelming amount of super silly junk. We also don’t want to intimidate them with niche, nerdy topics. A while back we introduced an onboarding flow with the ability to choose categories (groups) they are interested in. We also filter out some groups from the homepage by default that would otherwise flood certain feeds. You can view our opensource codebase to see the list of groups that are hard-coded to be excluded: https://github.com/manifoldmarkets/manifold/blob/main/common/envs/constants.ts.

However, there is still an interesting discussion of how we can segregate communities with different interests further. We are about to roll-out a new algorithm for recommending users more relevant markets on their home-feed and Maniswipe which is also exciting!

There have even been lengthy discussions in discord if allowing fun, silly markets on Manifold is a net positive, or if we should aim to reduce them to a minimum for everyone…

Moderation

During the moderation drama that went on with Twitter, we were having our own debates about how things should be handled on Manifold. Although we originally wanted to take an extremely hands-off approach, we realised that we still needed to protect our users from getting defrauded. And although we tried to add as much clarity as possible over what type of market crosses the line with edgy content, it is still a bit of a grey area. Should we allow markets about individual deaths? Does it matter whether they are a celebrity or not? Is discussing conspiracy theories fine when framed in ways that could be perceived as facilitating hate towards a group of people?

Site updates

Thanks for ending 2022 together with us. And all I have left to say is we hope you stick around for the traunching that will be 2023.

-David Chee

Above The Fold