James Grugett

Karma: 52

James Grugett 21 Jun 2024 17:45 UTC
6 points
1 ∶ 0
in reply to: Ben Millwood🔸’s comment on: Manifold markets isn’t very good
Ah, yes, then we have succeeded in normalizing the idea that you should be able to create a market, trade in it, and then resolve it according to your own judgment.
When we were getting started literally everyone we told this to, including YC partners, thought it was crazy.
Only Scott Alexander saw the value in users creating and resolving their own markets. He awarded us the ACX grant, shared the link to our prototype with his readers, and the rest is history!

James Grugett 20 Jun 2024 23:38 UTC
1 point
1 ∶ 1
in reply to: Jason’s comment on: Manifold markets isn’t very good
Allowing users to create any question they want with no moderation gate (unlike Metaculus or any other prediction market site) is a big part of Manifold’s success (even business success!). We further empower users to judge and resolve their markets. While not everyone is perfect or is always acting in good faith, this system largely works.
This openness has been key, but is it the same thing as allowing anyone to go to our conference? Not exactly, but they are related. Another example is that the scheduling software allowed anyone to book any time slot for any room at Manifest, and we basically didn’t have any problems there.
My take is that:
1. the “controversy” is way overblown based on an unfair connecting of dots from journalists at The Guardian who didn’t even attend. The actual Manifest was amazing and people who attended were nearly unanimous on that point, except:
2. There may have been one person at Manifest who was both edgy and impolite, bordering on aggressive. If we could have kicked that person out, then the anonymous EA forum poster wouldn’t have ever felt threatened, and we wouldn’t be having this conversation.
So, I do agree that some moderation can be helpful with business goals, just like we have had to ban some users from our site. But you need less moderation than you would think!

James Grugett 20 Jun 2024 19:43 UTC
48 points
8 ∶ 1
on: Manifold markets isn’t very good
Hi, I work on Manifold! Thanks for the critical take — we’re always happy to hear these issues so we can continue improving.
IMO this is the crux:
Metaculus’ self-measured Brier score is 0.111, compared to Manifold’s 0.168
While Metaculus does perform better (seriously impressive work!), we have >130,000 user-created markets across all matter of topics at various levels of seriousness. Personal questions tend to have higher brier scores than more objective science questions for example.
When compared to other prediction market sites (Kalshi, Polymarket), we generally perform better. See for example https://calibration.city
If you pick a topic that you think is important, I think you will find that Manifold is contributing a lot of relevant forecasts.
For example, we have >1,000 questions just in the topic “AI Safety”. I think getting a good forecast on all those questions is a valuable service to the world.
You mentioned that Manifold did worse this year in the ACX forecasting contest. That worries me as well, though I will note that the year before we did “extraordinarily well” placing in the 99.5% percentile: https://www.astralcodexten.com/p/who-predicted-2022
Markets are a different forecasting mechanism than polling and I think they have a lot of contribute to the forecasting space. In particular, I think that scaled up markets have hope to be the most accurate forecasting method, because there can be large incentives to find the most accurate price. Any bias you see in markets can also be rectified with trading strategies to correct them, which give feedback in terms of profitability.
We have recently changed the Manifold economy significantly to reduce bonuses and prevent the kind of exploits where accounts could go negative, so a lot of your points here have been addressed.
Regarding Manifest and controversial attendees, we kept the same ethos as a our site, where anyone can create markets. On balance, we think it creates more value for the world to not shut down ideas we might personally disagree with. Much more has been said on this topic elsewhere.

James Grugett 5 Sep 2022 19:48 UTC
1 point
0 ∶ 0
on: Cause Exploration: Governance Design & Formation
The solution I’ve been thinking of is competitive governance: Allow total freedom for new systems of governance wherever land owners opt in to it. Then, people vote with their feet.
Founders can raise money with a vision for their own ideal society. If they can offer an attractive set of laws and regulations that citizens like, from tax policy, social safety nets, education, policing, etc., then they can grow as more move in or more land owners opt-in to those rules.
This is actually how businesses work. The crucial part is not the corporate governance structure of shareholders and a board of directors IMO. It’s that 1. businesses have wide leeway to govern as they please and 2. customers opt-in to paying the business and employees opt-in to working there. If a business stops being effective, customers and employees can switch to a competitor.
The freedom to create and the freedom to opt in are important features of many successful mechanisms. E.g. This is how Manifold Markets works. Anyone can create any question, and you opt-in to bet on it. If someone resolves a question in a way you disagree with, then you just stop betting in their markets. Vote with your feet.
Markets are the ultimate opt-in mechanism: you and another party mutually agree to a transaction. It works because mutual agreement is positive-sum, otherwise at least one side would decline the exchange.
Utilitarians most of all should approve these positive-sum exchanges. The world could do worse than apply this mechanism everywhere. And the one place it’s very obviously missing is in governance!