Jonathan Mann
@jonathanmann.bsky.social
I'm a forecasting professional with Good Judgment (Superforecaster), RAND (INFER Pro), and Samotsvety. I currently work as a Security Architect and was previously a Data Scientist on Wall Street. I also teach computer science as adjunct faculty at NYU.
Discovering Better World Models
open.substack.com/pub/abstract...
open.substack.com/pub/abstract...
Discovering Better World Models
separating discernment from delusion
open.substack.com
April 24, 2025 at 5:35 PM
Discovering Better World Models
open.substack.com/pub/abstract...
open.substack.com/pub/abstract...
Reposted by Jonathan Mann
"How, exactly, could AI take over by 2027?"
Introducing AI 2027: a deeply researched scenario forecast I wrote with Daniel Kokoatjlo, Scott Alexander, and others.
The scenario: ai-2027.com
PDF version: ai-2027.com/scenario.pdf
Introducing AI 2027: a deeply researched scenario forecast I wrote with Daniel Kokoatjlo, Scott Alexander, and others.
The scenario: ai-2027.com
PDF version: ai-2027.com/scenario.pdf
AI 2027
A research-backed AI scenario forecast.
ai-2027.com
April 3, 2025 at 11:16 PM
"How, exactly, could AI take over by 2027?"
Introducing AI 2027: a deeply researched scenario forecast I wrote with Daniel Kokoatjlo, Scott Alexander, and others.
The scenario: ai-2027.com
PDF version: ai-2027.com/scenario.pdf
Introducing AI 2027: a deeply researched scenario forecast I wrote with Daniel Kokoatjlo, Scott Alexander, and others.
The scenario: ai-2027.com
PDF version: ai-2027.com/scenario.pdf
Reposted by Jonathan Mann
When you write your post about which experts to trust, make sure to mention exponential weights. Even when the experts select their predictions adversarially, exponential weights is only O(\sqrt{T}) worse than the best expert in hindsight. See the proof here:
www.mit.edu/~9.520/sprin...
www.mit.edu/~9.520/sprin...
www.mit.edu
January 17, 2025 at 4:42 AM
When you write your post about which experts to trust, make sure to mention exponential weights. Even when the experts select their predictions adversarially, exponential weights is only O(\sqrt{T}) worse than the best expert in hindsight. See the proof here:
www.mit.edu/~9.520/sprin...
www.mit.edu/~9.520/sprin...
Reposted by Jonathan Mann
What is everyone wrong about? Exaggerated, unnuanced takes 🧵
Forecasting.
Everyone is wrong about why forecasting is good for most people. It's a way to learn if they are bullshitting uncorrelated with the career success or social approval.
x.com/s_r_constan...
Forecasting.
Everyone is wrong about why forecasting is good for most people. It's a way to learn if they are bullshitting uncorrelated with the career success or social approval.
x.com/s_r_constan...
January 15, 2025 at 12:41 AM
What is everyone wrong about? Exaggerated, unnuanced takes 🧵
Forecasting.
Everyone is wrong about why forecasting is good for most people. It's a way to learn if they are bullshitting uncorrelated with the career success or social approval.
x.com/s_r_constan...
Forecasting.
Everyone is wrong about why forecasting is good for most people. It's a way to learn if they are bullshitting uncorrelated with the career success or social approval.
x.com/s_r_constan...
When should you actually spend time forecasting a decision?
Seldom. But when you should, you really should.
New post on knowing when forecasting matters (and when it doesn't):
abstraction.substack.com/p/when-to-us...
Seldom. But when you should, you really should.
New post on knowing when forecasting matters (and when it doesn't):
abstraction.substack.com/p/when-to-us...
When to Use Forecasting for Decisions
seldom needed, but don't neglect it when it is
abstraction.substack.com
December 13, 2024 at 4:01 PM
When should you actually spend time forecasting a decision?
Seldom. But when you should, you really should.
New post on knowing when forecasting matters (and when it doesn't):
abstraction.substack.com/p/when-to-us...
Seldom. But when you should, you really should.
New post on knowing when forecasting matters (and when it doesn't):
abstraction.substack.com/p/when-to-us...
Reposted by Jonathan Mann
New in NHB: We systematically replicated 26 of 41 PNAS social science studies that used MTurk. The approach provides a proof-of-concept for using decision markets to select findings to replicate.
Replication effect sizes were 45% of original effect sizes.
www.nature.com/articles/s41...
Replication effect sizes were 45% of original effect sizes.
www.nature.com/articles/s41...
Examining the replicability of online experiments selected by a decision market - Nature Human Behaviour
This study finds that decision markets can be a useful tool for selecting studies for replication. For a sample of 26 online experiments published in PNAS selected by a decision market, the authors fi...
www.nature.com
November 19, 2024 at 11:59 AM
New in NHB: We systematically replicated 26 of 41 PNAS social science studies that used MTurk. The approach provides a proof-of-concept for using decision markets to select findings to replicate.
Replication effect sizes were 45% of original effect sizes.
www.nature.com/articles/s41...
Replication effect sizes were 45% of original effect sizes.
www.nature.com/articles/s41...
Reposted by Jonathan Mann
psa: likes are public here
November 18, 2024 at 10:23 PM
psa: likes are public here
Reposted by Jonathan Mann
Make a hypothesis about what you expect to see, every step of the way. A manifesto:
statmodeling.stat.columbia.edu/2024/11/13/m...
statmodeling.stat.columbia.edu/2024/11/13/m...
Make a hypothesis about what you expect to see, every step of the way. A manifesto: | Statistical Modeling, Causal Inference, and Social Science
statmodeling.stat.columbia.edu
November 13, 2024 at 2:28 PM
Make a hypothesis about what you expect to see, every step of the way. A manifesto:
statmodeling.stat.columbia.edu/2024/11/13/m...
statmodeling.stat.columbia.edu/2024/11/13/m...
Reposted by Jonathan Mann
In these difficult times, it is worth remembering that the climate outlook, though dark, has improved over the last 10 years. We can change our course.
Source: www.economist.com/interactive/...
Source: www.economist.com/interactive/...
November 15, 2024 at 9:13 AM
In these difficult times, it is worth remembering that the climate outlook, though dark, has improved over the last 10 years. We can change our course.
Source: www.economist.com/interactive/...
Source: www.economist.com/interactive/...