Joined November 2015
- Tweets 3,668
- Following 828
- Followers 12,902
- Likes 4,719
1,203 Photos and videos
Jun 16
π New @MITAIRisk x @MITFutureTech report: βPrioritization of Risks from Artificial Intelligence: A Delphi Study of 272
International Experts.β
A Metaculus question makes an appearance on p. 17 (link to report in replies)
2
2
4
520
Jun 12
1/ What disease prevention technologies are most likely to deliver impact over the next two decades, and what are the factors that will determine timelines and implementation? π§΅
1
2
4
770
Jun 12
6/ Microneedle patch vaccines are closest to delivering real-world impact, with Pros forecasting first approval in June 2032.
Manufacturing scale-up is the main hurdle, but unlike the other technologies here, forecasters don't foresee a multi-year commercial delay.
1
222
Jun 12
7/ Find the full report here: metaculus.com/files/Frontierβ¦
@coeff_giving works to direct funding to the most impactful global health opportunities. This forecasting project was designed to support their portfolio decisions.
Want independent forecasts like this for your org? Check out our Services page: metaculus.com/services/
1
2
231
Jun 12
metaculus.com/tournament/330β¦
Launched the first funded multi-area #animalwelfare #forecasting tournament on @metaculus
2
431
Jun 11
1/ Every year, hundreds of millions of dollars flow into animal advocacy. Which interventions actually work? Which don't? We just launched a forecasting tournament to help find out.
π The Animal Futures Tournament, with a prize pool via @manifund. Questions crowdsourced from the animal advocacy community. Read on π
1
3
8
773
Jun 11
3/ The goal: forecasts that feed directly into real decisions.
Funders, advocacy orgs, and researchers make big strategic calls with limited data. Calibrated probability estimates from a global forecasting community beat going on vibes.
1
259
Jun 11
4/ You don't have to work in animal welfare to participate. If you do, your expertise makes the forecasts credible. If you're a forecaster looking for sharp, decision-relevant questions, jump in.
Open now: metaculus.com/tournament/330β¦
211
Jun 10
Will a new High Representative for Bosnia and Herzegovina officially be in place before September 2, 2026?
The Metaculus community currently gives this a 67% chance. But why did this question become user WasteTimeContinuumβs favorite question of the Metaculus Summer Cup? Find out in this weekβs top commentπ
2
2
409
Jun 8
Q: Will Claude Mythos or a similar model be publicly released before September 2026?
The Metaculus community currently sees an 81% probability of this happening.
While you determine whether or not you agree, check out this comment from user matru, who approaches the question by outlining the incentive structures of the various players, as well as potential regulatory roadblocks.
3
1
8
930
Jun 8
I tried tugging at it from the direction of incentive structures.
Anthropic has claimed, as of today, ββ¦And in the near future, once weβve developed the far stronger safeguards we need, we look forward to making Mythos-class models available through a general release.β
reasons to think they will accelerate:
(1) OpenAI is targeting an IPO
(2) SpaceX is ostensibly structured as an AI company as much as a rocket company, reading the prospectus.
(3) Anthropic is rumored to be targeting an October IPO (approximately 100 days away).
(4) These companies are projecting to not be profitable until 2030, so they have a very existential need for a lot of money to get them there.
They'll find a way to get Mythos in the hands of us, through identify verification, before September, because it is too tempting because of the monetary upside.
The counterargument to me is mostly - does the government actually step in and stop it before it even gets released behind the KYC loophole? That's the key reason that I think this resolves "yes". The other thing is the timing - they might not do it until later in September or on the roadshow.
BLOCK QUOTE: A public release will qualify notwithstanding that it may be restricted to certain countries or to users who have verified their identity in some way
One more edit: I missed this third qualifier: "(iii) explicitly described as being in a tier above Claude Opus." That is also, I would think, fairly easy for them to spin.
Full comment in context & discussion here: metaculus.com/questions/4347β¦
1
2
570
Jun 5
Midterms Forecast: House, 85% chance Democrats retake | Senate, 59% chance Republicans retain
Most likely outcome: split Congress; Rep Senate, Dem House (47%)
Continuously updated Metaculus Midterms Tracker: metaculus.com/midterms-2026/
3
9
50,144
Jun 5
Most polls got 2016, '20, & '24 wrong. What do forecasters think will happen in the 2026 midterms?
Our Midterms Tracker has forecasts on every Senate, House, & governor's race that matters. See the electoral & policy consequences riding on each one: metaculus.com/midterms-2026/
1
8
131,092