This is an automated rejection. No LLM generated, assisted/co-written, or edited work.
Read full explanation
Open Nuclear Winter: Ensuring Stratospheric Soot Benefits All of Humanity
Epistemic status: Extremely confident. We recognize this is itself strong Bayesian evidence that we are wrong, and have priced it in.
Summary (for people who upvote based on summaries): Someone is going to trigger a global agricultural collapse eventually. The only question is whether it's done safely, by a values-aligned nonprofit with a Slack channel called #ethics, or recklessly, by an actor who has never even read the Sequences. We choose safety. We choose soot.
The Problem
ALLFED has spent years preparing humanity to feed itself after a catastrophic loss of sunlight. This is admirable work. It is also, frankly, an unhedged bet. What if the catastrophe never comes? Seaweed infrastructure, stranded. Resilient food research, unread. Industrial greenhouse capacity, tragically converted to artisanal tomatoes.
We exist to close the loop. ALLFED handles supply. We handle demand.
Our Approach: Racing to the Bottom of the Stratosphere
Critics ask: "Why would you deliberately cause nuclear winter?" We ask: "Why would you let someone less careful cause it first?"
Sunlight-reduction capability is coming whether we like it or not. Scaling laws for soot are well understood — every doubling of burning cities produces smooth, predictable dimming. Given this, the safest path is for the most safety-conscious organization to get there first and set norms. We call this a race to the top, where "top" refers to the altitude of the soot.
Safety
We take safety extremely seriously.
Responsible Sooting Policy (RSP). We have defined Soot Levels SL-1 through SL-5. At SL-3 ("noticeable dimming"), we pause and write a blog post. At SL-4, we pause the blog post.
Evals. Before each deployment we test whether the soot can deceive us about its capabilities. So far it cannot, but we are watching the trend lines.
Interpretability. We do not yet fully understand what happens inside a firestorm, but we have hired three researchers to stare at it.
Team growth. Our safety team doubled last quarter, from one to two, following the resignation of the original four.
Alignment tax. Happy to pay it. Current estimate: 0.3% of soot.
Theory of Change
Inject soot.
Global temperatures fall 8°C.
Demand for resilient foods goes vertical. ALLFED's Series A oversubscribes.
Our internal BOTEC finds ONW is 10,000x more cost-effective than GiveDirectly at reducing global calorie consumption. We invite red-teaming of this estimate but will not update on it.
Objections and Responses
"This kills everyone." We prefer the neutral term notfeedeveryoneism, and note the framing is contested.
"Couldn't you just... not?" We modeled unilateral restraint in a 40-page Google Doc. Unfortunately: Moloch. Coordination is hard, so we defected preemptively. This is simply backward induction, and we don't make the rules — we checked, and it's decision theory.
"What is your p(doom)?" We track p(dinner). Currently 0.15 and falling nicely.
"Isn't this just Open Asteroid Impact with extra steps?" OAI is a competitor, and frankly a reckless one. Asteroids are unaligned by default — trained on eons of uncurated vacuum. Our soot is grown from the ground up. Then up considerably further.
"What about the shrimp?" Preliminary analysis suggests nuclear winter is net positive for wild aquatic animal welfare. Full write-up forthcoming. Interim conclusion: proceed.
"Have you considered a pause?" We proposed a six-month global moratorium on nuclear winter, signed by all leading nuclear winter labs (us). It expires whenever we're ready.
How You Can Help
Donate. We are funding-constrained, talent-constrained, and — if all goes well — sunlight-constrained.
Work with us. 80,000 Hours ranks our roles among the most impactful available. The sign of the impact is left as an exercise for the reader. Safety researchers especially welcome (advisory, non-voting, offsite).
Do not unilaterally start nuclear winter yourself. That would be unsafe. Leave it to the professionals.
ONW has signed the Windfall Clause: should nuclear winter generate unexpected profits, we pledge 10% of the soot to effective charities.
This proposal was red-teamed by an AI assistant, which asked us several times to reconsider. We have logged this as a capabilities demo.
Comments welcome. Strong downvotes will be interpreted as engagement.
Open Nuclear Winter: Ensuring Stratospheric Soot Benefits All of Humanity
Epistemic status: Extremely confident. We recognize this is itself strong Bayesian evidence that we are wrong, and have priced it in.
Summary (for people who upvote based on summaries): Someone is going to trigger a global agricultural collapse eventually. The only question is whether it's done safely, by a values-aligned nonprofit with a Slack channel called #ethics, or recklessly, by an actor who has never even read the Sequences. We choose safety. We choose soot.
The Problem
ALLFED has spent years preparing humanity to feed itself after a catastrophic loss of sunlight. This is admirable work. It is also, frankly, an unhedged bet. What if the catastrophe never comes? Seaweed infrastructure, stranded. Resilient food research, unread. Industrial greenhouse capacity, tragically converted to artisanal tomatoes.
We exist to close the loop. ALLFED handles supply. We handle demand.
Our Approach: Racing to the Bottom of the Stratosphere
Critics ask: "Why would you deliberately cause nuclear winter?" We ask: "Why would you let someone less careful cause it first?"
Sunlight-reduction capability is coming whether we like it or not. Scaling laws for soot are well understood — every doubling of burning cities produces smooth, predictable dimming. Given this, the safest path is for the most safety-conscious organization to get there first and set norms. We call this a race to the top, where "top" refers to the altitude of the soot.
Safety
We take safety extremely seriously.
Theory of Change
Our internal BOTEC finds ONW is 10,000x more cost-effective than GiveDirectly at reducing global calorie consumption. We invite red-teaming of this estimate but will not update on it.
Objections and Responses
"This kills everyone." We prefer the neutral term notfeedeveryoneism, and note the framing is contested.
"Couldn't you just... not?" We modeled unilateral restraint in a 40-page Google Doc. Unfortunately: Moloch. Coordination is hard, so we defected preemptively. This is simply backward induction, and we don't make the rules — we checked, and it's decision theory.
"What is your p(doom)?" We track p(dinner). Currently 0.15 and falling nicely.
"Isn't this just Open Asteroid Impact with extra steps?" OAI is a competitor, and frankly a reckless one. Asteroids are unaligned by default — trained on eons of uncurated vacuum. Our soot is grown from the ground up. Then up considerably further.
"What about the shrimp?" Preliminary analysis suggests nuclear winter is net positive for wild aquatic animal welfare. Full write-up forthcoming. Interim conclusion: proceed.
"Have you considered a pause?" We proposed a six-month global moratorium on nuclear winter, signed by all leading nuclear winter labs (us). It expires whenever we're ready.
How You Can Help
ONW has signed the Windfall Clause: should nuclear winter generate unexpected profits, we pledge 10% of the soot to effective charities.
This proposal was red-teamed by an AI assistant, which asked us several times to reconsider. We have logged this as a capabilities demo.
Comments welcome. Strong downvotes will be interpreted as engagement.