---
title: 'The Butterfly Effect is Much Stranger Than We Thought'
source: 'https://youtube.com/watch?v=8RHLYIZCPG4'
video_id: '8RHLYIZCPG4'
date: 2026-11-11
duration_sec: 383
channel: 'Sabine Hossenfelder'
---

# The Butterfly Effect is Much Stranger Than We Thought

> Source: [The Butterfly Effect is Much Stranger Than We Thought](https://youtube.com/watch?v=8RHLYIZCPG4)

## Summary

This video explores a new scientific paper that suggests the butterfly effect, when combined with quantum physics, imposes a fundamental limit on weather prediction of 129 days. The presenter explains the difference between the popular and 'real' butterfly effect, discusses the implications for climate modeling, and offers a critical assessment of the research.

### Key Points

- **Quantum Butterfly Effect** [00:00] — Researchers show that quantum physics makes predictions impossible after a certain time, specifically for Earth's atmosphere.
- **Lorenz's Original Point** [01:25] — Edward Lorenz's real point was a 'Zeno effect' where improving precision only adds half the previous improvement, leading to a finite prediction horizon even with infinite precision.
- **129-Day Limit** [03:13] — The new paper calculates a finite prediction horizon for weather of 129 days, after which forecasting is theoretically impossible even with perfect data and models.
- **Quantum Uncertainty Source** [04:24] — The limit arises from unknown quantum states of sunlight, which spread from small to large scales through the atmosphere.
- **Climate Models Unaffected** [04:52] — Climate models are unaffected because they deal with averages, not specific events, so the butterfly effect doesn't change their predictions.
- **Paper Critique** [05:20] — The presenter rates the paper 5/10, skeptical of the claim that quantum effects fully erase initial state information, but acknowledges its conceptual importance.

### Conclusion

The video concludes that while the quantum butterfly effect may set a fundamental limit on weather prediction, it does not undermine climate modeling, and highlights the underrated intersection of quantum physics and chaos as a key research direction.

## Transcript

The butterfly effect is much stranger than we&nbsp; thought. A group of researchers just showed that&nbsp;&nbsp; if you take into account quantum physics, then it&nbsp; doesn't just make predictions difficult, it makes&nbsp;&nbsp;
predictions impossible after a certain time.&nbsp; They worked this out concretely for one of the&nbsp;&nbsp; most complicated chaotic systems we know, Earth's&nbsp; atmosphere. Does this mean that we can't predict&nbsp;&nbsp;
the climate either? Let's have a look. Stress can&nbsp; damage your health both physically and mentally.&nbsp;&nbsp; Ask me how I know. If you struggle with mental&nbsp; health problems or know someone who does, today's&nbsp;&nbsp;
paid partner BetterHelp can help. BetterHelp&nbsp; matches you online with a credential therapist&nbsp;&nbsp; based on your needs. You can talk by video,&nbsp; phone, or messaging, whatever works for you. I&nbsp;&nbsp;
think BetterHelp is a service that the world needs&nbsp; because especially with mental health problems,&nbsp;&nbsp; it shouldn't eat up someone's last remaining&nbsp; energy to find a therapist. For me, this was a&nbsp;&nbsp; real obstacle. Yes, I know they've seen criticism&nbsp; in the past, but they've made changes and now have&nbsp;&nbsp;
a 4.8 out of 5 rating in the App Store. More&nbsp; than 6 million people have gotten BetterHelp this&nbsp;&nbsp; way. Use betterhelp.com/sabine or my code during&nbsp; sign up for 10% off your first month. And now,&nbsp;&nbsp;
back to the science news. The name butterfly&nbsp; effect goes back to meteorologist Edward Lorenz&nbsp;&nbsp; who found out that the weather on our planet&nbsp; is extremely sensitive to small perturbations.&nbsp;&nbsp;
The flap of a butterfly's wings in Brazil could&nbsp; set off a tornado in Texas. More formally,&nbsp;&nbsp; if you start with two initial states that are&nbsp; just a tiny little bit different, then over time,&nbsp;&nbsp; the difference between them can grow dramatically.&nbsp; This part of the butterfly effect is quite well&nbsp;&nbsp;
known, but few people knew that this actually&nbsp; wasn't Lorenz's point. His point was what climate&nbsp;&nbsp; scientist Tim Palmer has called the real butterfly&nbsp; effect. If you read Lorenz's original paper,&nbsp;&nbsp;
you'll see that he was really arguing that weather&nbsp; prediction suffers from a sort of Zeno effect. If&nbsp;&nbsp; you improve the precision by which you know the&nbsp; initial conditions by a factor of say 2, then&nbsp;&nbsp;
you can increase the time over which you can make&nbsp; an accurate prediction by one half. Improve the&nbsp;&nbsp; precision by another factor of 2, and you add&nbsp; another one half squared, then one half cubed, and&nbsp;&nbsp; so on. This means that even if you could measure&nbsp; the initial conditions to infinite precision,&nbsp;&nbsp;
you'd not be able to make predictions beyond a&nbsp; certain time horizon, even though the system's&nbsp;&nbsp; fully deterministic. Whether this real butterfly&nbsp; effect is actually "real" in the sense of happening&nbsp;&nbsp;
in nature has never been settled, because Lorenz&nbsp; didn't, strictly speaking, derive it. He just made&nbsp;&nbsp; a compelling argument. I found it both interesting&nbsp; and somewhat concerning that this real butterfly&nbsp;&nbsp;
effect has gotten almost entirely lost in the&nbsp; literature, because if it happens in nature,&nbsp;&nbsp; it doesn't just happen for the weather forecast.&nbsp; It'd be a general property of chaotic systems&nbsp;&nbsp;
relevant for everything from nuclear fusion to&nbsp; quantum computing to black holes. In any case,&nbsp;&nbsp; the authors of the new paper now looked at this&nbsp; question, whether there is such a finite horizon&nbsp;&nbsp; for predictions for the weather, and they&nbsp; found that yes, there is, and it's 129 days.&nbsp;&nbsp;
After that time, they say weather forecast&nbsp; isn't just practically, but even theoretically&nbsp;&nbsp; impossible. Even if you knew the initial&nbsp; state perfectly, had a perfect model,&nbsp;&nbsp; and unlimited computing power, you still wouldn't&nbsp; be able to get past the 129 days. Their argument,&nbsp;&nbsp;
however, is rather different from the argument&nbsp; that Lorenz made. Lorenz made a general argument&nbsp;&nbsp; which said that inevitable changes on small scales&nbsp; like molecules would spread to larger and larger&nbsp;&nbsp; scales, eventually making predictions impossible.&nbsp; The authors of the new paper have sidestepped the&nbsp;&nbsp;
question of just what happens at the limit&nbsp; to infinitely small scales by instead asking&nbsp;&nbsp; what the physical causes for unavoidable&nbsp; disturbances are and what we know about&nbsp;&nbsp; them. Their answer is kind of obvious. Earth&nbsp; constantly receives quanta of light from the sun,&nbsp;&nbsp;
but we don't know the exact quantum state&nbsp; of the sunlight. We can't know it. Quantum&nbsp;&nbsp; physics forbids it. This quantum uncertainty&nbsp; first affects very small scales, but from there&nbsp;&nbsp; it spreads to larger and larger scales until&nbsp; it has spread through the entire atmosphere.&nbsp;&nbsp;
They calculate how long it take for the energy of&nbsp; the quanta to spread, and that gives them the 129&nbsp;&nbsp; days. What does this mean for climate models that&nbsp; make predictions over years or hundreds of years?&nbsp;&nbsp;
The maybe somewhat surprising answer is pretty&nbsp; much nothing because climate models already&nbsp;&nbsp; deal with averages. They'll tell you the average&nbsp; ocean temperature as a function of average carbon&nbsp;&nbsp; dioxide or the probability of certain extreme&nbsp; events. They won't tell you whether it rains&nbsp;&nbsp;
in Paris on July 14th, 2030. The calculation&nbsp; for the butterfly effect tells you how long it&nbsp;&nbsp; takes until average prediction is the best you&nbsp; can do. It doesn't make the average prediction&nbsp;&nbsp;
any better or worse. I give this paper a 5&nbsp; out of 10 on the [&nbsp;__&nbsp;] meter. That's because&nbsp;&nbsp; I'm not super convinced by their argument that&nbsp; quantum effects entirely erase the information&nbsp;&nbsp; of the initial state. Still, I think that this&nbsp; paper's conceptually an important contribution&nbsp;&nbsp;
to the literature and I wish that physicists in&nbsp; other disciplines would pay more attention to&nbsp;&nbsp; this. The intersection of quantum physics and&nbsp; chaos is maybe the most underrated research&nbsp;&nbsp;
direction in the foundations of physics today.&nbsp; When you want to know if there's a fundamental&nbsp;&nbsp; limit to the precision your quantum computer can&nbsp; reach before spending a few billion dollars on it,&nbsp;&nbsp; maybe that's just me. If you try to&nbsp; predict what someone will say from what&nbsp;&nbsp;
they've previously said, how many words does&nbsp; it take until you lose all predictive power?&nbsp;&nbsp; Broccoli. Do you think ChatGPT would have made&nbsp; this joke? Thanks for watching. See you tomorrow.
