A sizable chunk of the endowment likely has legal restrictions that limit how funds can be spent. E.g., they could be earmarked for undergraduate scholarships or a specific lab at a specific department. The endowment isn't a slush fund.
It's also worth noting that the structural costs of research are far larger than what any single institution would be able to shoulder. For instance, MIT has extremely limited supercomputing resources under their own maintenance. Researchers would typically use such resources from centralized places funded by the NSF or DOE, where larger pools of money can be assembled.
And of course this doesn't even get into the reality that the annual operating costs of somewhere like MIT likely far exceeds the investment returns generated by the endowment.
You might as well argue that companies should never take venture capital - e.g. if they can't finance their growth through profits alone then they shouldn't raise any money. The whole point of grants or investment is to subsidize and incentive work which has payoffs on much longer timescales than what market dynamics can sustain alone.
> A sizable chunk of the endowment likely has legal restrictions that limit how funds can be spent. E.g., they could be earmarked for undergraduate scholarships or a specific lab at a specific department. The endowment isn't a slush fund.
Some of it has some restrictions, but money is fungible. I do not believe that MIT is actually limited (in practice) from writing their own grants because of donor restrictions (if they wanted to).
> And of course this doesn't even get into the reality that the annual operating costs of somewhere like MIT likely far exceeds the investment returns generated by the endowment.
Somehow they spend $1.2B/year on administration, so, yeah. Don't do that. But they easily have enough principal to cover grant funding for the remaining years of this administration. Especially if they can play on their lib donor heart-strings about how mean the current administration is being to them.
The vast majority of the endowment isn't money (dollars in bank accounts). University endowments work like private equity funds, most of the funds will be invested in assets, most of which hardly liquid enough to reasonably convert them into cash on short notice. They could try to borrow money against the valuations of those assets, but it's not sane to take on debt in order to sustain a level of expenditures that was adjusted to a much higher level of income (true more generally). Especially when the alternative of temporarily scaling back expenses is relatively easy.
Money is not fungible when you are a large organization. Many things that should be possible in principle are impossible in practice due to rules, politics, and institutional inertia.
MIT's endowment is ~80% earmarked to whatever purposes the donors considered important. The remaining ~20% is unrestricted, but unrestricted does not mean unallocated. Everything has already been allocated to some purpose, at least implictly. If you want to allocate more money towards something, you need to take that money from somewhere else. And then you get politics.
>practice due to rules, politics, and institutional inertia.
Isn't that the responsibility of the dean to fix? I think a lot of us have no idea how this actually works, but do understand the difference between impossible and hard. This seems more like it's on the hard side than impossible.
A dean is a mid-level manager in an organization, where effective power is widely diffused.
Many things are possible in the same sense as rewriting the US constitution. The mechanism for it exists, but using it in practice would require widespread agreement on the specifics. When there are many people making independent decisions, it's best to see the situation in statistical terms. Outcomes that are too many standard deviations away from the expected are effectively impossible.
> Especially if they can play on their lib donor heart-strings about how mean the current administration is being to them.
Yes, like those famous liberals the Koch family who paid for prime real estate across from Stata.
It's just not as simple as you lay it out to be. Do you _seriously_ think that if hunkering down and paying out of the endowment to sustain nominal operations for a few short years was a viable strategy that they wouldn't be doing just that?
I think this is a valid point, but if the talent pool shrinkage was truly a threat to your academic institution are you really going to just watch?
And the argument is that research funding is coming back but just not to MIT. So I think it is a serious long term issue that they have to consider going forward, and not something that they can just hope goes away.
It would actually be _more_ competitive, because what's driving the reduction in admissions is uncertainty in grant/funding availability.
That means fewer available slots overall. Kornbluth's comments don't explicitly state anything about _applications_, just _admissions_. Given the heightened economic uncertainty and poor job prospects for recent graduates, I'd expect more students to be looking for graduate school as a way to tide themselves over.
So a very, very bad picture for folks seeking graduate education and training.
> Weather forecasts have always been an inexact science
Weather forecasting is anything but "an inexact science." It's extremely exact up to the limitations and assumptions you impose on your model due to resource constraints.
And yes - I assume that this is what you mean by "an inexact science." But still in 2026 I regularly meet people who think that weather forecasting is the same as astrology, completely ignorant of massive amount of physical scientific understanding that goes into it.
> Weather forecasting is anything but "an inexact science." It's extremely exact up to the limitations and assumptions you impose on your model due to resource constraints.
It's "extremely exact" but our models are not good enough. So... inexact?
The reality is that we don't have the technology to model the physical world with extreme accuracy. If we did, we would be able to predict the future, and not just weather events. The world's most powerful supercomputers can model atmospheric conditions pretty well, and they've certainly improved over time, but there are still a lot of variables unaccounted for.
This is why I think that ~90% accuracy for a few days in advance[1] is good enough for most people. A smartphone app won't miraculously make this better, no matter how pretty or "fun" it is.
> It's "extremely exact" but our models are not good enough. So... inexact?
That's not the common way that the phrase "inexact science" is used. All modeling involves approximations at some levels, but you wouldn't turn around and call it "inexact science."
> ... but there are still a lot of variables unaccounted for.
Such as... ?
This is the problem with throwing away colloquialisms like "inexact science." What, specifically, is a "variable" that is unaccounted for that would unlock improved forecast accuracy or to push thresholds closer to the predictability limits?
> This is why I think that ~90% accuracy for a few days in advance[1] is good enough for most people. A smartphone app won't miraculously make this better, no matter how pretty or "fun" it is.
I agree, which is why the other portions of your comment come off poorly.
> No. But I'd suspect a tabula rasa approach to weather–particularly given it hasn't been rolled out globally in one go–incorporates satellite data, local measurements, et cetera.
There most likely won't ever be such an effort - even in companies that are targeting verticalization of the "weather supply chain" (proprietary observations + models + decision support tools) - if only because it would be utterly foolish to exclude the vast amounts of data collected by government agencies and the wide variety of players in the weather enterprise. At best, verticalized weather companies can produce niche value over baseline from the single modality of proprietary data they collect.
The infrastructure for observing and forecasting the weather is incredibly sophisticated, and has been evolving for about 150 years at this point. The quality of contemporary numerical weather prediction likely doesn't leave much headroom towards the threshold of fundamental physical limitations on predictability. This is why there are groans and eye rolls from the weather community each time a new player steps forward with yet-another-AI-model-trained-on-ERA5-reanalysis and boasts some comically small improvement in average forecast skill.
With all that being said, there's likely an exciting frontier opening up as the AI models push towards encompassing data assimilation. But the applications that start to become extremely interesting there won't have any noticeable impact on average forecast quality for your typical weather app.
> it would be utterly foolish to exclude the vast amounts of data collected by government agencies
Never suggested this. You use the government data. And you supplement with specialist sources. If you’re near any avalanche areas, for example, your snow forecasts typically have an additional layer of resolution available if you know where to look.
> Do any feel-like estimates take cloud cover into consideration?
No, usually not, because they're usually just simple toys combining a heat index and wind chill scale.
There _is_ an official metric used for estimating heat stress that accounts for cloud cover - the Wet-bulb Globe Temperature (e.g. https://www.weather.gov/tsa/wbgt). This is what is used, for instance, in literature analyzing the impact that future climate change might have on heat stress and mortality risk during heat waves. It's also used by some professional sports programs to monitor risk for crowds and athletes, as well as commonly used by OSHA and other regulatory agencies looking at worker exposure to heat hazards.
It would just be regurgitating information from numerical forecast models.
If you're in the Northeast and have questions about the significant winter storm that is impending, please check out the National Weather Service's forecast and decision support materials for your community on www.weather.gov.
He has a knack for being scarily prescient. I didn't expect we would seriously be discussing geoengineering in the 2020's (I gave it until at least the early 2030's, given the technical complexity of building the actual delivery system for any planet-scale intervention), but here we are.
> Wait... So, to undo it all we have to do is stop doing it? Doesn't this contradict the statement right before it?
It's not quite that simple.
The intuition that you're subtly relying on is the idea that the response or effect of one of these geoengineering treatments is linear. But unfortunately, that's not something you can assume about a dynamical system. In reality, the climate system can undergo certain types of hysteresis where "undoing" the forcing doesn't revert the initial perturbation, because you're suddenly on a different response curve. Probably the most famous example of this in climate dynamics is the way that the ice-albedo effect sets up a hysteresis in the trajectory towards a "snowball Earth" scenario. Apologies for the lack of links/references; Wiki has decent write-ups on this, and it's typically covered in the first chapter of a climate dynamics textbook.
The potential response to suddenly stopping a climate change mitigation strategy has a very well-popularized name: a "termination shock." In fact, Neal Stephenson used exactly this concept in his titular novel on the topic in 2021.
As a climate scientist, my mental model to better understand the risk of termination shocks and unintended consequences boils down to how fast the response of the climate system is. Marine brightening is "less risky" because the meteorological response to these interventions is extremely fast; a cloud-precipitation system will respond on the order of minutes to hours, and unless the intervention continues unabated, it will clean the air quickly, limiting the repsonse. Stratospheric aerosol injection is more complicated, but we have a very good analogue - very large scale volcanic eruptions like Mt Pinatubo. The response to these sorts of events is measured more on the timescale of 2-5 years, although knock-on effects (such as a shift towards more diffuse solar radiation reaching the surface, which has significant effects on terrestrial and oceanic biogeochemistry) could very much persist longer than that - and don't "snap back" nearly as quickly. A continuous, Pinatubo-like intervention would compound and introduce coupled atmosphere/ocean responses that could decade years or longer to fully play out. And that's _in addition_ to the near immediate (1-2 year) response in global average temperature, which would bounce back to most of the pre-intervention level very quickly.
These things are complex. There's a lot we don't know. But, there's also a lot we _do_ know. I would encourage anyone who does not have significant experience in climate dynamics to remain curious and avoid jumping to conclusions based on simple intuition; they're probably wrong.
Thank you for this response. Of those that replied to me, yours seems the most balanced and scientific, and I learned the most from. I wish more often people engaged on HN like you have here.
Given your expertise in this, I'm curious what your take is on CO2 capture, not in terms of economic viability, but in terms of climate risk...
For example, if we were to discover a process that removed CO2 from atmosphere and converted it into a product profitably such that there was an economic incentive/positive feedback loop to remove CO2.
My intuition is that if we removed the CO2 too quickly or too much of it we may have unwanted consequences, but if the rate was managed and we slowed down and stopped at a certain equilibrium, would this be a theoretically ideal way to address the problem?
First, what is "too quickly" with reference to CO2 removal from the atmosphere? At present, human civilization emits over 40 gigatons - or 40 trillion kilograms - of CO2 per year. And that increases the atmospheric burden by about 2.5 parts per million per year. So today, before you even start _reducing_ atmospheric CO2, you need to be sucking down at least 40 trillion kilograms of CO2. I struggle to imagine a scenario outside of total science fiction where that would be remotely possible.
Second, the equilibrium climate response to changes in greenhouse gas forcing take on the order of decades or centuries to realize. This is because the dynamics of the climate system are heavily buffered. For example, the ocean acts as a giant heat capacitor that slowly interchanges with the atmosphere. Were you to instantaneously halve the CO2 in the atmosphere, you'd likely see a pretty classic exponential decay in global average temperature (and other more nuanced responses); in the present climate, it's not clear we have already passed specific "tipping points" that would induce that hysteresis I described in the previous comment (in fact - one could read "climate tipping point" as a synonym for dynamical system hysteresis). Theoretically, one could try to "dial in" some particular equilibrium climate state, but it's not obvious over what timescale you'd have to intervene and what the cost of such an intervention would be.
The cool thing is none of this needs to be purely "theoretical." You could simulate all of this _today_ if you had a setup to run a global climate model. A "4X CO2" experiment where you branch from an equilibrium spin-up climate state and immediately apply a global quadrupling of CO2 has been a completely standard experiment as part of CMIP for over two decades. The opposite experiment is an established protocol for both the Carbon Dioxide Removal Intercomparison Project [1], which features an annual ramp down of CO2 at a 1% per year rate, and the Cloud Feedback Model Intercomparison Project [2], which features a more direct counterpart, with an abrupt decrease of atmospheric CO2 by 50%. There is a large body of literature discussing the results of these classes of experiments, but this is outside of my primary research focus and I can't turn you to particularly good papers off-hand. But they're easy enough to find.
It's also worth noting that the structural costs of research are far larger than what any single institution would be able to shoulder. For instance, MIT has extremely limited supercomputing resources under their own maintenance. Researchers would typically use such resources from centralized places funded by the NSF or DOE, where larger pools of money can be assembled.
And of course this doesn't even get into the reality that the annual operating costs of somewhere like MIT likely far exceeds the investment returns generated by the endowment.
You might as well argue that companies should never take venture capital - e.g. if they can't finance their growth through profits alone then they shouldn't raise any money. The whole point of grants or investment is to subsidize and incentive work which has payoffs on much longer timescales than what market dynamics can sustain alone.