A critique of longtermism: why you shouldn't worry about the far future
gautamcgoel.substack.com
gautamcgoel.substack.com
There's longtermism and then there's longtermism. Climate Change is a problem that will affect our kids and grand kids in this century and the next, hence it's a longterm problem.
But worrying about the future diaspora of "potentially hundreds of trillions or more" humans across the Milky Way Galaxy is so far off and unforeseeable I'm not even sure "longtermism" is the right word for it. That's more SciFi territory.
It's kinda tiresome all these -isms that people invent to argue against. More often they're strawman positions that few to no people actually hold. We should hold people to counter specific arguments made by specific people, not vague ideas held by nebulous -ists.
+ I am so excited about <potential of this development/timeline/discover>, what an amazing step towards Mars colonization.
- Mars colonization is a waste of everyone's time. It's a giant distraction, and a truly gargantuan resource sink, that doesn't get us anything meaningful. We should be focusing on immediate problems, like hunger and climate change and nuclear weapons and toilet paper shortages.
+ Mars colonization creates redundancy for humanity against extraterrestrial existential threats. It's important to direct our efforts to ensure that <some tiny fraction> of 'humanity' can survive an Earth-killing cataclysm.
- Extraterrestrial existential threats are unlikely and hypothetical on any appreciable timeline. We are spectacularly failing to deal with a short-term, man-made existential threat to our civilization. In the <near>/<long> term, I'd rather direct <gargantuan efforts required for building a self-sustaining Mars colony> towards saving <overwhelming majority> of humanity from <impending, obvious catastrophe>, than work to preserve <some tiny fraction> of humanity from <highly unlikely cataclysm that somehow kills off 100% of the people on Earth but 0% of the people on a hypothetical self-sustaining Mars colony>.
Rather, it would seem that solving climate change will give us the tool to deal with space colonization, as space colonization of new homes fundamentally involves managing new ecology and life support systems.
https://www.currentaffairs.org/2021/07/the-dangerous-ideas-o...
It's common for people concerned about an impending AI apocalypse to use this sort of "longtermist" argument. For example this article:
https://oxfordpoliticalreview.com/2019/08/25/is-ai-safety-ra...
If we reduce existential risk by mere one-millionth of one percentage point, it will be worth more than 100 times the value of saving a million human lives. The expected value of any other good actions – like helping people here and now – will be trivial compared to even the slightest reduction in existential risk.
Similarly, the recent book The Precipice argued that the most important risk for humanity is AI risk because the badness of AI destroying all humanity is much worse than, say, climate change killing off 99% of humans.
Not your primary point, but I think we can reduce the risk of losing some future war against aliens by discovering the aliens sooner. The Aztecs would have been a lot better off if they had realized Cortés was coming a decade earlier....
Ideologically 'longtermism' functions the same way as arguments about 'existential risk'. They're often highbrow ways to justify authoritarian stances. To give you a concrete example, Bostrom's writing contains a lot of arguments along the lines of "if there is an infinitesimally small risk of some terrorist making a bioweapon that wipes out life on earth, it's worth to introduce widespread surveillance measures".
I'd agree that it's not a particularly common belief system and it's not like the government will be full of draconian longtermists but it has gained some traction.
https://cherwell.org/2019/04/27/mass-surveillance-could-save...
In fact, it's really questionable if the human species will still exist in its biological form. AI might already meet human intelligence within decades and then exponentially grow in might from there. Add this exponential function to a timeline of thousands of years and where do you end up?
Current humans? I don't think so.
The DCF makes sense because receiving a dollar tomorrow doesn't have the same value as receiving one today because in the second case I could invest it in the risk free asset and get more than one dollar tomorrow. This doesn't apply to human life.
I understand the feeling of the author regarding designing policy aimed at future people but the argument is just plain wrong. I'm sure there is a non fallacious argument supporting this point though. Maybe a risk / black swan based one?
Yes it does, as the certainty even for someone healthy and alive today that they will remain alive as time goes can only decrease.
Look at how the best plans go to trash for people saving when they're young, with grand plans to retire early and gorge on leisure/travel/etc, when they suddenly find out they have a terminal disease or some other health problem that at best reduce their ability to travel or enjoy their leisure (trekking in the desert on a wheelchair at 70? Or on painkillers? Might be possible, but certainly not as fun as doing that same thing at 20, if only because you'll find fewer people of your same age group willing to join in on the adventure)
Nothing short of a few radical breakthroughts can fix a strictly decreasing function.
To make it even simpler, the DCF arises because you can simply show that receiving a dollar now is preferable to a dollar in the future. Can you even claim that about human life? If you cannot even find a good explanation for your theory how can you claim you can come with an equation for it? The DCF is a very precise way to compare mesurable quantities. How would you even define the value of a human life?
You cannot randomly pick known equations and apply them to unrelated topics because there is a vague resemblance. This is supermarket philosophy. That's why it's ridiculous.
What's ridiculous to me is that you insist:
> First, it seems you didn't read the article, the comparison is not between various ages of a given person but between a bunch of human life now and a bunch of other life in the future. (...) > To make it even simpler, the DCF arises because you can simply show that receiving a dollar now is preferable to a dollar in the future. Can you even claim that about human life?
Are you mathematically guaranteed these lives in the future will exist? If not, yes you can.
> If you cannot even find a good explanation for your theory how can you claim you can come with an equation for it?
Yes, as probability theory creates strictly decreasing functions.
> The DCF is a very precise way to compare mesurable quantities. How would you even define the value of a human life?
You don't have to. With a strictly decreasing function where any f(x)>0, you can say either that lim x->+inf f(x)=0 or there's a w such that f(w)=0 - even if you don't know anymore more about f(x)
If you are saying that's it's likely that the population will increase and humanity will not destroy itself, you have a partial argument, but then there's the counter argument that technology will progress alleviating all problems (a strong claim given the past performance of the human species fixing many ailments, poverty, etc) or that humanity will evolve into something else (say, gnostic ascention, computer uploading ala matrix, cybernetics ala battlestar galactica...) and back to square one: a bunch of real people exist now, for sure.
Sure it does, since it's Effective Altruism we're talking about, so it's (i) literally about socially investing money and (ii) it's choosing between spending it on things anticipated to generate compounding returns like economic growth or carbon control or health and education (all likely to have indirect long term benefits to solving whatever humanity's future challenges are too) versus spending it on stuff which serves no purpose in the foreseeable future but is hypothetically critical to the hypothetical risks faced by hypothetical humans in the hypothesised future.
And also because even if cash couldn't be invested [risk free], humans would still apply a discount rate to the future because of uncertainty and time preference the author discusses. Needless to say, there's a lot of uncertainty about the needs of future generations, and just as it's pretty normal to care more about your own child than kids in a faraway land, it's pretty normal to consider hypothetical third millenium transhumans on Alpha Centurai essentially irrelevant to your priorities.
(I mean, trying to avoid human extinction is pretty worthwhile for its own sake anyway; invoking billions upon billions of future humans spread across the galaxy because everyone thinks your project has an astronomically small chance of actually saving humanity is just Pascal's Wager in scifi form)
We should put off problems now because they will be cheaper and easier to solve in the future because technology would have improved.
Putting resources into technology now instead of solving problems that can be put off will help to ensure that.
Which is all well and good, but not really relevant to effective altruism or utilitarian ethics. Effective altruism starts with the premise that people have inherent worth in some universal sense. If we're just thinking about people's value relative to me, then sure, humanity thousands or millions of years from now isn't worth much. I'll never know or interact with them, and nor will anyone I know or care about. But the whole appeal of utilitarian ethics is that it's not based on who I happen to like.
I actually agree with you that longtermism is problematic; I'm not fond of ethical systems that give wildly shifting directives based on guesses about the shape of the future. But that's why I'm not utilitarian. Grant all the assumptions and the logic seems pretty sound to me.
It's true that longtermists view climate change as relatively less of a threat than other existential risks. The reason is that climate change seems unlikely to literally make humanity extinct or to trap us in an unrecoverable and inescapable bad position. Plus, climate change is relatively slow moving compared to other existential risks. The book does better than I will be about to do here in explaining this
I could see a grease fire on my kitchen stove and decide to put it out even while acknowledging that a fire is relatively less important than villains abducting me and my family and torturing us to death.
I find the logic flawed as it seems to imply that extrem suffering of some is justified if it benefits many others. "You can kill a person to harvest its organs to save many" seems implicit, but I just read the summary. The book may contain much more nuance.
"Climate change is relatively unimportant" - Only to existential risks. The likelihood of humanity becoming extinct from climate change is low and it will happen slowly enough that we could probably react to it.
"Not able to predict what Coke will do in the future" - True, we can't predict what humanity will be doing in the future, but we can say that either humanity will exist or it won't. If it does exist then that is what we are concerned with. If it doesn't exist then the bad part of existential risk has already been realized.
"Discounting future value of human lives" - You can play with numbers in your DCF formula and find highly unintuitive things like it would be better to kill a billion people a billion years in the future than a million people a million years in the future. That doesn't really seem right. It's also true that the future value of humans isn't "ours" exactly in the way that the future earnings of a company might be, we aren't trying to maximize "our" value, but rather value generally - so I'm not sure discounting formulas like this are relevant.
But it sounds like a great excuse to ignore almost all of the world's current or "in our lifetimes" problems, focus on a few very theoretical problems - of serious interest to only a few SciFi fans and longtermists - and proclaim it a virtue to do that. (While probably doing nothing whatever about any problem, beyond talking big and puffing themselves up.)
Like what?
Just adjusting and staying relevant in our fast-paced world requires an increasingly short term focus, if anything. Paycheck to paycheck and business quarter to business quarter. Further, many people aren't very hopeful for the future as everything seems to be breaking down.
We're also absolutely terrible at even "short term" problems. Climate change was warned about since the 70s. We had decades to curb it but did nothing. We can foresee population booms as well as aging decades before they happen but again do absolutely nothing.
We're a short term species. And getting even more short term. Instant gratification and all that.
Personally, I don't care at all about the prospect of people populating the galaxy. Imagine the mess they make. It wouldn't be long before the sky looked like an ad.
Am I meant to believe that the average human currently alive values an intergalactic future at thousands of times the value derived from their own existence? For the vast majority of people, that factor is probably in the ones and twos at most. I can think of many people I know personally for whom that factor is marginally negative.
One of the problems seems to be that even if you agree that the utility of unlikely but great outcomes for humanity in the far future is something like (very small probability of perfect scenario) x (very small discount factor) x (very large utility), people might disagree fundamentally about the order of magnitude of all three quantities, which will lead to vastly different judgements? I'm curious whether this sort of argument is too naive, too, and whether there are convincing arguments why this ought not to be the case that are prevalent in the longtermist community?
I don't think the argument is to not do anything about climate change. It is to continue to work with urgency to tackle climate change _and_ to do something appropriate about risks that could wipe out civilisation completely, where we have reason to believe our actions could to some degree mitigate the risks.
What if ‘meh’ is the great filter?
about "the future is distant and we can't know anything about it", see e.g. this[2] post on the effective altruism forum.
> Possible misconception: “Trying to influence the far future is pointless because it is impossible to forecast that far.”
> My response: “Considering far future effects doesn’t necessarily require predicting what will happen in the far future.”
> [...]
> Nuclear war could feasibly happen tomorrow. Climate change is an ongoing phenomenon and catastrophic climate change could happen within decades. In The Precipice, Toby Ord places the probability of an existential catastrophe occuring within the next 100 years at 1 in 6, which is concerningly high.
> Ord is not in the business of forecasting events beyond a 100-year time horizon, nor does he have to be. These existential threats affect the far future on account of the persistence of their effects if they occur, but not on account of the fact that they might happen in the far future. Therefore whilst it is true that a claim has to be made about the far future, namely that we are unlikely to ever properly recover from existential catastrophes, this claim seems less strong than a claim that some particular event will happen in the far future.
i recommend op engage with some longer longtermist writings and try to understand the view from the inside. another poster mentioned toby ord's the precipice, but there is also plenty of material online. for example, i recommend this interview with carl shulman about extinction risk.[3]
[1] https://slatestarcodex.com/2017/04/07/yes-we-have-noticed-th...
[2] https://forum.effectivealtruism.org/posts/ocmEFL2uDSMzvwL8P/...
[3] https://80000hours.org/podcast/episodes/carl-shulman-common-...