Existential Risk and Pascal’s Wager

This is a reply to Professor Christian Munthe’s question: Why Aren’t Existential Risk / Ultimate Harm Argument Advocates All Attending Mass?

When assessing risks from a personal point of view, there is a pretty clear limit as to how bad or good things can get (as long as you exclude the possibility of infinite amounts). You may die an agonizingly slow death. While hardly pleasant to think about, it’s not very difficult to imagine and compare to other outcomes, such as dying a quick death or living comfortably. Considering risks from an impersonal point of view is much more difficult. When doing this you need to consider (among other things) whether or not people who don’t exist, but are merely potential, matter. If “they” do, as many philosophers tend to believe, you have a real quandary.

There are some possible futures where lots of people exist, and have at least pretty ok lives. Such futures are, if you believe that value aggregates (such that ”more is more”, rather than less) and is time-neutral, very desirable (from an impersonal point of view). Consequently, an event that makes such a future impossible, such as an existential catastrophe, is very, very undesirable. So far so good.

Now at this point we may realize that we may be stumbling into Pascal’s wager. An event, no matter how unlikely, is a risk worth taking seriously if the consequences are bad enough. By “taking seriously” I mean “worth spending resources in making either less likely or less bad”. However, since an existential catastrophe could (with a non-zero probability) be caused by almost anything, we face a situation of total paralysis. Any technology could be the one causing disaster, or preventing one.

As in Pascal’s case, believing in the Christian God may save us from eternal torment. But it could be just as likely that this belief condemns us to hell.

What to do? One thing would be to ignore the whole problem and focus on more pressing problems, such as fighting poverty. This is generally the answer to Pascal. We have no idea as to how to find out (in a non-biased way) which God exists and is likely to torment us if we fail to adhere to its commandments. While this answer is sometimes appealing, this option should only be considered if we honestly do not believe that there is anything that we could do to reduce the risks.

I don’t believe that this is the case with regards to  some existential risks. While existential catastrophes are unlikely and uncertain, some are hardly as unlikely as being tormented by the Spaghetti Monster. And they are also possible to study to some degree, while not necessarily with the same methods we use to study other risks.

If we would like to know the probability of suffering a car accident, we may consult the frequency of such events in the relevant demography and arrive at a decent approximation. This is not possible for existential risks, since we have never suffered an existential catastrophe. Moreover, this fortunate lack of experience cannot tell us anything about the probability of such events, since it is true for every observer that they have the same (lack of) experience, regardless of how probable such events are. This is sometimes called an “observation selection effect”.

So what do we do? As professor Sven-Ove Hansson has argued, it is time we return to philosophical methods of investigating the concepts of “risk” and “safety”, or at least that philosophers contribute to this discussion.

One such contribution is Hansson’s ideas on “mere possibility- arguments”, where a multitude of speculative scenarios are devised and eliminated from consideration through applying a number of heuristics. When such exercise is over, we may find that some previously underestimated risks are well worth taking seriously.

For example, some contemporary economists (Nordhaus, Tol) believe that the claim that current projections of greenhouse gases may cause a catastrophe is too unlikely to consider. While these are normative claims, these economists fail to consider that climate risk is asymmetrical. This means that while there are considerable risks at one end of the probability tail (six degrees warmer) such risks are non-existent at the other end of the tail (a slight temperature increase).

This means that any effort that limits the amount of greenhouse gases in the atmosphere can only reduce, and never increase, climate risk. This makes climate risk different from other plausible sources of risk. For example, consider research in microbiology. Such research could, in theory, be used to produce a powerful and cheap weapon of mass-destruction. But it could also be used to protect us against such a weapon, or some naturally occurring pathogen. In this regard, biotechnological risk is, contrary to climate change, symmetrical. We don’t know enough about the risks and possibilities to be justified in promptly stopping all research in this area.

We do have a reason though, if you believe (contrary to some economists) that risk should not be discounted, or that the consequences of a warming of six degrees might be much worse than a 30% decline in GDP (Nordhaus), to reduce the risk of a six-degree scenario even at a considerable cost.

In summary, I have argued that we indeed face a Pascalian dilemma with regards to some risks, and cannot do anything else than suspend judgement and, if possible, gather more information. However, other risks are more likely to be understood and prevented. If this is the case, it is reasonable that, among other things that a philosopher could study, this topic is well worth spending your time on.

I have written more extensively on this topic here and here.

Publicerad onsdag, februari 4th, 2015 i etik, filosofi, Karim Jebari.

3 kommentarer

  1. Hey, that’s a clever way of thinking about it.

  2. This video is a bad representation for you and your company. The audio quality if poor and lighting is really bad! Get it together. This should be the easiest part of it.

Vad tycker du?