Notes on not dying


Earlier I wrote about the importance for EAs of not dying, suggesting that EAs should be much less willing to risk their own life than Egoists. The case where this consideration applies robustly is when you’re considering doing some risky activity X, which gives you a pure pleasure or happiness boost, or makes your life better in some way that doesn’t affect your ability to do good.

The first case where this consideration applies in a mitigated form is where that risky activity X also benefits your productivity. For example, suppose that you have some medical condition that is interfering with your ability to work, and in order to cure it you need to undergo general anaesthetic. General anaesthetic has 10 micromorts, giving you a 1 in 100,000 chance of dying — not a sort of risk you want to take lightly. But if this condition really is affecting your productivity (and therefore your ability to do good), and you’d be more than 0.001% more productive by curing your condition, then the expected value of curing the condition is positive.

The second case where this consideration applies in a mitigated form is where that risky activity X saves you time. Let’s suppose that you can either drive or cycle to some destination 1 mile away. And let’s suppose you’re somewhere like Oxford where driving around the centre is slower than cycling. Cycling 1 mile creates 0.04 micromorts; driving one mile creates 0.005 micromorts. So, by the “don’t die” argument you might think that you should drive. But you’ve got to weigh that against the good you do in the time saved. Let’s suppose that you’d save 10 minutes by taking the bike, which you would spend working. What’s the cost of those micromorts? Let’s suppose you’ve got 50 more productive years, that you work 60 hours a week, 48 weeks a year. In which case the expected cost of cycling is (0.04/100000)*(50*60*48*60) – (0.005/100000)*(50*60*48*60) = 3 minutes.

So, to a very first approximation (and bearing in mind that these numbers are very sketchy!), it’s worth it to cycle if it gives you more than 3 extra minutes of working time. I stipulated in the above case that cycling gives you 10 extra minutes of life. So, in that above case, you should cycle.

I say ‘to a first approximation’ because you should apply a discount rate: almost certainly, you should value giving yourself 1hr in the future either more or less than you should value giving yourself 1hr now. If your discount rate is positive (and you value time in the future less than you value time now), then comparing time saved against expected life lost due to potential death will overestimate the badness of the risk of death.