Through the doomsday argument people try to calculate the number of humans that will ever live based only on the number of humans that have lived until now. To me this arguments seems very flawed, but I might be misunderstanding it.

Let's say there is a fixed number of all members of the human species in the past and future (H). We want to estimate this number. And let's say we have a simple counter that represents the number of humans born yet (n). The counter starts from 1 with the first human. Every time the counter increases to include a new human (n+1), there is a slight chance (c) that this one human being comes up with the doomsday argument. This one human being that just came up with the DA will look at the current value of the counter (a), which represents their number in the succession of all humans. Using this number a, the human will try to estimate H, while the counter keeps counting.

I don't see how the current value a in any way depends on H in this case. It only depends on the chance c (which depends on how easy it is to come up with the doomsday argument for a human). Am I missing something?

Edit: a depends on H in that a <= H

  • It isn't very clear to me how the possibility 'c' directly affects the counter 'n'.
    – user31740
    Commented Jun 18, 2019 at 14:44
  • 1
    Since humans talk to each other once one of them came up with the doomsday argument the rest can just hear about it, they need not come up with it themselves. It is unclear why the chance (c) has anything whatsoever to do with the estimates.
    – Conifold
    Commented Jun 18, 2019 at 16:04
  • c is the probability of one human coming up with the doomsday argument (and then telling everyone). If c is 0.1, it might take about 10 people until the first one comes up with the argument. If c is 0.01, it might take 100 people. The possibility does not effect the counter, but the first person to come up with the argument will look at the counter and base the estimation on the current counter value.
    – alexander
    Commented Jun 19, 2019 at 11:16

1 Answer 1


I agree it's a flawed argument. The base concept comes from Bayes' theorem, to the effect that if you randomly choose a member of a set, the characteristics of that member are most likely to be the most common characteristics found in the set as a whole. In other words, if you have a bag full of red and blue balls, and the ball you randomly pick out is red, it's (non-conclusive, statistical) evidence towards the conclusion that there are more red balls than blue balls in the bag. That's fairly uncontroversial, although it's important to note that it's a statistical probability, not an entailment. Your random draw could always be that one-in-a-million long shot, although probably only once in a million tries. (More about this here.)

A more controversial application of this theorem is the idea that you can use yourself in place of the randomly chosen ball and extrapolate info about all humanity from your own personal traits. Personally, I think the argument breaks down right here. One of the major, obvious difficulties, is that you aren't necessarily a random, representative member of humanity. Most people in the world right now live in poverty. Most people in history didn't have access to the internet. The fact that you've even heard of the doomsday theory and are considering it may already entail that you are a special person with non-representative traits.

Another objection against the doomsday theory is that it's simply mathematically unsound. It stems from the idea that there are more people alive now than have ever been alive at one time in the past. Imagine a graph of people throughout all time. A basic assumption of the theory is that population rises to a maximum, and then doomsday, the graph cuts off. All other things being equal, you (supposedly!) as a random datapoint, should statistically be found right at the righthand edge of the graph, where the population is highest.

But that misunderstands the math. This is the moment in time when the population is the highest it has ever been. But there are more people who have lived and died than are alive right now. So a random datapoint is likely to be on the righthand side of the graph but NOT necessarily all the way to the edge. We also have no way of knowing what the full shape of the graph would be --why must it come to a peak and then cut off? Couldn't it gradually diminish again? In that case the most likely place to find a random datapoint would definitely not be the end.

With all that said, I think your objection fails against this argument, because the putative relationship isn't causal at all, it's purely statistical. And the piece that you're missing is that there isn't a constant number of people born each year --there are more people born each year. So the probability that the event "someone comes up with the DA this year" happens is bigger each year --all other things being equal, and as long as population continues to increase.

  • Thanks for the explanation. But I'm still not convinced that my "simulation" of the argument is wrong. It's just another perspective. "you aren't necessarily a random, representative member of humanity" is exactly what my simulation shows. To think about the DA and then consider yourself as "drawn" requires you to discover the DA first. Cleopatra has 0 chance of being "randomly drawn", because the DA was not discovered back then. Your "drawing" mostly depends on the time the DA was discovered. The chance of discovery is not affected by H.
    – alexander
    Commented Jun 19, 2019 at 11:49
  • @alexander - I have edited to address Commented Jun 19, 2019 at 14:57

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .