31 ms·

So, I read this paper and was pretty unconvinced. I think the premise of using a Monte Carlo analysis is flawed because of course multiplying together wide ~iid

by frisco 8y ago

So, I read this paper and was pretty unconvinced. I think the premise of using a Monte Carlo analysis is flawed because of course multiplying together wide ~iid distributions that have big near-zero components asymptotes to zero; unless I'm misunderstanding something, this is essentially the central limit theorem. (All of the parameter samples below 1e0 drag the total outcome towards zero since probabilities are bounded >= 0) Thus, the real information is still strongly dominated by beliefs about the width and centering of the distributions, which we still don't know, and as long as enough of the parameters have high uncertainty with some potentially having near-zero values, the result will always tend towards zero. I think what this analysis tell us is that if you don't think it's possible to get better, more specific values for the parameters, the Drake Equation is just a bad tool. Also, the Drake Equation doesn't describe the number of alien civilizations, it describes the likely number of civilizations. Thus, extremely pessimistic estimates have to somehow correct for the fact that we know N is not zero, since we are here. That seems like a highly informative update that is not addressed in an interesting way.

mcbutterbunz 8y ago
I agree. Just because there might be a large number of stars without a habitable planet does not decrease the number of alien civilizations; it only reduces the likelihood that a given solar system would have such a civilization. Or to put it in other words, just because the haystack is large doesn't mean there is no needle present.
aqsalose 8y ago
> Thus, the real information is still strongly dominated by beliefs about the width and centering of the distributions, which we still don't know, and as long as enough of the parameters have high uncertainty with some potentially having near-zero values, the result will always tend towards zero. You state this like it is a problem. I think it's more like a feature? My take on the main point of the paper: it is commonplace to derive probabilities for being alone in the universe by plugging in point estimates into Drake Equation, but one should take into account that, in fact, we have lots uncertainty in our estimates for the parameters, and without quantifying that uncertainty, the Drake's Equation indeed will produce biased estimates. And if we quantify the uncertainty, we end up with wide distributions (that have some of their mass near zero). edit. Your point about N = 1 is interesting, though. edit2. Concerning N = 1: They don't elaborate on it in writing, but if I'm reading the sec 4.0.2 correctly, they do account for our existence in their spatial Poisson update: P(no detection | N, d) = 1 - exp(-(4/3) pi d^3 lambda) = 1 - P("number of civilizations in sphere of radius d when rate is lambda" = 0) = P("number of civilizations in sphere ..." > 0). edit3. wait, that does not make sense, does it. shouldn't P( no detection | N, d) be P("number of civilizations in sphere when rate is lambda" = 1) = (4/3) pi d^3 lambda exp(-(4/3) pi d^3 lambda), as per spatial Poisson process P(K = 1). now I'm just confused.
cottsak 8y ago
Why do we keep hanging around the Drake Equation then? or: What is a better tool?
bambax 8y ago
Yes, the Drake equation seems quite pointless because multiplying many unknowns by one another doesn't reduce uncertainty, it increases it. I don’t know anything about the history of the Drake equation, but it looks possible it was devised to demonstrate the unknownability of the result, rather than a tool to help find it. Also, the article seems to use "our galaxy" and "the universe" as interchangeable terms, which of course they are not. I don’t see how the Fermi paradox is a paradox: given the distances involved and what we know about the speed of light, an alien civilization would have had to develop a very long time ago in order to send signals that we would be able to detect today. Just because, maybe, no civilization was able to do that a billion years ago doesn’t mean nothing happened since (as you say, at least we happened). We just don’t know.
- raducu 8y ago
  Actually Fermi's paradox makes total sense to me: - Presume that even with something like travel at 1% o speed of light, the Milky Way is colonizable in a couple of million years. - Presume that we are not special at all and there exist more technological advanced civilizations that predate us by at least a couple million years - Presume such a large scale colonization would be easy to observe If you accept the 3 presuppositions, the only logical conclusion is that there are no other technologically advanced species in the Milky Way or that we're the first(very unlikely). Or you can choose to attack any of the presuppositions, but that's also very hard in my humble oppinion.
  - red75prime 8y ago
    P(at least one technologically advanced civilization observes itself | there exists at least one technologically advanced civilization) = 1 We will observe ourselves, no matter how it is unlikely for us to exist. We may not be special, but highly improbable and therefore the first and the only. But such explanation isn't that interesting.
    - ubernostrum 8y ago
      You're just stating an anthropic-principle argument here, and unfortunately anthropic arguments tend conceal a fallacy of equivocation. You can look up some of William Lane Craig's work for a formal logical analysis showing exactly where it goes wrong, but the traditional plain-language analogy is: Imagine you're sitting in a prison cell, and you've been told you're going to be executed by a firing squad of a dozen expert marksmen, who will shoot you from a distance of only a few feet away. You're led out of your cell, tied to a stake, the marksmen line up, they raise their rifles and aim at you, and you hear the mass bang of their simultaneous firing. And... you discover you are unharmed. All twelve expert marksmen completely missed you, at close range! The anthropic argument says you should be completely unsurprised; after all, the only way for you to be observing the marksmen now is if they all missed, so the probability of them having missed, given that you are observing them, is 1. We should not be surprised by something which has probability 1. While this is trivially true, it is also utterly unconvincing and utterly pointless. You are perfectly justified in being surprised that all the marksmen missed, and here is the equivocation: the anthropic argument tells you not to be surprised, but the thing it tells you not to be surprised at is not the thing that was actually surprising. Your prior for all the marksmen missing and leaving you completely unharmed should have been very, very low, and no amount of ex-post-facto wrangling can change that. Or, put more simply, anthropic arguments can be used to launder any past event to a probability of 1; after all, the probability of event X having happened, given that you observed event X, is always trivially 1. But it tells you nothing useful whatsoever.
gpsx 8y ago
The fact that there is at least one intelligent civilization in "the" universe is not necessarily that informative. This depends on the number of universes. If there is only one universe, then it is informative. If there are many universes, the probability of life in one can still be very small but it can still arise, in some universes. The question then is if there is life on our universe.
- millstone 8y ago
  Talking about the number of universes feels like talking about the number of wheels on a unicycle. The count is one. Either the "multiple universes" are causally disconnected and therefore irrelevant, or are actually connected and therefore part of the same universe.
  - gpsx 8y ago
    Yes, these are not causally connected, so we can call the different universes. I interpreted the original post to be saying that if life is very improbable then it is unlikely we would be here. And If there were only one universe then yes, it would be unlikely we are here. But if there are many universes, this could support a vanishingly small probability of life with most universes being empty, but a small fraction having life. So these different universes are meaningful only to the extent you are asking philosophical questions like this, but not to most practical considerations.
  - vidarh 8y ago
    It's a argument about how close to 0 the probability of life might be, given that we know of one instance of life. It is tempting, because we exist, to assume the probability of life arising can't be that low. But if life is so rare that the odds of one arising in any given (causally disconnected) universe means life almost never arises, and more than one civilisation is unimaginable more rare than that again, if you imagine an infinite number of universes there will still be an infinite number of civilisations. But being alone might be the most likely scenario if life is present, and nobody would be present in the empty universes to take them into account. In which case our existence might easily trick us into over-estimating the odds of life arising. Framed in a less depressing way: It might be we're not seeing the aliens because something is preventing us from observing them rather than because they don't exist.
szemet 8y ago
have to somehow correct for the fact that we know N is not zero, since we are here The anthropic principle is the correction. Observing ourselves is a prerequisite of analysis - independently of probabilities which can be any low! So it is not an "independent" observation. Otherwise we could easily say from armchair, that: 1. Probability of civilization in a certain place is extremely low 2. Universe is huge, with plenty of places 3. https://en.wikipedia.org/wiki/Poisson_limit_theorem https://en.wikipedia.org/wiki/Poisson_limit_theorem 4. Expected number of civilizations in universe (without noticing anthropic principle): 1 (at least!) 5. So take Poisson dist. with lambda=1, as a conservative estimate 6. Probability of more than 1 civilization, if we know that there is at least 1 (again without noticing anthropic principle): P(X>1 | X>0) = P(X>1)/P(X>0) /Bayes/ = 0.718 So with around 72% probability there is a another civilization. ;)
- cristianpascu 8y ago
  If you assume that a storm put together a Boeing 707 randomly, and there are many storms in the Universe, then yes there will be a good chance for there being Boeings all over. Not to creatroll the discussion, but the probability of occurrence for each individual independent event that led up to what is life on Earth, in each branch of the tree of life, on each evolutionary path, is so low that number of planets in the Universe multiplied by all the milliseconds in the history of the Universe will not get very far toward reasonable chances. Take, for instance, a protein of 100 aminoacids. 23^100 possibilities, to put it naively, yields a quite low probability. And there a many types of proteins needed for life. IF one knows of any study that touches seriously on this topic, I'd happily want to know and be corrected!
  - narag 8y ago
    I can't cite you a study, but every time I've read about life appearance, the authors believed that the primitive forms were much simpler. Once the current more effective forms evolved, the original ones disappeared or got assimilated.
  - dogma1138 8y ago
    Well the question is just how physics plays into this. Proteins are not random because we have laws of physics that govern their structure and even if they were then once you have a protein which is self replicating and it’s better at doing that job in its environment that’s all you need, “natural selection” does not need life it just needs a process that is iterative, repeatable and is dependent on its environment. Overall life might be an inevitable outcome of our universe as our universe is predisposed to create carbon and life is the most efficient process of hydrating carbon we currently know off if increasing entropy is a natural lawful state of the universe it will be predisposed for creating life to serve that purpose. That said it’s unclesr if complex and not to mention intelligent life is a predisposition or an anomaly as out of the nearly 4 billion years life existing on Earth and now it seems it started only a few 100Ms of years after the early bombardment period technological intelligence evolved so late into the game that it might have been a fluke. Other types of intelligence exist we have pretty darn smart animals but despite many of them predating mankind and our ancestors none of them has ever evolved further.
jcims 8y ago
The Drake Equation actually describes the likely number of detectable alien civilizations.
zerostar07 8y ago
You probably mean the multiplicative analogue of the central limit theorem
DennisP 8y ago
Seems to me the point is mainly that the Fermi paradox is no paradox at all, since it's entirely plausible that we're alone in the universe. This doesn't mean we are alone, given the uncertainties. It just means that if we don't see anyone, we don't have to come up with all sorts of crazy scenarios to resolve the paradox, as people often do. (If you find it implausible that we're the only ones, note we could also be one of just a few, separated by billions of light years, in which case we still wouldn't see anyone.)
- Semirhage 8y ago
  If advanced technological civilizations are on average, short-lived, then the separation could be orders of magnitude smaller and your point would still stand. All of human evolution is in the millions of years, modern civilization in the thousands, technology in the hundreds.
  - lisper 8y ago
    This seems to me to be the most likely scenario given what we see here on earth. Our civilization figured out how to remove limits on its growth a lot faster than it is figuring out that we need to self-impose limits on our growth in order to avoid a catastrophic failure of our ecosystem. I see no reason to expect anything different. Evolution does not plan ahead.
    - andybak 8y ago
      > faster than it is figuring out that we need to self-impose limits on our growth in order to avoid a catastrophic failure of our ecosystem We "figured out" unconstrained growth could be catastrophic a couple of hundred years ago at the very least. And in fact we guessed prematurely as technology extended the date of the catastrophe multiple times. So we then bounced into a phase of overconfidence whereby some of us decided that technology could overcome all problems. (Arguably we are still in this phase - or at least some of us hope this turns out to be true as we've lost hope in the possibility of political solutions to problems caused by growth)
TTPrograms 8y ago
How is the premise of monte carlo analysis flawed because "of course multiplying together..."? Monte carlo is just a way to estimate the total distribution of the resulting drake probability given uncertainties about the other values. The resulting distribution they depict is just the distribution of outcomes according to our uncertainties. It's precisely bayesian using the empirical estimates according to the literature.

by frisco 8y ago

mcbutterbunz 8y ago

I agree. Just because there might be a large number of stars without a habitable planet does not decrease the number of alien civilizations; it only reduces the likelihood that a given solar system would have such a civilization. Or to put it in other words, just because the haystack is large doesn't mean there is no needle present.

aqsalose 8y ago

> Thus, the real information is still strongly dominated by beliefs about the width and centering of the distributions, which we still don't know, and as long as enough of the parameters have high uncertainty with some potentially having near-zero values, the result will always tend towards zero. You state this like it is a problem. I think it's more like a feature? My take on the main point of the paper: it is commonplace to derive probabilities for being alone in the universe by plugging in point estimates into Drake Equation, but one should take into account that, in fact, we have lots uncertainty in our estimates for the parameters, and without quantifying that uncertainty, the Drake's Equation indeed will produce biased estimates. And if we quantify the uncertainty, we end up with wide distributions (that have some of their mass near zero). edit. Your point about N = 1 is interesting, though. edit2. Concerning N = 1: They don't elaborate on it in writing, but if I'm reading the sec 4.0.2 correctly, they do account for our existence in their spatial Poisson update: P(no detection | N, d) = 1 - exp(-(4/3) pi d^3 lambda) = 1 - P("number of civilizations in sphere of radius d when rate is lambda" = 0) = P("number of civilizations in sphere ..." > 0). edit3. wait, that does not make sense, does it. shouldn't P( no detection | N, d) be P("number of civilizations in sphere when rate is lambda" = 1) = (4/3) pi d^3 lambda exp(-(4/3) pi d^3 lambda), as per spatial Poisson process P(K = 1). now I'm just confused.

cottsak 8y ago

Why do we keep hanging around the Drake Equation then? or: What is a better tool?

bambax 8y ago

Yes, the Drake equation seems quite pointless because multiplying many unknowns by one another doesn't reduce uncertainty, it increases it. I don’t know anything about the history of the Drake equation, but it looks possible it was devised to demonstrate the unknownability of the result, rather than a tool to help find it. Also, the article seems to use "our galaxy" and "the universe" as interchangeable terms, which of course they are not. I don’t see how the Fermi paradox is a paradox: given the distances involved and what we know about the speed of light, an alien civilization would have had to develop a very long time ago in order to send signals that we would be able to detect today. Just because, maybe, no civilization was able to do that a billion years ago doesn’t mean nothing happened since (as you say, at least we happened). We just don’t know.

raducu 8y ago
Actually Fermi's paradox makes total sense to me: - Presume that even with something like travel at 1% o speed of light, the Milky Way is colonizable in a couple of million years. - Presume that we are not special at all and there exist more technological advanced civilizations that predate us by at least a couple million years - Presume such a large scale colonization would be easy to observe If you accept the 3 presuppositions, the only logical conclusion is that there are no other technologically advanced species in the Milky Way or that we're the first(very unlikely). Or you can choose to attack any of the presuppositions, but that's also very hard in my humble oppinion.
- red75prime 8y ago
  P(at least one technologically advanced civilization observes itself | there exists at least one technologically advanced civilization) = 1 We will observe ourselves, no matter how it is unlikely for us to exist. We may not be special, but highly improbable and therefore the first and the only. But such explanation isn't that interesting.
  - ubernostrum 8y ago
    You're just stating an anthropic-principle argument here, and unfortunately anthropic arguments tend conceal a fallacy of equivocation. You can look up some of William Lane Craig's work for a formal logical analysis showing exactly where it goes wrong, but the traditional plain-language analogy is: Imagine you're sitting in a prison cell, and you've been told you're going to be executed by a firing squad of a dozen expert marksmen, who will shoot you from a distance of only a few feet away. You're led out of your cell, tied to a stake, the marksmen line up, they raise their rifles and aim at you, and you hear the mass bang of their simultaneous firing. And... you discover you are unharmed. All twelve expert marksmen completely missed you, at close range! The anthropic argument says you should be completely unsurprised; after all, the only way for you to be observing the marksmen now is if they all missed, so the probability of them having missed, given that you are observing them, is 1. We should not be surprised by something which has probability 1. While this is trivially true, it is also utterly unconvincing and utterly pointless. You are perfectly justified in being surprised that all the marksmen missed, and here is the equivocation: the anthropic argument tells you not to be surprised, but the thing it tells you not to be surprised at is not the thing that was actually surprising. Your prior for all the marksmen missing and leaving you completely unharmed should have been very, very low, and no amount of ex-post-facto wrangling can change that. Or, put more simply, anthropic arguments can be used to launder any past event to a probability of 1; after all, the probability of event X having happened, given that you observed event X, is always trivially 1. But it tells you nothing useful whatsoever.

gpsx 8y ago

The fact that there is at least one intelligent civilization in "the" universe is not necessarily that informative. This depends on the number of universes. If there is only one universe, then it is informative. If there are many universes, the probability of life in one can still be very small but it can still arise, in some universes. The question then is if there is life on our universe.

millstone 8y ago
Talking about the number of universes feels like talking about the number of wheels on a unicycle. The count is one. Either the "multiple universes" are causally disconnected and therefore irrelevant, or are actually connected and therefore part of the same universe.
- gpsx 8y ago
  Yes, these are not causally connected, so we can call the different universes. I interpreted the original post to be saying that if life is very improbable then it is unlikely we would be here. And If there were only one universe then yes, it would be unlikely we are here. But if there are many universes, this could support a vanishingly small probability of life with most universes being empty, but a small fraction having life. So these different universes are meaningful only to the extent you are asking philosophical questions like this, but not to most practical considerations.
- vidarh 8y ago
  It's a argument about how close to 0 the probability of life might be, given that we know of one instance of life. It is tempting, because we exist, to assume the probability of life arising can't be that low. But if life is so rare that the odds of one arising in any given (causally disconnected) universe means life almost never arises, and more than one civilisation is unimaginable more rare than that again, if you imagine an infinite number of universes there will still be an infinite number of civilisations. But being alone might be the most likely scenario if life is present, and nobody would be present in the empty universes to take them into account. In which case our existence might easily trick us into over-estimating the odds of life arising. Framed in a less depressing way: It might be we're not seeing the aliens because something is preventing us from observing them rather than because they don't exist.

szemet 8y ago

have to somehow correct for the fact that we know N is not zero, since we are here The anthropic principle is the correction. Observing ourselves is a prerequisite of analysis - independently of probabilities which can be any low! So it is not an "independent" observation. Otherwise we could easily say from armchair, that: 1. Probability of civilization in a certain place is extremely low 2. Universe is huge, with plenty of places 3. https://en.wikipedia.org/wiki/Poisson_limit_theorem https://en.wikipedia.org/wiki/Poisson_limit_theorem 4. Expected number of civilizations in universe (without noticing anthropic principle): 1 (at least!) 5. So take Poisson dist. with lambda=1, as a conservative estimate 6. Probability of more than 1 civilization, if we know that there is at least 1 (again without noticing anthropic principle): P(X>1 | X>0) = P(X>1)/P(X>0) /Bayes/ = 0.718 So with around 72% probability there is a another civilization. ;)

cristianpascu 8y ago
If you assume that a storm put together a Boeing 707 randomly, and there are many storms in the Universe, then yes there will be a good chance for there being Boeings all over. Not to creatroll the discussion, but the probability of occurrence for each individual independent event that led up to what is life on Earth, in each branch of the tree of life, on each evolutionary path, is so low that number of planets in the Universe multiplied by all the milliseconds in the history of the Universe will not get very far toward reasonable chances. Take, for instance, a protein of 100 aminoacids. 23^100 possibilities, to put it naively, yields a quite low probability. And there a many types of proteins needed for life. IF one knows of any study that touches seriously on this topic, I'd happily want to know and be corrected!
- narag 8y ago
  I can't cite you a study, but every time I've read about life appearance, the authors believed that the primitive forms were much simpler. Once the current more effective forms evolved, the original ones disappeared or got assimilated.
- dogma1138 8y ago
  Well the question is just how physics plays into this. Proteins are not random because we have laws of physics that govern their structure and even if they were then once you have a protein which is self replicating and it’s better at doing that job in its environment that’s all you need, “natural selection” does not need life it just needs a process that is iterative, repeatable and is dependent on its environment. Overall life might be an inevitable outcome of our universe as our universe is predisposed to create carbon and life is the most efficient process of hydrating carbon we currently know off if increasing entropy is a natural lawful state of the universe it will be predisposed for creating life to serve that purpose. That said it’s unclesr if complex and not to mention intelligent life is a predisposition or an anomaly as out of the nearly 4 billion years life existing on Earth and now it seems it started only a few 100Ms of years after the early bombardment period technological intelligence evolved so late into the game that it might have been a fluke. Other types of intelligence exist we have pretty darn smart animals but despite many of them predating mankind and our ancestors none of them has ever evolved further.

jcims 8y ago

The Drake Equation actually describes the likely number of detectable alien civilizations.

zerostar07 8y ago

You probably mean the multiplicative analogue of the central limit theorem

DennisP 8y ago

Seems to me the point is mainly that the Fermi paradox is no paradox at all, since it's entirely plausible that we're alone in the universe. This doesn't mean we are alone, given the uncertainties. It just means that if we don't see anyone, we don't have to come up with all sorts of crazy scenarios to resolve the paradox, as people often do. (If you find it implausible that we're the only ones, note we could also be one of just a few, separated by billions of light years, in which case we still wouldn't see anyone.)

Semirhage 8y ago
If advanced technological civilizations are on average, short-lived, then the separation could be orders of magnitude smaller and your point would still stand. All of human evolution is in the millions of years, modern civilization in the thousands, technology in the hundreds.
- lisper 8y ago
  This seems to me to be the most likely scenario given what we see here on earth. Our civilization figured out how to remove limits on its growth a lot faster than it is figuring out that we need to self-impose limits on our growth in order to avoid a catastrophic failure of our ecosystem. I see no reason to expect anything different. Evolution does not plan ahead.
  - andybak 8y ago
    > faster than it is figuring out that we need to self-impose limits on our growth in order to avoid a catastrophic failure of our ecosystem We "figured out" unconstrained growth could be catastrophic a couple of hundred years ago at the very least. And in fact we guessed prematurely as technology extended the date of the catastrophe multiple times. So we then bounced into a phase of overconfidence whereby some of us decided that technology could overcome all problems. (Arguably we are still in this phase - or at least some of us hope this turns out to be true as we've lost hope in the possibility of political solutions to problems caused by growth)

TTPrograms 8y ago

How is the premise of monte carlo analysis flawed because "of course multiplying together..."? Monte carlo is just a way to estimate the total distribution of the resulting drake probability given uncertainties about the other values. The resulting distribution they depict is just the distribution of outcomes according to our uncertainties. It's precisely bayesian using the empirical estimates according to the literature.