I saw it as more of a warning about the limits of maps—when something happens that you think is impossible, then it is time to update your map, and not rail against the territory for failing to match it.
(Of course, it is possible that you have been fooled, somehow, into thinking that something has happened which has, in actual fact, not happened. This possibility should be considered and appropriately weighted (given whatever evidence you have of the thing actually happening) against the possibility that the map is simply wrong.)
If you were to tell me, for example, that you had seen a man flying through the air like Superman, then I think I could reasonably call that “impossible” and conclude that you were lying. (If I happened to be in Metropolis at the time, then I might soon be proven wrong—nonetheless, the conclusion that you are lying is significantly more probable than the conclusion that someone has suddenly developed the power of flight).
On the other hand… if you were to hold an object, and then let go, and that object were to fall up instead of down, then calling that “impossible” would be useless; I have seen the object fall up, I can see it there on the roof, I can walk under it. (And it is, indeed, not impossible; the object could be a helium balloon, or you might have concealed a powerful magnet in the roof and used a metal object).
...hmmm. I think the difference here is that in the first case, the thing has not clearly happened; I merely have an eyewitness report, which is easily forged, to say that it took place. In the second case, I have far more data to show that the object really did fall upwards, and I can even (perhaps with the aid of a ladder) retrieve the object and drop it myself, confirming that it continues to fall upwards; it has clearly happened, calling it “impossible” is indeed futile, and the only question is how.
No observation is false. Any explanation for a given observation may, with finite probability, be false; no matter how obvious and inarguable it may seem.
This is one of those things that seems like it ought to be true, if only humans weren’t so human. I’ve seen enough attempted bug reports based on events which—upon going through the logs—never actually happened, to disabuse me of the notion. Certainly some class of “false observations” might be better thought of as “false explanations”, but sometimes people are just plain wrong about what they saw.
Your observation of the reading on the scale is true, of course. Your observation that the weight is 51 grams is false.
“This weight masses 51 grams” is not an observation, it’s a theory attempting to explain an observation. It just seems so immediate, so obvious and inarguable, that it feels like an observation.
I think this “rabbit hole” is basically reality. In other words “there is a physical world which we see and hear etc” is a theory which is extremely well supported by our observations. Berkeley’s explanation that there is no physical world, but God exists and is directly causing all of our sensations is an alternate theory, although a rather unlikely one.
Berkeley’s explanation that there is no physical world, but God exists and is directly causing all of our sensations is an alternate theory, although a rather unlikely one.
After I wrote that comment, I realized that the only way of distinguishing that from the physical world hypothesis is by the prior. Because “there is a physical chair which is responsible for some of my experiences when I am in my room and continues to exist even at the times when I’m not experiencing it” predicts entirely the same things as “God has an idea of a chair in my room, and He causes experiences according to that idea.”
So if one is more probable than the other, it would be according to the prior. On the other hand, someone might argue that “there is a physical world that is defined by a certain mathematical theory” and “God exists and has a mental model of a world defined by that same mathematical theory, and produces experiences according to that mental model” may not even be distinct hypotheses. In other words, what exactly does it mean to say that God exists and has a mental mathematical model? And what exactly does it mean to say that a physical world exists according to a mathematical model? Someone might assert that insofar as these predict entirely the same experiences, they are not even different theories, but just different ways of describing the same thing. According to this, Berkeley’s theory would not imply that the chair in my room does not really exist, but rather that “there is a chair in my room” means exactly the same thing as “God’s mental model includes a chair in my room”. So it would still be true to say the chair exists and so on.
Not sure how one would refute that. But assuming they are two different theories, it sure seems like the physical world theory should have a higher prior.
[...] everything beyond photons striking the retina becomes a “theory”.
Isn’t that itself a theory to explain our qualia of vision? If, for example, some versions of the simulation hypothesis were true, even photons and eyes would be a false map, though a useful one.
The confusion here has nothing to do with the meaning of “false,” or the distinction between accuracy and precision.
If I’m using a known 50-g weight to calibrate a scale, and I look at the scale reading (which says “51g”), and thereby conclude that the scale is off by 1g, I don’t think you’re at all justified in concluding that I’ve observed that the weight is 51g.
I mean, I agree that if I had made such an observation, it would be a mistaken observation.
But I don’t agree that I made any such observation in the first place. For example, if you asked me after weighing the weight “What is the mass of the weight?” I would most likely answer “50g,” because being able to say that with confidence is the whole point of using standard-mass callibration weights in the first place.
I am confused. In your example what are you saying your observation is, and do you consider it true or false? Also, what do you consider “known” before the observation?
This is getting stuck in the morass of trying to distinguish between observations and interpretations. I don’t particularly want to discuss the philosophy of qualia.
My point is much simpler. It’s quite common for data points which everyone calls “observations” to be false. Trying to fix that problem is called cleaning the data and can be a huge hassle. In practical terms, if you get a database of observations you cannot assume that all of them are true.
I certainly agree that such data points can be false.
When you chose to disagree with khafra’s claim I thought you were making an actual counterassertion, rather than simply challenging their use of the label “observation” in an indirect way.
That may be, but if you label them ‘impossible’ and dismiss them, you won’t gather more evidence to prove it. And if something you consider impossible has actually happened, you’re missing an opportunity to improve your model significantly.
This is in fact what happens in-context. With a preposterously-detailed description of observable events (via magic hypnosis; I didn’t say the novel made sense), Gently concludes that something has happened which could not have happened as described, and that the only explanation which would explain the results involves time travel; the other person says that it’s impossible, to which Gently replies this.
I don’t think the quote is talking about “hypothesizing” anything; I read it more as “You have to update on evidence whether that evidence fits into your original model of the world or not”. Instead of “hypothesizing time travel when things don’t make sense”, it’d be more like a stranger appears in front of you in a flash of light with futuristic-looking technology, proves that he is genetically human, and claims to be from the future. In that case it doesn’t matter what your priors were for something like that happening; it already happened, and crying “Impossible!” is as illegal a move in Bayes as moving your king into check is in chess.
Not that such a thing is likely to happen, of course, but if it did happen, would you sit back and claim it didn’t because it “doesn’t make sense”?
Yes. And then I would go see a psychologist. Because I find it more likely that I’m losing my grip on my own sanity than that I’ve just witnessed time travel.
Alright, so you bring this alleged time traveler with you to visit two or three different psychologists, all of whom are appropriately surprised by the whole ‘time travel’ thing but agree that you seem to be perceiving and processing the facts of the situation accurately.
Furthermore you have a lot of expensive tests run on the health and functionality of your brain, and all of the results turn out within normal limits. Camera-phone videos of the initial arrival are posted to the internet and after millions of views nobody can credibly figure out how it could have been faked. To the extent that introspection provides any meaningful data, you feel fine. In short, by every available test, your sanity is either far beyond retrieval down an indistinguishably perfect fantasy hole, or completely unmarred apart from perhaps a circumstantially-normal level of existential anxiety.
Then I accept that there’s a time traveler. The evidence in this second situation is quite a bit stronger than a personal observation, and would probably be enough to convince me.
Well, the insanity defense is always a possibility, but then again, you have no proof that you’re not insane right now, either, so it seems to be a fully general counterargument that can apply at any time to any situation. Ignoring the possibility of insanity, would you see any point in refusing to update, i.e. claiming that what you just saw didn’t happen?
It’s always a possibility that I’m insane, but normally a fairly unlikely one.
The baseline hypothesis is (say) p = 0.999 that I’m sane, p = 0.0001 that I’m hallucinating. Let’s further assume that if I’m hallucinating, there’s a 2% chance that hallucination is about time travel. My prior is something like p = 0.000001 that time travel exists. If I assume those are the only two explanations of seeing a time traveler, (i.e. we’re ignoring pranks and similar), my estimate of the probability that time travel exists would shift up to about 2% instead of 0.0001% -- a huge increase. The smart money (98%) is still on me hallucinating though.
If you screen out the insanity possibility, and any other possibility that gives better than 1 in a million chances of me seeing what appears to be a time traveler with what appears to be futuristic technology, yes, the time traveler hypothesis would dominate. However, the prior for that is quite low. There’s a difference between “refusing to update” and “not updating far enough that one explanation is favored”.
If I was abducted by aliens, my first inclination would likewise be to assume that I’m going insane—this is despite the fact that nothing in the laws of physics precludes the existence of aliens. Are you saying that the average person who thinks they are abducted by aliens should trust their senses on that matter?
Ah. In that case, I think we’re basically in agreement. To clarify: I only used the time travel as an example because that was the example that VAuroch used in his/her comment. I agree that even taking into account your observation of time travel, the posterior probability for your insanity is still much larger than the posterior probability for genuine time travel. You do agree, however, that even if you conclude that you are likely insane, the probability of time travel was still updated in a positive direction, right? It seems to me that Nominull (the person to whom I was originally replying) was implying that your probability estimate shouldn’t change at all, because that’s “clearly impossible”/”fictional evidence” or something along those lines. It is that implication which I disagree with; as long as you’re not endorsing that implication, we’re in agreement. (If Nominull is reading this and feels that I am mistaken in my reading of his/her comment, then he/she should feel free to clarify his/her meaning.)
Unfortunately this still suffers from the whole “Time Traveller visits you” part of the claim—our language doesn’t handle it well. It’s a realistic claim about counterfactual response of a real brain to unrealistic stimulus.
I’ll be sure to ask you the next time I need to write an imaginary comment.
It’s not like anyone didn’t know what I meant. What do you think of the actual content? How much do you trust faul_sname’s claim that they wouldn’t trust their own senses on a time-travel-like improbability?
I’ll be sure to ask you the next time I need to write an imaginary comment.
I wasn’t the pedant. I was the tangential-pedantry analyzer. Ask Lumifer.
It’s not like anyone didn’t know what I meant. What do you think of the actual content? How much do you trust faul_sname’s claim that they wouldn’t trust their own senses on a time-travel-like improbability?
Your comment was fine. It would be true of most people, I’m not sure if Faul is one of the exceptions.
That seems like a failure of noticing confusion; some clear things are actually false.
I saw it as more of a warning about the limits of maps—when something happens that you think is impossible, then it is time to update your map, and not rail against the territory for failing to match it.
(Of course, it is possible that you have been fooled, somehow, into thinking that something has happened which has, in actual fact, not happened. This possibility should be considered and appropriately weighted (given whatever evidence you have of the thing actually happening) against the possibility that the map is simply wrong.)
If you’ve been fooled, there’s still no point to calling it impossible, given that you’re trying to find out what actually happened.
Hmmm.
If you were to tell me, for example, that you had seen a man flying through the air like Superman, then I think I could reasonably call that “impossible” and conclude that you were lying. (If I happened to be in Metropolis at the time, then I might soon be proven wrong—nonetheless, the conclusion that you are lying is significantly more probable than the conclusion that someone has suddenly developed the power of flight).
On the other hand… if you were to hold an object, and then let go, and that object were to fall up instead of down, then calling that “impossible” would be useless; I have seen the object fall up, I can see it there on the roof, I can walk under it. (And it is, indeed, not impossible; the object could be a helium balloon, or you might have concealed a powerful magnet in the roof and used a metal object).
...hmmm. I think the difference here is that in the first case, the thing has not clearly happened; I merely have an eyewitness report, which is easily forged, to say that it took place. In the second case, I have far more data to show that the object really did fall upwards, and I can even (perhaps with the aid of a ladder) retrieve the object and drop it myself, confirming that it continues to fall upwards; it has clearly happened, calling it “impossible” is indeed futile, and the only question is how.
No observation is false. Any explanation for a given observation may, with finite probability, be false; no matter how obvious and inarguable it may seem.
This is one of those things that seems like it ought to be true, if only humans weren’t so human. I’ve seen enough attempted bug reports based on events which—upon going through the logs—never actually happened, to disabuse me of the notion. Certainly some class of “false observations” might be better thought of as “false explanations”, but sometimes people are just plain wrong about what they saw.
That’s trivially not true—consider e.g. measurement error.
OK, let’s consider measurement error.
I have, let’s say, a weight.
It actually masses 50 g.
I put it on a scale and observe that the reading on the scale is “51g.”
On your account, is my observation false?
If so, does your judgment change if it’s a standard weight that I’m using to calibrate the scale?
Your observation of the reading on the scale is true, of course. Your observation that the weight is 51 grams is false.
The distinction between accuracy and precision is relevant here. I am assuming your scale is sufficiently precise.
No, it does not. I am using “false” in the sense of the map not matching the territory. A miscalibrated scale doesn’t help you with that.
“This weight masses 51 grams” is not an observation, it’s a theory attempting to explain an observation. It just seems so immediate, so obvious and inarguable, that it feels like an observation.
I feel this leads into a rabbit hole where everything beyond photons striking the retina becomes a “theory”.
I think this “rabbit hole” is basically reality. In other words “there is a physical world which we see and hear etc” is a theory which is extremely well supported by our observations. Berkeley’s explanation that there is no physical world, but God exists and is directly causing all of our sensations is an alternate theory, although a rather unlikely one.
Yeah, this is basically why probability matters.
What evidence lead you to this conclusion?
After I wrote that comment, I realized that the only way of distinguishing that from the physical world hypothesis is by the prior. Because “there is a physical chair which is responsible for some of my experiences when I am in my room and continues to exist even at the times when I’m not experiencing it” predicts entirely the same things as “God has an idea of a chair in my room, and He causes experiences according to that idea.”
So if one is more probable than the other, it would be according to the prior. On the other hand, someone might argue that “there is a physical world that is defined by a certain mathematical theory” and “God exists and has a mental model of a world defined by that same mathematical theory, and produces experiences according to that mental model” may not even be distinct hypotheses. In other words, what exactly does it mean to say that God exists and has a mental mathematical model? And what exactly does it mean to say that a physical world exists according to a mathematical model? Someone might assert that insofar as these predict entirely the same experiences, they are not even different theories, but just different ways of describing the same thing. According to this, Berkeley’s theory would not imply that the chair in my room does not really exist, but rather that “there is a chair in my room” means exactly the same thing as “God’s mental model includes a chair in my room”. So it would still be true to say the chair exists and so on.
Not sure how one would refute that. But assuming they are two different theories, it sure seems like the physical world theory should have a higher prior.
Isn’t that itself a theory to explain our qualia of vision? If, for example, some versions of the simulation hypothesis were true, even photons and eyes would be a false map, though a useful one.
Hey hallucinations are totally a thing.
The confusion here has nothing to do with the meaning of “false,” or the distinction between accuracy and precision.
If I’m using a known 50-g weight to calibrate a scale, and I look at the scale reading (which says “51g”), and thereby conclude that the scale is off by 1g, I don’t think you’re at all justified in concluding that I’ve observed that the weight is 51g.
I mean, I agree that if I had made such an observation, it would be a mistaken observation.
But I don’t agree that I made any such observation in the first place. For example, if you asked me after weighing the weight “What is the mass of the weight?” I would most likely answer “50g,” because being able to say that with confidence is the whole point of using standard-mass callibration weights in the first place.
I am confused. In your example what are you saying your observation is, and do you consider it true or false? Also, what do you consider “known” before the observation?
I observe that the reading on the scale is “51g,” as I said in the first place.
Yes. True.
All kinds of things. In the case with a standard 50g callibration weight, that includes the mass of the weight.
This is getting stuck in the morass of trying to distinguish between observations and interpretations. I don’t particularly want to discuss the philosophy of qualia.
My point is much simpler. It’s quite common for data points which everyone calls “observations” to be false. Trying to fix that problem is called cleaning the data and can be a huge hassle. In practical terms, if you get a database of observations you cannot assume that all of them are true.
I certainly agree that such data points can be false.
When you chose to disagree with khafra’s claim I thought you were making an actual counterassertion, rather than simply challenging their use of the label “observation” in an indirect way.
My apologies, and I’m happy to drop it here.
That may be, but if you label them ‘impossible’ and dismiss them, you won’t gather more evidence to prove it. And if something you consider impossible has actually happened, you’re missing an opportunity to improve your model significantly.
This is in fact what happens in-context. With a preposterously-detailed description of observable events (via magic hypnosis; I didn’t say the novel made sense), Gently concludes that something has happened which could not have happened as described, and that the only explanation which would explain the results involves time travel; the other person says that it’s impossible, to which Gently replies this.
Yeah, I feel like in real world situations, hypothesizing time travel when things don’t make sense is not likely to be epistemically successful.
Wasn’t there a proverb about generalizing from fictional evidence? Especially from fiction that intentionally doesn’t make sense?
I don’t think the quote is talking about “hypothesizing” anything; I read it more as “You have to update on evidence whether that evidence fits into your original model of the world or not”. Instead of “hypothesizing time travel when things don’t make sense”, it’d be more like a stranger appears in front of you in a flash of light with futuristic-looking technology, proves that he is genetically human, and claims to be from the future. In that case it doesn’t matter what your priors were for something like that happening; it already happened, and crying “Impossible!” is as illegal a move in Bayes as moving your king into check is in chess.
Not that such a thing is likely to happen, of course, but if it did happen, would you sit back and claim it didn’t because it “doesn’t make sense”?
Yes. And then I would go see a psychologist. Because I find it more likely that I’m losing my grip on my own sanity than that I’ve just witnessed time travel.
Alright, so you bring this alleged time traveler with you to visit two or three different psychologists, all of whom are appropriately surprised by the whole ‘time travel’ thing but agree that you seem to be perceiving and processing the facts of the situation accurately.
Furthermore you have a lot of expensive tests run on the health and functionality of your brain, and all of the results turn out within normal limits. Camera-phone videos of the initial arrival are posted to the internet and after millions of views nobody can credibly figure out how it could have been faked. To the extent that introspection provides any meaningful data, you feel fine. In short, by every available test, your sanity is either far beyond retrieval down an indistinguishably perfect fantasy hole, or completely unmarred apart from perhaps a circumstantially-normal level of existential anxiety.
Now what?
Then I accept that there’s a time traveler. The evidence in this second situation is quite a bit stronger than a personal observation, and would probably be enough to convince me.
Well, the insanity defense is always a possibility, but then again, you have no proof that you’re not insane right now, either, so it seems to be a fully general counterargument that can apply at any time to any situation. Ignoring the possibility of insanity, would you see any point in refusing to update, i.e. claiming that what you just saw didn’t happen?
It’s always a possibility that I’m insane, but normally a fairly unlikely one.
The baseline hypothesis is (say) p = 0.999 that I’m sane, p = 0.0001 that I’m hallucinating. Let’s further assume that if I’m hallucinating, there’s a 2% chance that hallucination is about time travel. My prior is something like p = 0.000001 that time travel exists. If I assume those are the only two explanations of seeing a time traveler, (i.e. we’re ignoring pranks and similar), my estimate of the probability that time travel exists would shift up to about 2% instead of 0.0001% -- a huge increase. The smart money (98%) is still on me hallucinating though.
If you screen out the insanity possibility, and any other possibility that gives better than 1 in a million chances of me seeing what appears to be a time traveler with what appears to be futuristic technology, yes, the time traveler hypothesis would dominate. However, the prior for that is quite low. There’s a difference between “refusing to update” and “not updating far enough that one explanation is favored”.
If I was abducted by aliens, my first inclination would likewise be to assume that I’m going insane—this is despite the fact that nothing in the laws of physics precludes the existence of aliens. Are you saying that the average person who thinks they are abducted by aliens should trust their senses on that matter?
Ah. In that case, I think we’re basically in agreement. To clarify: I only used the time travel as an example because that was the example that VAuroch used in his/her comment. I agree that even taking into account your observation of time travel, the posterior probability for your insanity is still much larger than the posterior probability for genuine time travel. You do agree, however, that even if you conclude that you are likely insane, the probability of time travel was still updated in a positive direction, right? It seems to me that Nominull (the person to whom I was originally replying) was implying that your probability estimate shouldn’t change at all, because that’s “clearly impossible”/”fictional evidence” or something along those lines. It is that implication which I disagree with; as long as you’re not endorsing that implication, we’re in agreement. (If Nominull is reading this and feels that I am mistaken in my reading of his/her comment, then he/she should feel free to clarify his/her meaning.)
Factually speaking, I think if you saw that happen, you would believe, regardless of your protestations now.
I don’t think it’s literally factually :-D
I think you’re right. It’s closer to, say… “serious counterfactually speaking”.
Realistically speaking?
Unfortunately this still suffers from the whole “Time Traveller visits you” part of the claim—our language doesn’t handle it well. It’s a realistic claim about counterfactual response of a real brain to unrealistic stimulus.
I’ll be sure to ask you the next time I need to write an imaginary comment.
It’s not like anyone didn’t know what I meant. What do you think of the actual content? How much do you trust faul_sname’s claim that they wouldn’t trust their own senses on a time-travel-like improbability?
Anecdotal evidence of the reaction of normal people to seeing something impossible:
http://www.bcgreen.com/comments/not_kansas.html
I wasn’t the pedant. I was the tangential-pedantry analyzer. Ask Lumifer.
Your comment was fine. It would be true of most people, I’m not sure if Faul is one of the exceptions.
Generalization from fictional evidence