I did take the survey, however I found something I was unsure of what to put down and had to type in an explanation/question about:
It was for the question: “By what year do you think the Singularity will occur? Answer such that you think there is an even chance of the Singularity falling before or after that year. If you don’t think a Singularity will ever happen, leave blank.”
If I think the singularity is slightly less than 50% likely overall, what should I have put? It seemed off to leave it blank and imply I believed “I don’t think a Singularity will ever happen” because that statement seemed to convey a great deal more certainty than 50+epsilon%. However, if I actually believed there was a less than 50% chance of it happening, I’m not going to reach an even chance of happening or not happening on any particular year.
As a side note, after taking that test, I realized that I don’t feel very confident on a substantial number of things.
I think that there need to be two separate questions here. Probability of Singularity, and year it happens if it does. For instance, I’d guess about 70% chance of a singularity at all, but if it happens, 2040 would be about my expected date. You can’t describe these two statements in just one number.
I interpreted this as “there is an even chance of the Singularity falling before or after, [assuming it does]”. That is, if you think the probability that the Singularity will happen is something low like 1%, you should answer a year such that the probability it happens by that year is 0.5%. The only way you can’t answer it is if you’re sure it won’t ever happen.
(For example, if I thought a Singularity is very [...] very hard to achieve, I might answer 5000 AD or 500000 AD, depending on how many “very” there are, even though I might put a very low probability on our civilization actually surving that long.)
Given the expected date would be skewed to infinity by a non-zero estimate of the Singularity not occurring, you can probably put your estimate of the year X so that P( S ⇐ X | C ) = 0.5, where S is the statistic “Year singularity will occur” and C is the event “Singularity will occur”.
I did take the survey, however I found something I was unsure of what to put down and had to type in an explanation/question about:
It was for the question: “By what year do you think the Singularity will occur? Answer such that you think there is an even chance of the Singularity falling before or after that year. If you don’t think a Singularity will ever happen, leave blank.”
If I think the singularity is slightly less than 50% likely overall, what should I have put? It seemed off to leave it blank and imply I believed “I don’t think a Singularity will ever happen” because that statement seemed to convey a great deal more certainty than 50+epsilon%. However, if I actually believed there was a less than 50% chance of it happening, I’m not going to reach an even chance of happening or not happening on any particular year.
As a side note, after taking that test, I realized that I don’t feel very confident on a substantial number of things.
I think that there need to be two separate questions here. Probability of Singularity, and year it happens if it does. For instance, I’d guess about 70% chance of a singularity at all, but if it happens, 2040 would be about my expected date. You can’t describe these two statements in just one number.
Same here. But I voted 2150 because I think it’s 50% that it happens before 2150, 20% that it happens later, and 30% that it never happens.
Oooh, good answer. I hadn’t thought of that method.
I interpreted this as “there is an even chance of the Singularity falling before or after, [assuming it does]”. That is, if you think the probability that the Singularity will happen is something low like 1%, you should answer a year such that the probability it happens by that year is 0.5%. The only way you can’t answer it is if you’re sure it won’t ever happen.
(For example, if I thought a Singularity is very [...] very hard to achieve, I might answer 5000 AD or 500000 AD, depending on how many “very” there are, even though I might put a very low probability on our civilization actually surving that long.)
Given the expected date would be skewed to infinity by a non-zero estimate of the Singularity not occurring, you can probably put your estimate of the year X so that P( S ⇐ X | C ) = 0.5, where S is the statistic “Year singularity will occur” and C is the event “Singularity will occur”.