97%: Savings continue to grow (1 util) 3%: Savings wiped out to prevent hardship for John. (0 util)
The first bet comes out at E(util): 17.19, the second at 0.97 utils.
You need to be very very risk-averse for the second option to be preferable. So risk-averse that I would not consider you rational (foregoing 16.22 utils to avoid a 3% chance of neg 300 utils?)
So risk-averse that I would not consider you rational
Er, you’ve implied the level of risk aversion by assigning utils, so of course it would be irrational to act more risk averse than John actually is, but it’s a weird way to phrase it. If John were more risk averse, the utils for hardship might be considerably lower.
Assuming we can speak cardinally of utility, John has to value his hardship as 33 times as bad as SIAI getting money is good for the first bet not to come out positive at all. If John cares not one whit for the SIAI that makes sense. If John is a codeword for the case that started this discussion that doesn’t make sense. Maybe I should have just made my point with the ratios.
I think most people would evaluate the hardship of having their credit ruined and being evicted as far greater than 33 times as bad as SIAI getting money (or SIAI getting money in 1 year when a 6-month cash reserve safety net has been built up). Also, it’s actually greater than 33 times, because you also have to include the probability that they won’t get another income source before they hit the catastrophe point, but even including that, I think most people would rate their life being ruined as orders of magnitude more negative utils than SIAI getting the donation is positive utils.
John is definitely not the case that started this discussion. My entire point is that Rain has a bunch of other compensating factors, which one could easily argue make it rational in that case. The issue under debate is whether somebody with none of those addtional compensating factors that Rain has would be rational to do the same, given just a reliable source of income and overdraft protection.
If you are a consequentialist and decide that donating $10,000 to SIAI is a good idea, then you already believe the benefit of $10,000 of the SIAI’s work significantly exceeds the benefit of saving the life of a child in the developing world. So now, would you say it is obvious that being evicted is 33 times worse than letting a child die because they lack basic medical care? Would you have even a handful of children die so that you can keep your credit rating?
I don’t know. I am pretty unconvinced that the SIAI will this much good in the world, but I would not call someone irrational if they risked everything they had to help ameliorate abject poverty and I would not necessarily call someone irrational if they believed that working on AI was more important than saving or improving lives directly.
To answer your questions, I don’t think it’s obvious that being evicted is 33 times worse than letting a child die (ignoring that the original question was about $2700, not $10,000), but it might actually turn out to be the case, since if somebody is evicted and has their credit ruined (and by hypothesis has none of the oher safeguards), it’s quite possible that they will never recover, and thus will never be financially secure enough to make future donations of vastly more consequence than the difference between a large donation now and a large donation in 1 year (after they’ve established an emergency fund).
I think the question is really whether it’s rational to donate all your savings now (if you have no reliable way of handling the unexpected case of losing your income source). Doing so greatly increases the probability of a personal catastrophe that one might not properly recover from. A more rational alternative, I would submit, is to donate a smaller amount now, while continuing to save until you have a sufficient emergency fund, and then donate more at that point. It is more rational, I believe, because the end results are quite similar (the same amount donated over the long term), but the personal risk (and the risk of not being able to make future donations) is greatly lessened.
I would call someone irrational if they risked a significant part of their potential future income (donations) to help ameliorate poverty, or help SIAI, immediately. Altruists need to look out for themselves.
I’d like to see somebody find a financial advice site or book that says you can periodically wipe out your sayings if you have a reliable source of income and overdraft protection on the empty account (and no other compensating factors, to beat the dead horse).
Sometimes it amazes me the things that people on LW will argue for just for the sake of argument.
Sometimes it amazes me the things that people on LW will argue for just for the sake of argument.
I may be motivated to argue the point because I just recently wiped out my savings to purchase a car and an amp—and I don’t have any overdraft, or even a credit card.
Hmm, I thought I was correcting an obvious error you made, and I expected you to immediately explain what you really meant or add some extra condition or retract your claim, and then we would have been done with the discussion.
When the question boils down to, “should someone with completely different circumstances do the exact same thing?”, my guess is the answer will typically be, “no.”
I challenge most hypotheticals by pointing out my method of averting crises: I build the appropriate circumstances such that the conflict will not occur.
The question, of course comes down to what utils are reasonable to assign. I too could choose numbers that would make either option look better than the other.
There’s also a wide range of available options between the two extremes you consider, and risk aversion should make one of them preferable..
And so John bets his current lifestyle that he won’t lose his job. That bet looks like:
97%: SIAI gets 2700 dollars. (27 utils)
3%: SIAI gets 2700 dollars, hardship for John. (neg 300 utils)
The bet you’re recommending is:
97%: Savings continue to grow (1 util)
3%: Savings wiped out to prevent hardship for John. (0 util)
The first bet comes out at E(util): 17.19, the second at 0.97 utils.
You need to be very very risk-averse for the second option to be preferable. So risk-averse that I would not consider you rational (foregoing 16.22 utils to avoid a 3% chance of neg 300 utils?)
Er, you’ve implied the level of risk aversion by assigning utils, so of course it would be irrational to act more risk averse than John actually is, but it’s a weird way to phrase it. If John were more risk averse, the utils for hardship might be considerably lower.
Assuming we can speak cardinally of utility, John has to value his hardship as 33 times as bad as SIAI getting money is good for the first bet not to come out positive at all. If John cares not one whit for the SIAI that makes sense. If John is a codeword for the case that started this discussion that doesn’t make sense. Maybe I should have just made my point with the ratios.
I think most people would evaluate the hardship of having their credit ruined and being evicted as far greater than 33 times as bad as SIAI getting money (or SIAI getting money in 1 year when a 6-month cash reserve safety net has been built up). Also, it’s actually greater than 33 times, because you also have to include the probability that they won’t get another income source before they hit the catastrophe point, but even including that, I think most people would rate their life being ruined as orders of magnitude more negative utils than SIAI getting the donation is positive utils.
John is definitely not the case that started this discussion. My entire point is that Rain has a bunch of other compensating factors, which one could easily argue make it rational in that case. The issue under debate is whether somebody with none of those addtional compensating factors that Rain has would be rational to do the same, given just a reliable source of income and overdraft protection.
If you are a consequentialist and decide that donating $10,000 to SIAI is a good idea, then you already believe the benefit of $10,000 of the SIAI’s work significantly exceeds the benefit of saving the life of a child in the developing world. So now, would you say it is obvious that being evicted is 33 times worse than letting a child die because they lack basic medical care? Would you have even a handful of children die so that you can keep your credit rating?
I don’t know. I am pretty unconvinced that the SIAI will this much good in the world, but I would not call someone irrational if they risked everything they had to help ameliorate abject poverty and I would not necessarily call someone irrational if they believed that working on AI was more important than saving or improving lives directly.
To answer your questions, I don’t think it’s obvious that being evicted is 33 times worse than letting a child die (ignoring that the original question was about $2700, not $10,000), but it might actually turn out to be the case, since if somebody is evicted and has their credit ruined (and by hypothesis has none of the oher safeguards), it’s quite possible that they will never recover, and thus will never be financially secure enough to make future donations of vastly more consequence than the difference between a large donation now and a large donation in 1 year (after they’ve established an emergency fund).
I think the question is really whether it’s rational to donate all your savings now (if you have no reliable way of handling the unexpected case of losing your income source). Doing so greatly increases the probability of a personal catastrophe that one might not properly recover from. A more rational alternative, I would submit, is to donate a smaller amount now, while continuing to save until you have a sufficient emergency fund, and then donate more at that point. It is more rational, I believe, because the end results are quite similar (the same amount donated over the long term), but the personal risk (and the risk of not being able to make future donations) is greatly lessened.
I would call someone irrational if they risked a significant part of their potential future income (donations) to help ameliorate poverty, or help SIAI, immediately. Altruists need to look out for themselves.
My apologies. I didn’t check the ancestors. I would note that wedrifid’s comment can be repaired with “sufficiently” in front of reliable.
I think the rational “fix” is to make sure you can stay afloat for at least a few months if a catastrophe happens. That is also the standard advice of every financial planning book I’ve ever read. And a Google search finds plenty of sites like: http://www.mainstreet.com/article/moneyinvesting/savings/how-much-should-you-save-rainy-day
I’d like to see somebody find a financial advice site or book that says you can periodically wipe out your sayings if you have a reliable source of income and overdraft protection on the empty account (and no other compensating factors, to beat the dead horse).
Sometimes it amazes me the things that people on LW will argue for just for the sake of argument.
I may be motivated to argue the point because I just recently wiped out my savings to purchase a car and an amp—and I don’t have any overdraft, or even a credit card.
To be honest that is what I saw you doing, hence the disengagement.
Hmm, I thought I was correcting an obvious error you made, and I expected you to immediately explain what you really meant or add some extra condition or retract your claim, and then we would have been done with the discussion.
When the question boils down to, “should someone with completely different circumstances do the exact same thing?”, my guess is the answer will typically be, “no.”
I challenge most hypotheticals by pointing out my method of averting crises: I build the appropriate circumstances such that the conflict will not occur.
The question, of course comes down to what utils are reasonable to assign. I too could choose numbers that would make either option look better than the other.
There’s also a wide range of available options between the two extremes you consider, and risk aversion should make one of them preferable..