Rather than get into the details, Iâll make the meta-level point that the impact of your action here is likely to be very small in one direction or another.
At best, you are one more computer in a network of millions*; at worst, youâve added a tiny amount of pollution to the air, which might take a few minutes to an hour off of humanityâs collective lifespan, if we stick to Gwernâs reasoningâyou might waste more human life in the course of spending time to install the software than you would actually running the program.
Meanwhile, the âindirect costsâ are based mostly on money you could otherwise donate to charity, a consideration which could come up every time you spend money on anything (and which is generally better to ignore unless youâre making a big spending decision; I wouldnât worry about $10/âyear).
Given the complexity of the issue (e.g. trying to calculate your computerâs extra electricity usage, evaluating the expected value of papers produced through FAH), I would recommend against trying to make a serious calculation of your impact. As with many questions people ask in EA spaces, âdonât worry about itâ is a reasonable answer.
----
*There are only about 100,000 machines in the FAH network right now, but many of those were designed specifically for high-performance computation; Iâd be unsurprised if an average home machine contributed one-millionth or less of the projectâs processing power.
I agree that the impact of this decision is likely to be very small, but trying to analyze a complicated phenomenon can be personally beneficial for improving your skills at analyzing the impact of other phenomenon. In general, it seems good for EAs to practice analyzing the impact of various interventions, as long as they keep in mind that the impact of the intervention and the direct value of the analysis might be small.
This might be the case, though if someone has the time to analyze a complicated phenomenon and wants to get practice, I think they should take a bit more time to choose a phenomenon to start with, so that they can get one with other useful characteristics. For example, they might try to find something with a larger expected magnitude of impact, positive or negative, or to choose a question that is of direct relevance to the EA community (e.g. something which is an active topic of debate, or involves some very common thing many people in EA do).
Along those lines, I like Gwernâs study of melatonin, which involves a bit of self-experimentation but also expected-value calculations. Various other productivity tools/âstrategies could also be solid candidates.
Sometimes I do blatantly useless things so I can flaunt my rejection of the often unhealthy âalways optimizeâ pressures within the effective altruism community. So today, Iâm going to write about rock music criticism.
I certainly donât endorse âalways optimizeâ! I spend far too much time reading manga and trying to win Magic: the Gathering tournaments for that. I fully endorse analyzing things that are interesting/âentertaining. But it seems bad to get stuck with something that is both low-expected-impact and low-interest. Someone who really likes Folding@Home should totally give the analysis a go; someone who doesnât care and just wants evaluation practice has many other options.
Rather than get into the details, Iâll make the meta-level point that the impact of your action here is likely to be very small in one direction or another.
At best, you are one more computer in a network of millions*; at worst, youâve added a tiny amount of pollution to the air, which might take a few minutes to an hour off of humanityâs collective lifespan, if we stick to Gwernâs reasoningâyou might waste more human life in the course of spending time to install the software than you would actually running the program.
Meanwhile, the âindirect costsâ are based mostly on money you could otherwise donate to charity, a consideration which could come up every time you spend money on anything (and which is generally better to ignore unless youâre making a big spending decision; I wouldnât worry about $10/âyear).
Given the complexity of the issue (e.g. trying to calculate your computerâs extra electricity usage, evaluating the expected value of papers produced through FAH), I would recommend against trying to make a serious calculation of your impact. As with many questions people ask in EA spaces, âdonât worry about itâ is a reasonable answer.
----
*There are only about 100,000 machines in the FAH network right now, but many of those were designed specifically for high-performance computation; Iâd be unsurprised if an average home machine contributed one-millionth or less of the projectâs processing power.
I agree that the impact of this decision is likely to be very small, but trying to analyze a complicated phenomenon can be personally beneficial for improving your skills at analyzing the impact of other phenomenon. In general, it seems good for EAs to practice analyzing the impact of various interventions, as long as they keep in mind that the impact of the intervention and the direct value of the analysis might be small.
This might be the case, though if someone has the time to analyze a complicated phenomenon and wants to get practice, I think they should take a bit more time to choose a phenomenon to start with, so that they can get one with other useful characteristics. For example, they might try to find something with a larger expected magnitude of impact, positive or negative, or to choose a question that is of direct relevance to the EA community (e.g. something which is an active topic of debate, or involves some very common thing many people in EA do).
Along those lines, I like Gwernâs study of melatonin, which involves a bit of self-experimentation but also expected-value calculations. Various other productivity tools/âstrategies could also be solid candidates.
cf. Gwernâs study of catnip.
Also Lukeâs post on Scaruffi:
I certainly donât endorse âalways optimizeâ! I spend far too much time reading manga and trying to win Magic: the Gathering tournaments for that. I fully endorse analyzing things that are interesting/âentertaining. But it seems bad to get stuck with something that is both low-expected-impact and low-interest. Someone who really likes Folding@Home should totally give the analysis a go; someone who doesnât care and just wants evaluation practice has many other options.