If I ever have an executioner, I want him to be Scott Alexander. Alexander has such a winning way with words that I and his many fans enjoy him even when we disagree. I’d hardly notice my destination as his pleasing patter entranced me while we took the long way around to the gallows.
You only need to optimize a single em a little to let it and its copies take over a whole industry. There's no reason a company would not take an em that does the same work but uses 10% less processing power. If the modified ems pay 5% of their wages back to the researchers because of the licensing, then the researches get to keep a 5% of the industry's worth of wages while there's no better em. And if they're the only such researchers, that would continue indefinetely. Which means such research is lucrative. Which means it would be done by a lot of people and would be much faster if not so lucrative.
Hm. I'm still unclear though as to what reason is there then to think that ems will have any "leisure time", in the sense of time to work on whatever they choose to, regardless of whether they can find a buyer for it.
The background assumption is that we will be emulating whole brains becase we don't know how to engineer AGIs out of components, so we have to brute force a solution. So why would we know how to re engineer brains...? It is a harder problem, since they are not cleanly designed.
I've been reading some Durkheimian stuff, and it occurs to me that em society repels people because it abolishes the individual self as a sacred object.
I have a hard time finding much substantive about the daily life of subsistence agriculture but what I found ( https://books.google.de/boo... ) seems to indicate that during subsistence farming almost no economical processes (wage competition) were at work. I'mnot clear what to make out of that.
I think this ignores opportunity costs. It might well be possible to gain some extra performance from ems by spending lots of time and money trying to modify their brains. But in order to conclude that this will be done you also need some reason to expect such experimentation to be a better use of those resources than any of the other possible ways they could be used.
Additionally, since research is largely a public good, I think you would expect a sub-optimal amount of it to be done even in a world of ems that are better at coordinating than we are.
We aren't living in the conditions where we can profit from every CPU operation we've freed and doing that is a matter of life and death. A corporation can easily experiment on ems, and if they're able to get rid of a half of an em's brain without it affecting work performance more than twice, they'll do it as fast as they can. It's even possible to propagate some changes, given that human brains are similar. And we aren't even talking about writing specialized algorithms from scratch and then replacing ems with them.
Today we have many legacy software systems that are hard to usefully change, even though we can change any bits we want to. Hard to figure out which bit changes would help.
I think farmers had lots of leisure because if you only have an acre of land and you want to grow corn on it, after planting the corn, watering the corn, and so on there's not that much extra you can do. Getting more land wasn't an option because most farmland was already occupied early in the farming era, and because there were different cultural norms around capitalism and reinvesting your proceeds back into making more money.There were also long fallow seasons when you couldn't grow crops even if you wanted to.
Ems are data, why should they be expensive to modify? Even if they are, the gains could be extremely large given their numbers. I don't in general care much about individual ants, so hunter-gatherer algorithms aren't in general too valuable either. And no, I don't really know why I value humans and not fruit flies.
Depends on the cost of modifying ems. If ems are expensive to modify, doing so won't be worth it.
Regarding your point on value, I'm curious: do you think specialized hunter-gathering algorithms have intrinsic value? Or in other words, what is it about humans that you think gives them value, that you don't expect to see in other intelligent entities?
Wouldn't there be companies that create more productive modified ems? It may be profitable, for example if any such em is paying the company some percent of the em's gains. The results would be the optimization of an em.for its particular area of expertise and cutting down on all the other processes to improve the rate of money earned to the processor-time. (I would start with cutting, it's probably easier). In time such ems turn into optimized algorithms for its job. And somehow I don't think a specialized contract-drafting algorithm has an intrinsic value.
It would at that, wouldn't it? Datura was also a suggestion.It's hard to know whether we're overfitting a convenient theory in retrospect since the evidence that exists isn't very good. Personal accounts are unreliable and even if it was practised the lack of actual records means we're still just dealing in rumours.
You only need to optimize a single em a little to let it and its copies take over a whole industry. There's no reason a company would not take an em that does the same work but uses 10% less processing power. If the modified ems pay 5% of their wages back to the researchers because of the licensing, then the researches get to keep a 5% of the industry's worth of wages while there's no better em. And if they're the only such researchers, that would continue indefinetely. Which means such research is lucrative. Which means it would be done by a lot of people and would be much faster if not so lucrative.
Hm. I'm still unclear though as to what reason is there then to think that ems will have any "leisure time", in the sense of time to work on whatever they choose to, regardless of whether they can find a buyer for it.
"productive" here is in an evolutionary sense of leading to more descendants (weighted by resources like speed).
The background assumption is that we will be emulating whole brains becase we don't know how to engineer AGIs out of components, so we have to brute force a solution. So why would we know how to re engineer brains...? It is a harder problem, since they are not cleanly designed.
I've been reading some Durkheimian stuff, and it occurs to me that em society repels people because it abolishes the individual self as a sacred object.
Well maybe in the em-world there will be a market for Scott Alexander copies who alk to other ems just before those ems get scrapped/executed...
I have a hard time finding much substantive about the daily life of subsistence agriculture but what I found ( https://books.google.de/boo... ) seems to indicate that during subsistence farming almost no economical processes (wage competition) were at work. I'mnot clear what to make out of that.
I think this ignores opportunity costs. It might well be possible to gain some extra performance from ems by spending lots of time and money trying to modify their brains. But in order to conclude that this will be done you also need some reason to expect such experimentation to be a better use of those resources than any of the other possible ways they could be used.
Additionally, since research is largely a public good, I think you would expect a sub-optimal amount of it to be done even in a world of ems that are better at coordinating than we are.
We aren't living in the conditions where we can profit from every CPU operation we've freed and doing that is a matter of life and death. A corporation can easily experiment on ems, and if they're able to get rid of a half of an em's brain without it affecting work performance more than twice, they'll do it as fast as they can. It's even possible to propagate some changes, given that human brains are similar. And we aren't even talking about writing specialized algorithms from scratch and then replacing ems with them.
Today we have many legacy software systems that are hard to usefully change, even though we can change any bits we want to. Hard to figure out which bit changes would help.
Ever played the board game Agricola? There are SO many useful tasks & chores one can do in a farming society, even in winter.
I think farmers had lots of leisure because if you only have an acre of land and you want to grow corn on it, after planting the corn, watering the corn, and so on there's not that much extra you can do. Getting more land wasn't an option because most farmland was already occupied early in the farming era, and because there were different cultural norms around capitalism and reinvesting your proceeds back into making more money.There were also long fallow seasons when you couldn't grow crops even if you wanted to.
Ems are data, why should they be expensive to modify? Even if they are, the gains could be extremely large given their numbers. I don't in general care much about individual ants, so hunter-gatherer algorithms aren't in general too valuable either. And no, I don't really know why I value humans and not fruit flies.
Depends on the cost of modifying ems. If ems are expensive to modify, doing so won't be worth it.
Regarding your point on value, I'm curious: do you think specialized hunter-gathering algorithms have intrinsic value? Or in other words, what is it about humans that you think gives them value, that you don't expect to see in other intelligent entities?
Wouldn't there be companies that create more productive modified ems? It may be profitable, for example if any such em is paying the company some percent of the em's gains. The results would be the optimization of an em.for its particular area of expertise and cutting down on all the other processes to improve the rate of money earned to the processor-time. (I would start with cutting, it's probably easier). In time such ems turn into optimized algorithms for its job. And somehow I don't think a specialized contract-drafting algorithm has an intrinsic value.
It would at that, wouldn't it? Datura was also a suggestion.It's hard to know whether we're overfitting a convenient theory in retrospect since the evidence that exists isn't very good. Personal accounts are unreliable and even if it was practised the lack of actual records means we're still just dealing in rumours.