Hi. Your big point (the amount of steering we can hope to do is more like swimming in a river than altering its course) is a good one which I pretty much take for granted, so you won't get any argument from me there. Instead, I'll nitpick about ems.

Much as I like the general idea ("Humans 3.0" in my essay) I have yet to see a convincing description of the technology. A couple of things...

view entire post

Hi. Your big point (the amount of steering we can hope to do is more like swimming in a river than altering its course) is a good one which I pretty much take for granted, so you won't get any argument from me there. Instead, I'll nitpick about ems.

Much as I like the general idea ("Humans 3.0" in my essay) I have yet to see a convincing description of the technology. A couple of things jumped out at me while reading your essay:

"Ems would often spin off copies to do short term tasks and then end when those tasks are done."

How? A straight copy would be indistinguishable from the original. So you'd have two identical individuals, both thinking that the other one should just do its assignment and then have the good taste to commit suicide. If you propose to modify the copy to behave as expected, you lose the proposed advantage of ems over other approaches to AI, not needing to understand their inner workings in detail. Or is your proposal to create the copy in a torture chamber, suitably equipped to compel obedience and then kill upon task completion?

"Ems would split by speed into status castes, with faster ems being higher status."

Why would any em willingly run slower than technically possible? Would you willingly take a drug which slows you down? You mention gender imbalances as a possible cause, but if copying is so trivial, the gender in low supply could just duplicate itself.

Another speed-related problem: are your ems running on general purpose computers? The implied trivial ease of copying suggests as much. Have you tried to estimate the computational requirements for emulating even a single human brain in real time? Which technology are you suggesting will be used to run trillions of them, as you say, within a century?

When I inventoried computing technologies currently on the horizon, the only realistic candidate I found was neuromorphic hardware. It is far from obvious that it could be used for the kind of general purpose computing needed to run virtual worlds, and if it can't, ems will have a big speed problem.

view post as summary

show all replies (1 not shown)

hide replies

Robert,

Thank you for your original, entertaining and thought-provoking essay. I really like your metaphor of the future being like a wide, fast, murky river at night. I've been thinking a lot about your opening statement, that we should work hard to clearly see the future we will get if we do NOTHING to steer it (because most likely humanity will do little to steer it).

If what...

view entire post

Robert,

Thank you for your original, entertaining and thought-provoking essay. I really like your metaphor of the future being like a wide, fast, murky river at night. I've been thinking a lot about your opening statement, that we should work hard to clearly see the future we will get if we do NOTHING to steer it (because most likely humanity will do little to steer it).

If what you're saying is true, it should also apply to the past: for instance, people in 1900 should have thought very hard to clearly see the year 2000 they would get if humanity did nothing in particular to steer the future. The interesting question to ask ourselves, then, is this :

In retrospect, how inevitable was the general course taken by the history of the world during the 20th century?

This suggests a fascinating (but difficult) thought experiment: imagine that we rewind the clock to 1900, and consider the range of the most plausible possible 20th centuries that humanity could have had. If we eliminate "wild cards" like a major asteroid strike or an extraterrestrial invasion, do we always get to a year 2000 that looks reasonably like the one we got --- 6 billion people, capitalism "victorious", reasonably cheap personal computers, the Internet growing rapidly, cell phones, cars running on fossil fuel everywhere, almost no use of nuclear power in warfare but also in electricity generation (except in a few countries), a timid space program with only a few humans in low Earth orbit, huge inequalities between countries, but rising wealth overall, falling death rates and birth rates, etc. --- or was there a fairly wide range of plausible 20th centuries that could have happened instead, if humanity (or some of its most dominant actors) had steered the future differently?

Studying this alternate-historical thought experiment could help in finding out the degree of "steerability" of the stretch of river that we have ahead of us! I think it could make for an interesting side-project for the Institute for the Future of Humanity, or for other groups working in futures studies.

In my essay, "To Steer Well We Need to See Clearly: the Need for a Worldwide Futurocentric Education Initiative", I propose that we should work hard to put together a "futurocentric curriculum" aimed at schools but also at lifelong education, in order to raise the level of public interest and knowledge about the topics that are the most relevant to the future of humanity. I think that the question of the degree of "steerability" of the future that you raise, and the alternate-past thought experiments that it suggests, would be quite interesting for the history component of the futurocentric curriculum!

Marc

view post as summary

show all replies (2 not shown)

hide replies