The Ghosts Of Em

Our basic concept of “death” is binary, so that one is either dead or not. But we often metaphorically extend the concept to a continuum. For example, people who have more strength, energy, passion, and awareness are said to be “more alive,” and those who have more power, prestige, influence, or wealth are also said to be “more” in many ways, including more central and alive. Since sleepers have less of all of these things, sleep is often seen as a partial death.

We have a related mythical concept of “ghost,” which is also sometimes made into a continuum of ghostliness. A ghost was once human, but then died, and now is an active agent with death-related features. So ghosts tend to be cold, sick, in low mood, and have a weak influence on the physical world. They are typically distracted, unaware of, and disinterested in humans. Ghosts are anti-social, avoid groups of more than a few humans, and don’t collect into ghost gangs or ghost cities. They are reluctant to move away from their old haunts, and remain obsessed with old issues. Ghosts are heard more than seen, rarely speak words, and are seen more in unusual viewing modes such as night, shadows, and mirrors.

Slow em retirees share many features with people we see as “less alive,” including ghosts. Not only are they literally closer subjectively to dying soon due to civilization instability, their minds are also more inflexible and stuck in their ways. Compared to faster working ems, slow retirees have less awareness, wealth, status, and influence, and they are slower to respond to events, including via speaking words or coordinating with others. Retirees may often watch and judge working ems, and in such roles may only be visible only in special views.

Thus ems may come to see slower ems as ghostly, and more ghostly when slower. Such em ghosts are real, and with trouble one can talk to them, but like ghosts they aren’t very useful as allies, they sometimes hurt people they interact with, and so since one is usually free to ignore them, that is usually the wise strategy. Since ems must pay for faster speeds, for ems being more alive is more directly related to having more money to spend.

If “beneath” each em are many layers of a ghostly underworld, just how deep does this abyss go? Katja Grace at AI Impacts just helped me out by estimating the ratio of costs, using today’s technology, to store a brain state and to run a human-speed brain emulation. This ratio equals the “base” em speed as a fraction of human speed. This is near the lowest reasonable speed for ems, since well above it cost is proportional to speed, and well below it cost is independent of speed.

Apparently, plausible estimates of this base speed range from one hundredth of a trillionth of human speed up to one millionth of human speed, with a middle estimate of one tenth of a billionth of human speed. This ratio apparently hasn’t changed much over four decades, giving reason to hope it can help us estimate the future base speed. I’ve separately estimated typical em speed to be one thousand times human speed, and the maximum speed where speed is still proportional to cost to be a million times human speed.

Thus the range of speeds over which em speeds are about proportional to cost is at least a factor of a trillion, and may be a billion trillion. Thus for typical speed ems the underworld abyss of slower ghostly speeds is very deep! If your career and investments go badly, and you are forced to cut back and slow down, there is a very long hill to slide down before you finally reach bottom, where the only lower place to go is to be erased. Em inequality in speeds is immense.

I just added this stuff about ghostly ems to my book The Age of Em: Work, Love and Life When Robots Rule The Earth. And since I’m to turn in the final draft by Saturday, this will be the last thing I add. Publication date still not set; I’ll tell you when I know.

GD Star Rating
Tagged as: ,
Trackback URL:
  • endril

    Another theme of ghosts is fixation and hangups about their previous existence, grudges, misery and anguish. I hope slow ems can mostly chill out and enjoy a high-pleasure afterlife.

    • OK, I added a clause about that to the post.

  • Seattle Steve

    I shall remain happily beholden to my bacterial overlords in the physical realm.


    I wonder if some ems would choose to slow down, perhaps to wait for something they want or just to see what the “world” will be like in the future… Would there be mechanisms that can reliable protect an ems assets and investments while “hibernating” like that?

    On the other end of the spectrum some slow or average ems might save up for a temporary speedup, say to get extra time to study for an exam.

    • sflicht

      It’s a good question. Humans often choose to “unplug” and go on vacation without access to modern communication infrastructure. We also like to appreciate animals, e.g. at a zoo or aquarium. An em would have to slow down immensely to do so (or else it would be like the way we appreciate statuary or glaciers). But I doubt many would make this choice. If we found some cool animals in the Tau Ceti system, and couldn’t bring them back to Earth, the only way to see them would be to go on a “safari”, but by the time you returned the world and all your friends and family would be totally different. Hard to imagine that being a particularly popular form of tourism. The situation for ems choosing to slow down to hang out with ghosts, humans, or animals would be even more extreme.

  • Dave Lindbergh

    And yet historically the ratio of processing to memory in real machines has been steadily dropping (not constant).

    See the second graph here (one of my blogs):

    If that continues, does your thesis hold up?

    • Your graph looks at RAM instead of disk memory, and seems to show the ratio as pretty constant for sixty years. During the twenty years before that processing got cheap faster than memory, so during that period the base speed was rising at least in terms of RAM. That doesn’t threaten the idea of a future base speed, it just makes it harder for us to estimate that future speed.

  • Cambias

    Wow! I got a couple of story ideas just from this brief summary. Now I can’t wait for the book.

  • Sam Dangremond

    It makes me feel skeptical when I read something that the leading experts can’t estimate within an accuracy range of less than one million.

    Guessing about the number of powers of ten makes all of this sound more like “counting the number of angels that can dance on the head of a hard drive.”

    • So you think there is only high accuracy (low error) or zero accuracy (infinite error), and nothing between?

      • Sam Dangremond

        Pretty much actually.

        Test: “is this information, given the margin of error, useful for doing anything further with it?”

        Another way of putting it… guesstimates with a margin of error of a million are closer to philosophy than to science, engineering, or (what lots of us think of as) economics. Nothing wrong with philosophy though.

  • Lord

    Unless they are individual automatons, it seems more likely processor speed would be a common utility and improvements would benefit all though some would be more productive with it, though speed would be less important than currency and usefulness, older, less useful ones moving into storage, archived for history and study. If they had their 20 year old selves saved, maybe they would even choose reversion followed by current updates. The relationship between selves could become complex.

  • Proper Dave

    I don’t know if you have red Permutaion City by Greg Egan published around 1993, but it basically deals with this problem about ems being unemployable and much more.

    • I read it when it first appeared, and upon rereading it recently was disappointed in how little it dealt with.