Interstellar Travel and Uploaded Humans
Many difficulties of space travel go away when human consciousnesses have already been translated into software. For example,
“Suspended animation” is a simple matter of not running the software program and sticking with a copy
Even if the stored human consciousnesses are physically traveling on a spaceship, the danger from radiation can be taken care of by having many backup copies and doing periodic error correct. Other dangers such as muscle wasting from zero gravity are absent.
Even if the stored human consciousness are physically traveling on a spaceship, the total weight needed is likely to be much, much smaller than the weight of humans.
There is no need to worry about food and oxygen for the journey. And power needs should be relatively low if for most of the human consciousnesses on the journey error correction is the only operation being performed.
This should make even interstellar travel realistically possible without any space travel technology beyond what we can already foresee. Michio Kaku raises the additional possibility of transmitting a human consciousness on a laser beam. Here is a summary of that idea plus a little background from Adam Kirsch’s June 20, 2020 Wall Street Journal article “Looking Forward to the End of Humanity”:
Ultimately, however, the hope is that we won’t just use computers—we’ll become them. Today, cognitive scientists often compare the brain to hardware and the mind to the software that runs on it. But a software program is just information, and in principle there’s no reason why the information of consciousness has to be encoded in neurons.
The Human Connectome Project, launched in 2009 by the National Institutes of Health, describes itself as “an ambitious effort to map the neural pathways that underlie human brain function.” If those pathways could be completely mapped and translated into digital 0s and 1s, the data could be uploaded to a computer, where it could survive indefinitely. The physicist Michio Kaku has theorized that this is how humanity will overcome the logistical challenges of deep-space travel: “We’re going to put the connectome on a laser beam and shoot it to the moon. In one second, our consciousness is on the moon. In 20 minutes we’re on Mars, in eight hours we’re on Pluto, in four years our consciousness has reached the nearest star.”
However, Robin Hanson’s book The Age of Em, which I feature in “On Being a Copy of Someone's Mind,” gives economic arguments for why interstellar travel of software humans, while it is likely to happen, wouldn’t be central to early historical developments after technology to upload human consciousness’s becomes available. Basically, there are a lot of economic reasons why human beings, whether flesh and blood or software humans, want things done fast. And by the time we have the technology to upload human consciousness at all, it is likely that software humans can operate at a much faster pace than flesh-and-blood humans: a thousand or a million times faster.
What would a trip to Pluto look like if the typical software human is operating at 1000 times flesh-and-blood speed? Traveling at the speed of light on a laser beam, it then takes 16000 subjective hours of the software humans back home to go to Pluto and return, which is 22 months. That would seem like a long time to be gone. On the other hand, one copy of you could go while the other stays behind to keep working. So that could be quite attractive.
But even if a receiver station was in orbit around Alpha Centauri at the beginning of the age of software humans, a roundtrip of 8.6 years of flesh-and-blood human time would be 8,600 years of software human time. That would make it seem like a one-way trip. The one-way trip could be quite attractive if copies of all one’s family and friends and many other potential friends took the trip too, but anyone who went to the Alpha Centauri system wouldn’t have any influence on software human civilization for a period of 8,600 subjective years (of the average working software human).
Thus, interstellar travel, while a lot easier when there human consciousness can be uploaded, and likely quite consequential for the galaxy, would have very little effect on human history on earth for what will seem like a long, long time for many software humans.
In closing, let me several background points about the predicted scenario.
First, flesh-and-blood humans would not be central to this future simply because the ease of copying software humans would mean that there could easily—and profitably would be—trillions and trillions of software humans, while the number of flesh-and-blood humans would be measured in billions. Flesh-and-blood humans willing to be uploaded would be important as the source of real variety among software humans. Also, from the perspective of flesh-and-blood humans, the subsistence cost for a software human would be tiny compared to the subsistence cost for a flesh-and-blood human. That means that from the perspective of a software human, a flesh-and-blood human would seem incredibly rich. (The only alternative to a flesh-and-blood human seeming incredibly rich to a software human would be for the flesh-and-blood human to be at below a subsistence level for a flesh-and-blood human and therefore to die. Big, hulking things such as flesh-and-blood humans are expensive.)
Second, one should not think of software humans as being incorporeal. It would be very easy for a software human to move back and forth between being embedded in some kind of physical android form (though that physical form need not be human-shaped) and being embedded in a virtual world—that is, being purely electronic in form. A software human might specialize in one role or another, but fundamentally doesn’t need to choose. It is an easy transition back and forth. Generally speaking, a physical form not needed for work purposes is likely to seem relatively expensive to a software human, so most wouldn’t have a physical form outside of work hours. And as Robin Hanson writes, a large share of software humans, like most flesh-and-blood humans in advanced countries now, are likely to do office work, which leaves them without much reason to incur the expense of a physical form.
A scene from “Upload”
I have started in on the fun TV series “Upload.” It is interesting to compare what happens in the series with Robin Hanson’s predictions (conditional on plausible future technology):
The idea in “Upload” that there might be a period of time when the only connectome scanning technology available destroys the brain that is scanned is plausible enough. Probably at some later period, non-destructive scanning of the connectome will become possible.
“Upload” raises the issue of software humans working, but assumes a law prohibiting it. The commercial advantages of letting software humans work are likely to make such a law difficult to maintain politically in all jurisdictions. And the jurisdictions that allow software humans to work will get much richer than jurisdictions that don’t. As Robin Hanson points out, giving flesh-and-blood humans a small share of the gains from software humans working—even through something as simple as an income tax on software humans—is likely to make the flesh-and-blood humans to not only acquiesce in allowing software humans to work but to actively encourage it.
One of the plot-driving challenges in “Upload” is low-quality virtual reality for software humans. This is quite unlikely for one simple reason: human-brain-emulating software will be vastly more complex— and therefore vastly more expensive to run—than quite high-quality virtual reality software. Some kinds of virtual reality will have an expense that is a substantial fraction of the cost of running a software human—for example, running a scanned cat or dog as opposed to a simplified virtual-reality cat or dog—but many, many high-quality virtual reality features will cost a tiny fraction of running software needed to create human experience for that software.
Finally, “Upload” shows zero recognition in the first few episodes that software humans can operate at a different speed than flesh-and-blood humans—either faster or slower. This possibility would change so many decision points in the first few episodes that it will be hard for the writers to introduce this possibility later on without creating many retrospective plot holes in the first few episodes. As it is, they can hope viewers simply don’t think of this possibility. But technologically, once you have software, it is easy to rig it to run at different speeds. For example, I am now running Peter Attia’s podcasts at 1.2 speed. No problem. One important consequence of the different speeds of software humans and flesh-and-blood humans is that while interactions between these two groups will be easy, ordinary relationships might well be difficult: running at a speed that allows a software human to have a deep relationship with a flesh-and-blood human would mean sacrificing ease of having relationships with the bulk of other software humans. (And even among software humans, there would be divisions based on different operating speeds—perhaps driven by occupational differences in economically optimal speed. But with trillions of software humans likely in existence, there would be many software humans in each speed category to form relationships with.)
I find thinking about the future that Robin Hanson has conditionally predicted fascinating. It is a good topic to return to again and again. And as should already be clear, I highly recommend The Age of Em.
Related Posts: