This paper argues that at least one of the following propositions is true ...
Ah Nick Bostrom’s simulation hypothesis! Though you can argue it’s inspired more by popular fiction - Bostrom wrote his famous paper in 2003, 4 years after the mildly popular move The Thirteenth Floor
and the wildly popular The Matrix
first thrilled moviegoers – than Bostrom’s philosophical imagination, he put a wonderful spin on it by casting his questions in terms of probability: in short, if any people have managed to create an Thirteenth Floor-like simulation, they’ve likely run such programs huge numbers of times, so there is 1 original reality vs. a gigantic number of simulated ones, thus the probability we’re in the original reality rather than a simulation is tiny.
Before poking into this rabbit hole
, and your twists to it ,SP, I’ve got to address some science-y stuff:
Keep in mind nothing has ever left our solar system, no physical man-made object.
Voyager 1, launched in 1977, left the solar system – that is, passed the heliopause, about 120 AU out, to entered space no longer dominated by the Sun’s radiation - sometime in 2013. It’s currently over 133 AU out, moving away at more than 17 km/s. If, instead of the heliopause, you consider the solar system’s boundary to be the Oort cloud, which ends somewhere from 50000 to 200000 AU (over 3 ly, closer to another star than the Sun) out, Voyager won’t be out of it for 30000+ years – but it will get there.
Even though we've put a man on the moon over 50 years ago we haven't done anything since - the only time in history our expeditions have been stagnant is with the frontier of space.
While the 1969-1972 Apollo missions placed humans further from Earth than ever before or since, I wouldn’t say we humans have done no space exploration since. We’ve put at least one spacecraft in orbited around 6 of the 8 planet, flown by the other 2, and flown by many dwarf planets and minor bodies. We’ve put landers on 4 planets, 3 moons, and 2 asteroids and a comet. It’s just that these spacecraft have been robotic, not manned.
Manned spaceflight thrills the popular imagination, and a tool-equipped human is more versatile and adaptive than any present-day robot, but we’re also fragile, and burdened with ethical requirements. We think nothing of sending robots on one-way “suicide missions” and “faster, better, cheaper” missions with low probabilities of success – this kind of mission plan is by far more common than ones where a spacecraft returns to Earth – but can’t ethically do this with people. Manned spacecraft need to carry systems to protect and sustain people, and must minimize travel time, so must be bigger and more expensive than robotic ones, so we get much more scientific data for the same amount of money from robots than from human in space.
Science and technology enthusiasts tend, I think, to fail to understand that most people consider spaceflight a poor use of wealth, preferring programs of direct benefit to them. Present day scientific space mission planners must struggle to get money to do the most and best science what they can, which precludes manned missions. I think manned spaceflight is, with rare exception, a product of political posturing, nations showing off their ability to do the difficult. In the 1950s, ‘60s, and ‘70s, this was the “space race” between the US and the USSR. The race has de-intensified somewhat since then, and admitted more nations, and become more internationally cooperative, but I think it’s still more international sport completion than science.
(sources: Wikipedia articles List of Solar System probes
, List of landings on extraterrestrial bodies
)Now, back to the simulation hypothesis:
However, wouldn't we notice if we were living in a virtual reality? At best they'd have an imperfect imitation of their world and everything within it.
A key part of the definition of a virtual reality is that it’s a kind of computer program, so the answer to all these questions is: it depends on how the program is written.
The “ancestor simulation” Bostrom writes about are programs intended to allow people to learn about their past (and themselves) by simulating it. In a program like this, the people – the “we” in your question, let’s call them “agents” to avoid confusion – are collections of data accessed and manipulated by the program. Assuming the computer being used has an architecture similar to present-day ones (not, I think, an unreasonable assumption), then the program can be stopped, data inspected and manipulated, at the will of its user. So, while an agent might enter a state of noticing they were living in a VR, the user could be aware of this, and if it were undesirable, undo it, or the programmer modify the program to prevent it.
As post-humans, perhaps strong AI built by taking from the structure of living people themselves, physically assimilating man with machine, wouldn't know all the variables that created baseline reality.
So my whole thesis here is that they'd need a pristine, genuine human from the past to plant in this world, and go off of his expectations to reconstruct other aspects of the bygone human civilization to know how their ancestors thought and interacted to form culture, religion, society, and all the things that eventually led to their existence.
That being said, and that is an important point, we have a real human mind being put in a pseudo-real world; would this human not be able to recognize that something is off?
I think most people imagine that programmers in a posthuman civilization would not need an actual ancestor – a “pristine, genuine human from the past” – to create an ancestor simulation, though the exact circumstances of this scenario are not, I think, given. If there were no data-losing breakdowns of civilization between the ancestors’ time and the programmers’, and the simulation was of the about 1980 or later, the complete genomes of many humans and animals would be available to the programmer, so presumably they could practically perfectly simulate an ancestor from it. High-quality data from the past – statistics, documentary recordings, etc. – would also be available If there data-losing breakdowns of civilization, this data might be lost, and the programmer might have to build the simulation based on scientific principles and available data. They might program and run a huge number of variations based on different assumptions, never knowing which if any of them best matched actual past events.
I don’t think precisely reproducing the past would be a major goal of ancestor simulating, especially if high-quality data of that past was available. The greatest value might be to answer “what if” questions about pasts that never happened. Ancestor simulation might not be the most important simulations, because presumably, like us, our distant descendants will be more interested in their future than their pasts, so will find descendent
simulations more important than ancestor
Thinking more deeply, a major flaws I find in Bostrom’s idea are
- In focusing on ancestor simulations, it doesn’t speculate widely enough about what the most important simulations might be, and that they might intentionally be very different than accurate ancestor of descendent simulations
- In assumes that the simulation is run on a physical computer, it failing to speculate that our current physical reality might be an artifact, a kind of computer. This idea is pretty far down the rabbit hole.