It is clear that the following disjunction is true: posthuman civilizations either tend to create simulations of conscious humans or they don’t. In the latter case, there is presumably some simblocker, of which Bostrom delineated two: that intelligent species rarely reach the ability to create these simulations or that they aren’t interested in doing so. There are other simblockers such as it not being possible in the first place, but the point is that the disjunction which I just spelled out is a tautology and must be true. It either happens or it doesn’t.
However, Bostrom makes some naive assumptions when analyzing the former case. He first argues that if posthuman civilizations tend to create simulations of humans, then the fraction of people currently living in a simulation is close to 1 (1). Secondly, he argues that if that claim is true, then we are most likely living in a simulation (2).
This first leap (1) that he makes is fallacious in that it confuses possibility with probability. Bostrom essentially claims that if it is possible to make trillions of simulations, then they already exist. Consider the possibility that the reality that we exist in is the only reality that exists (i.e. we don’t live in a simulation and there are not currently any conscious simulations of humans). If the probability of this being the case were P, then the probability that we are in a simulation drops to at most 1 - P. Since Bostrom claims that close to 100% of humans live in simulations, he is saying that 1 = 1 - P, or that P = 0. Bostrom arbitrarily assigns a probability of 0 to us living in the top layer of reality. Since this is the only reality we know and we have no information about anything outside of or even tangential to this reality (like say the probability of this being the only reality), it makes no sense to assign 0 to the probability of our reality being the only one that exists. Bostrom groups our reality into a collection with all possible simulated realities and assumes they are all equally likely when generating the figure “close to 1”; but this is illogical simply because our reality being nonunique is a prior for even the existence of simulated realities. This would be like saying since there are an infinite number of things I could find in my pocket right now and only one way to find nothing, the probability that there is something there is close to 1. This obviously makes no sense because I know there’s nothing in my pocket right now, and if I didn’t know that, it would make no sense to assume a uniform distribution of probabilities for all different objects I could find there including nothing. In particular, it is fallacious to treat finding nothing as comparable to finding an object since finding something (not finding nothing) is an important prior to finding any object that might be there.
Bostrom defends his second implication (2) with the example of a junk gene with no physical marker or way to be detected. If x% of the human population has this gene, then you yourself have an x% chance of having this gene. He says this is analogous to the simulation argument in that if close to 100% of people live in simulations then you can be pretty sure that you are in a simulation. This evidence is sound, but only assuming you know this percentage “x”. The only way to acquire this percentage “x” is by counting frequencies. For the gene case, this would mean literally sequencing all of humanity’s DNA or at least of a random sample of humanity. But notice that in order to create this probability, one must at least observe the gene in humans. Bostrom asserts the probability of ~1 without ever having encountered any concrete evidence of a simulation. We can only acquire a percentage “x” if we at the very least actively observe humans in a simulation. Creating a probability from anything else would be like saying that since its possible to have DNA that encodes for this gene, someone probably has it.
The ultimate problem with Bostrom’s reasoning here is that he is postulating about realities beyond our own of which we have no knowledge. Until we have knowledge about things that occur outside of our reality, it is naive to assume that simulations being possible implies that they currently exist.