Assume two planets, Aleph and Beth, are one light year apart. If Aleph wants to communicate with Beth, it must send a signal to Beth, which travels for a year before it is received. Beth’s response takes a year to travel back to Aleph, giving a total of two years that Aleph needs to wait before hearing an answer. This latency imposes a serious constraint on communication over very long distances. There is in fact a way to get around this without violating the premise that information cannot travel faster than the speed of light. The sketch is as follows. Suppose Aleph has a simulation of Beth and Beth has a simulation of Aleph. For the time being assume that these simulators are entirely accurate (when we fully develop the idea we will weaken this assumption considerably). This means that if something happens on a planet, it is observable on the other’s simulator, and further if something is observable on a planet’s simulator, then it must be happening on the other planet. Given such a set up, Alphonse on Aleph can communicate instantaneously with Bethan on Beth, as long as they are both in front of the simulator. Alphonse asks Bethan a question. Bethan sees the question on her simulator and responds. Alphonse sees the response on his simulator and the exchange is complete. The distance between Aleph and Beth did not impact how long the exchange took (although it will matter to some degree when we move from this sketch to a more realistic model). Furthermore it did not violate the premise that information can not travel faster than the speed of light, as the information was already inside the simulators. The above argument relies on the premise that the simulators are entirely accurate. We will show in what follows, that the argument still works if we assume far milder (and more reasonable) capabilities of the simulators. The above is only intended to give some idea of where things are going. Besides weakening this assumption we also discuss the problems of two planets simulating one another, specifically how to get around the apparent contradiction of simulating the others simulator. This will all be discussed in detail below. The overarching goal of this piece is to show that realtime conversations over large distance are consistent with special relativity. Whether or not it is actually possible depends on whether or not simulators with certain properties are realizable. ## One-Sided Simulations For the time being let us restrict our view to the planet Aleph simulating the planet Gimel, which does not itself contain a simulator. It seems reasonable to assume that this simulator can not simulate the entire universe, but instead can only simulate some finite region around Gimel. This assumption alone allows us to prove that the simulation of Gimel will not remain perfectly accurate. The light from stars outside of the simulated region will eventually reach Gimel and the simulator will not have access to this information. Eventually we expect that all of this accumulated, unaccounted for information will lead to some minor divergence from the simulation, which in turn will snowball into a major one. It is also possible for some sort of celestial event outside of the simulated region to cause a major divergence by itself, for instance, if the people of Gimel were to witness a supernova. And so it seems that perfectly accurate simulations are but a fairy tale. We now see that, at the very least, the accuracy of a simulation must decrease with time. If this were the end of the story then the proposed communication system could only ever be useful for a brief window. Thankfully this need not be a fatal blow. Suppose a supernova witnessed on Gimel causes a major divergence. Gimel could immediately send a message to Aleph, alerting them of the event. Assuming Alpeh and Gimel are $y$ light years apart this message would take $y$ years to reach Aleph, but once recieved Aleph could incorporate this information into the simulator and increase its accuracy. The above example generalises to the idea of Gimel sending calibrating data to Aleph to keep the simulator on track. Any data sent in this way takes $y$ years to travel and so we must be sure that this resolution to the problem does not actually reduce this setup to the naive situation where Aleph needs to wait $y$ years before it can hear about events that occured on Gimel. Suppose Aleph’s simulation $S$ remains sufficiently accurate for $s$ years of Gimel that it simulates. To illustrate what this means, suppose $s=7$ and that $S$ starts simulating Gimel at the year 2000 (which is to say that in the simulation it is the year 2000 on Gimel; it need not be the year 2000 on Aleph where this simulation is being run). Then this says that $S$ is still sufficiently accurate while it is simulating the year 2007 on Gimel. Sufficiently accurate here means some level of accuracy, high enough for the simulation to be deemed useful. Now if the distance $y$ between Aleph and Gimel is less than the number $s$ of years that the simulators are accurate for, then it doesn’t matter that it takes $y$ years for the calibrating information to reach its target as the simulator will remain accurate while the information is traveling, and by the time the calibrating data is relevant it will have already arrived at its destination. There are many ways that this calibrating data idea can be implemented. One option is to send data that will tweak the running simulator, steering it towards a more accurate description. Another would be to send a large batch of data to be used as initial conditions for a simulator to use and run from scratch. Below we describe a system that implements the latter. It is a bit technical and can be skipped if you are already convinced that a calibration idea could be used to over come the divergence problem. ### Concrete Example *Simulators (at least in the sense we mean) must start with some initial conditions, and use these to determine what will happen. For instance, in the case of Gimel, the initial conditions might be the location, momentum, temperature (and so on) of all objects in the simulated region, as well as the mental states of all sentient beings on Gimel. Then using this, the simulator can determine what it thinks the next state the world will be in, which it will use to determine what the state after that will be (and so on). For this example we assume Gimel sends initial conditions to Aleph at regular intervals.* *Let $k$ denote the time it takes for a simulator on Aleph, which has just received initial conditions from Gimel, to catch up to real time. Assume that there are $s/(s-y-k)$ (rounded up) simulators on Aleph (where $s$ is the amount of time the simulator remains accurate for and $y$ is the distance in light years between the two planets). Then if Gimel sends initial conditions every $s-y-k$ years, there is a way for Aleph to always have a sufficiently accurate simulation of Gimel running.* *The basic idea is that there is a primary simulator running the simulation that is being used by Aleph. The other simulators are initially on standby. Initial conditions are received by Aleph and fed into a simulator on standby. When this simulator catches up to real time (after $k$ years) it becomes the primary simulator and the one before it is put onto stand by mode. The number of simulators was chosen so that there will always be a simulator on standby when initial conditions are recieved. The interval in which initial conditions are sent was chosen so that the primary simulator is always sufficently accurate.* *To demonstrate this let’s proceed with a concrete example. Assume $s = 10$ (the time the simulator remains sufficiently accurate for), $y=1$ (the time it takes for the data to reach Aleph from Gimel) and $k=3$ (the time it takes for a simulator to catch up to real time). This means that Gimel will send calibrating data every $6$ years and that there are another $2$ simulators in addition to $S$. Call them $S_1$ and $S_2$.* *From the time Gimel sends the first batch of calibrating data, it takes $4$ years for $S_1$ to catch up and become the new primary simulator. Since Gimel waited $6$ years to send this information, $S_1$ catches up after $S$ had been running for $10$ years, which is just in time for the swap over. Note however that at the time of the swap over, the calibrating data being used by $S_1$ is $4$ years old, so $S_1$ will only be good for another $6$ years. Thankfully Gimel sent another batch of initial conditions $6$ years after it sent the first batch and so $S_2$ will be ready to swap over just as $S_1$ stops being accurate. By the time the third batch arrives, the original simulator $S$ will be in standby and ready to receive it. The process can continue in this way indefinitely.* The above system requires that a simulator can run its simulation faster than reality, for if it could not the number $k$ would not be finite. This section demonstrates that if a simulator is not perfectly accurate, this does not mean that the mistakes will snowball until it is unusable. We are now ready to consider two planets, both simulating one another. ## Entangled Simulation Entangled simulation is the situation where Aleph simulates Beth and Beth simulates Aleph. From Aleph’s perspective the only difference between simulating Beth and simulating Gimel, is that Beth has a simulator of Aleph. Hence if there were to be some problem with Entangled simulation it must be due to the fact that the planet being simulated has a simulator of the planet doing the simulating, and indeed, such a problem arises. For most of this section we assume that the simulators used are of the form described in the last section (highly accurate but not perfect), however if we briefly assume again that the simulators are perfectly accurate we can more clearly demonstrate the problem of simulating the others simulators. In all that follows we talk primarily about Aleph’s simulator. Due to the symmetric nature of the arrangement, all the arguments that follow will equally apply to Beth’s simulator. Let $S_{\alpha}$ and $S_{\beta}$ be Aleph’s and Beth’s respective simulators and assume they can mutually simulate one another perfectly (i.e. $S_{\alpha}$ simulates $S_{\beta}$ in addition to the rest of Beth). Clearly $S_{\alpha}$ and $S_{\beta}$ use some fraction (less than one!) of their resources to simulate one another other. This gives that $S_{\alpha}$ simulates itself with only a fraction of its own resources, as it is simulating $S_{\beta}$ which is itself simulating $S_{\alpha}$. Hence the resources used by $S_{\alpha}$ can be safely reduced to this fraction. This process can be repeated, each time reducing the resources needed by each simulator. This number will go arbitrarily low. This does not seem possible and so we arrive at a contradiction. One way to resolve things might be to weaken the assumption of perfect simulators to highly accurate ones. In this case the simulation $S_{\alpha}$ has of itself will only have some percentage chance of being accurate, and what they make up for in resources, they lose in accuracy. There may be some way to make this idea work, but we shall abandon it here in favour of a method that involves not simulating the other simulator at all. The reason Aleph’s simulator appears to need a simulation of Beth’s simulator is so that it can predict how people on Beth will react after recieving information from their simulator. For instance, suppose Aleph’s simulator simulates Bethan moving to her simulator and querying it. What Bethan sees will determine her next course of action and without access to this information, Aleph’s simulator is unlikely to predict her next action correctly. However, is it true that Aleph’s simulator needs to be simulating Beth’s simulator in order to have access to this information? Since Beth’s simulator is giving information about what is (more than likely) occurring on Aleph, perhaps the information Aleph’s simulator needs can be obtained directly from Aleph itself. If Bethan looked and saw Alphonse say in her simulator: “Hi Bethan, what are you doing?”, then Aleph’s simulator need only get access to what Alphonse said to be able to continue accurately running its simulation. Alphonse could freely give this information, or there could be a recording device in the room he is in. The point being that it is not necessary for Aleph’s simulator to simulate Beth’s simulator. This is similar to the concept of bootstrapping in computer science. For this idea to work, Aleph’s simulator needs to assume that Beth’s simulator will correctly predict what happens on Aleph. This is an okay assumption because we have already assumed that both simulators are highly accurate. Below we describe a concrete implementation of two entangled simulators, that would allow communication to occur. We look at a very restrictive system where the simulator can not be perused freely but can only be used for text communication. There are many other ways this could be implemented. We describe the setup on Aleph specifically. Beth has the same set up. ### Concrete Example *There is only one room from which Aleph’s simulator can be accessed. In that room is a computer and a keyboard. The computer is running a chat program only, which allows the user to send messages to and read messages from Beth. It works in the following way.*  *If Aleph’s simulator, simulates Bethan typing a message on Beth’s corresponding computer, then it sends the contents of her message to Aleph’s computer. Alphonse on Aleph can see the message Bethan typed on the computer. If Alphonse types a message on the computer, then the contents of his message are sent to the simulator. The simulator can use this information when it simulates someone on Beth reading the messages on their computer.* *For completeness let us describe what is happening on Beth in this scenario. Bethan walks into the room and types a message onto the computer. The contents of this message are sent to Beth’s simulator. Beth’s simulator then simulates Alphonse walking into the room and looking at the computer. It then simulates Alphonse reading the message Bethan just typed (the information it just received from the computer) and sees Alphonse type a message in its simulation, the contents of which it displays on the computer in front of Bethan.* Notice that Aleph’s simulator has access to all the information it needs to simulate the people of Beth, and at no point did it need to simulate the machinations of Beth’s simulator. Also this idea can be broadened to allow for many more computers. In fact it could be made into an internet chat room that anyone could log in to, as long as Aleph’s simulator receives all the messages the people on Aleph send. A final remark on this topic concerns ensuring that entangled simulation is compatible with the sending of calibration data. If Beth is sending calibration data, it seems likely that it will have to include whatever information Beth’s simulator gave them as this is not information Aleph will have access to. Beth not simulating Aleph’s simulator and vice versa appears to have no bearing on the calibration data that must be sent. ## Wrapping Things Up So as we have shown, it is possible to communicate quickly over large distances using entangled simulation. This does not violate information not traveling faster than the speed of light as the information is already contained inside the simulators, constantly being replenished with the sending of calibration data. Of course the simulators are assumed to be extremely powerful and its not clear that computing will ever advance to a stage where these simulators are realizable. The problem of collecting and sending calibration data may be an equally difficult problem. What this article has achieved is to give a concrete condition, on which quick communication over large distances depends. ### Miscellaneous Remarks **This cannot be used to look very far into the future:** One might think that Aleph could speed the simulation up and look into Beth’s future. This can be done up until someone on Beth uses their simulator. Because recall that Aleph will not have access to the information that this person on Beth received as it is not simulating Beth’s simulator. Aleph will only have access to this information once the event witnessed on Beth actually occurs on Aleph. This entails waiting until reality catches up with reality. **Less restrictive use of the simulator:** The above idea can be used allow the simulator to be used less restrictively. If Aleph’s simulator sees that Bethan on Beth will use the simulator in ten days to observe Alphonse in his room. Then a drone or some other surveillance equipment can be prepared in advance to observe Alphonse at that moment and send the information to the simulator. In general, the more a planet allows its simulator to be used, the less privacy the people of the other planet should expect. **Weakened assumptions on the simulators power:** In the section on calibration we assume that the distance $y$ between the two planets is less than the time $s$ that the simulator remains accurate for. This condition may not be necessary. What may happen is that there would be a delay of $y-s$ years, or some function depending on this. Someone may want to investigate this further.