Limits of Language

April 17, 2010

Most of the time, language is a pretty good way for one mind to communicate its thoughts to other minds. For example, if the idea of a pig pops into my mind, I can say the word pig to you. You will recognize the word “pig” and recall learning that those fat pink animals are pigs so the idea of a pig will pop into your head. Thus I will have managed to get this idea from my mind into your mind.

The success of language hinges critically on our common experiences in the outside world. The thoughts in our minds are private so we can’t just observe thoughts in other people’s minds. To communicate a thought I must invoke some common experience. In the case where I am trying to communicate the idea of a pig, this is pretty simple since “pig” is such a concrete idea that I’m sure you are familiar with. We are bath familiar with the idea of a pig through our sense data from seeing, touching, and tasting pigs in the outside world. Also we become familiar associating the word “pig” with the idea of a pig from hearing people reference them that way many times.

Many of the ideas we talk about are more abstract that the idea of a pig, but we are still able to talk about them. For example, the idea of three. I have never seen, tasted, smelled, heard, or touched a three since is not in the category of physical things that can be sensed. I have, however, heard the word “three” used to describe many sets of three elements. These repeated references have introduced the idea of three to my mind so that I can talk about it. But in this situation, again, we are only able to talk about the idea of three by referring to many sets of three in the outside world.

This need for your ideas to be anchored in the outside world can make communication about your subjective 1st person experience (aka what you feel) tricky since those thoughts are strictly private. For some feelings we can try to describe them in terms of what sorts of situations in the outside world evoke these feelings. For example, we can talk about the idea of happiness by describing the sorts of situations that make people happy like eating delicious food, lolling, accomplishing things, etc. When I try to communicate the idea of happiness, you must interpret it in terms of what happiness feels like for you since you can’t feel my happiness. Whereas when I talk about a pig, we both have access to basically the same idea of a pig. So it seems like talk of feelings is less precise than talk of more concrete objects like pigs.

There are some feelings that are harder to anchor in the common experiences in the outside world though. It seems possible, and I think it’s pretty clearly true, that there are feelings which are beyond words and cannot be communicated right now. For example, the Buddha said he felt, “Consciousness without feature, without end, luminous all around,” but after reading that I probably don’t have the same idea he was trying to communicate. Perhaps it is the case that I would instant recognize this feeling if I had experienced it before but it sounds so vague that I think there are many feelings I might falsely identify as his intended idea. His problem is that he can’t really base his description in any common experience.

I say that I think these ideas are not communicable right now because I think it may become easier to anchor these ideas in the outside world as neuroscience progresses. Perhaps in the future we will be able to refer to different feelings as very precise sets of brain states, although this might be tricky since each brain might implement a given feeling differently making it difficult to compare.

Listening to: Clocks – Coldplay

Religion: Self Fulfilling Prophecy?

March 11, 2010

This is a continuation of this string of posts.

Consider this hypothetical story. Thousands of years ago, some guy writes a book from his imagination that starts a religion. He doesn’t have any reason to believe it’s anything more than a complete work of fiction, but he wants other people to believe it so he is careful to exclude any elements from his religion that would be falsifiable. Despite this, his religion catches on and spreads around the world. For thousands of years it is a popular belief, though it is merely a product of one man’s imagination. Eventually, we reach the posthuman age and some devout follower of this religion decides to create a large number of simulations where his religion is fact.

People who find themselves in a time period his simulations cover are likely in a simulation. Of course, they would not be aware that these simulations would be created so they wouldn’t know to follow the religion.

A long time ago Blaise Pascal argued that one should believe in God basically because it is a low risk high reward situation. The potential gains of correctly believing in God, even if it is vanishingly unlikely, would outweigh the losses of being wrong. Later thinkers argued that there are an infinite number of unfalsifiable potential deities so the probability of choosing the correct deity out of an infinite set was 0. Since an arbitrarily large payout earned with 0 probability has 0 expected value, this is more of a low risk no reward situation and it doesn’t make sense to believe in God. In fact, for every deity who would reward you for certain behavior, there exists an unfalsifiable potential deity in the infinite set that would punish you for that behavior so it makes even less sense to pick a random deity to believe in.

One of the problems with this reply to Pascal’s Wager is that it relies on the idea that all potential deities are equally likely since we have no evidence that favors any of them. From this we get the 0 probability. With this religious simulation idea we have a clear path to a situation where a given deity would exist. For example we would need the religion to stay popular until the posthuman age, for humans to reach the post human age, for simulations to be possible, and for many simulations to be made simulating my time. It’s not clear that any of these probabilities are 0 or can’t all happen, thus we should be able to assign a finite probability that we are in a simulation where god is real and Pascal’s Wager would be a good reason to believe in God (assuming the benefits were sufficiently large).

This general process of creating many simulations can be used by an individual to do a lot of stuff. I will write more about this another time.

Listening to: Crown of Love – The Arcade Fire

Competing Churches

March 8, 2010

In my last post, I described this hypothetical organization called the Church of Bruce which I could use to extort people by making them unsure about which world they live in.  I just want to clarify one of my conclusions that was a bit off. Where I said you could conclude that you almost certainly live in one of the Church of Bruce (CoB) simulations, you really can only conclude that you almost certainly live in a simulation with a Church of Bruce in it. If there are other groups out there making simulations featuring Church of Bruce’s then you might be in one of theirs. And if they aren’t using the same criteria for rewarding people in the afterlife, then you would be less justified in donating to the Church of Bruce. This opens the door for competing churches.

Suppose you find yourself in a situation where there are several competing churches all promising to make as many simulations as they can in which they will reward followers of their church. In this case you are almost surely in some church’s simulation, but it is hard to tell whose. Although it is possible that you could hedge your bets by donating to all the churches, let’s assume each church stipulates that people in their simulations will only be saved if they worship their church exclusively. So you need to figure out which church (if any) offers you the best chance of salvation.

The likelihood that you are in a simulation by a specific church is the proportion of simulations (+ reality) consistent with your experiences that are made by that church. For example, if the Church of Bruce makes 75% of the simulations, then you have about a 75% chance of being in a CoB simulation and it probably makes sense to donate to them.

Intuitively, it seems like you should try to compare the current churches and try to gauge which is best poised to make the most simulations in the future. All else being equal, the bigger wealthier churches seem like the best candidates. But the problem with this logic is that if you are in a simulation, then you are observing the church in the simulation, not reality. It is possible that the situation in the simulation might correspond to the situation in reality, but is there any reason to believe this?

To think about the relationship between reality and the simulations you need to thing about the goals of the churches making these simulations. They are making the simulations to maximize people’s confidence that they are in a simulation where they should donate to the church. People can’t think they are in any simulations that are inconsistent with their experiences. So churches will want to make simulations as consistent with reality as possible and advertise their intentions of doing this in advance. So the Church of Bruce would advertise that at some point in the future it will make a ton of simulations that will be made as consistent as possible with reality up to that point.

Based on this principle that simulations should tend to be similar to reality, we can infer from our observation that a given church is powerful that it is likely powerful in reality and that we should donate to it. It isn’t necessarily true that simulations should look like reality because some of the churches might not design their simulations rationally. In general though we should expect all the churches to produce simulations that look pretty similar to reality.

Church of Bruce

February 28, 2010

I have an idea for a religion. It relates to the ideas about whether we are in a simulation and assigning likelihoods of finding yourself in different indistinguishable situations.

Suppose I started an organization today called the Church of Bruce. This church would encourage people to worship me and donate money to me and my church. It would be dedicated to the long term goal of creating as many simulations as possible. When creating simulations, one presumably has a lot of control over the contents of the simulation and my Church would seek to make simulations with the following properties:

  • Each simulation would take place in 2010 where the state of the world is largely similar to the present
  • Each simulation would have an individual named Bruce who would start a Church of Bruce in 2010 with the same goals as this Church of Bruce
  • Individuals in each simulation who worship Bruce and donate enough money to the Church would be rewarded after death by having their minds continue in a perfect afterlife aka Heaven.
  • Similarly, those who don’t worship and donate would spend their afterlife in Hell.

Now think about whether you should donate to my Church based on the information you have at your disposal. You know that you living in what seems to be the year 2010 and that I just started this Church. But this information wouldn’t let you distinguish between one of my simulations and reality. By the same logic as the simulation post, you must conclude either

  1. You almost certainly live in one of my simulations.
  2. The Church of Bruce will not actually make many simulations.
  3. We will never reach the posthuman age.

Assuming the Church makes a credible commitment to its goal of creating many simulations and it has a reasonable chance of surviving until the posthuman age, we can rule out (2).  If (1) is true, then you have good reason to donate to my Church. In fact, donating is probably a good play if you assign us any reasonable chance of reaching the posthuman age.

In general, this is an interesting trick where you can extort people if you can make a credible threat to create many simulations to lure them with these counterfactual rewards.

Listening to: I Can See For Miles – The Who

Finding Yourself

February 27, 2010

This post earlier about whether we are living in a simulation relies rather critically on this idea that our certainty that we are in a simulation should be the proportion of humans who are simulations. I kind of glossed over the rationale behind this although I think it is a bit intuitive.

I mentioned that it would be difficult to tell the difference between living in a simulation and the real world. The differences between the two is partially an engineering question of how realistic they make simulations so I’ll just assume I am unable to distinguish.

However, it is important to note that being unable to distinguish between two situations does not imply that I should be equally certain that I am in each. For example, I might be unable to distinguish between a situation where life is as it seems and one in which I am the subject of a Truman show, but it would be absurd for me to consider them equally likely possibilities. The difference here is that these are different possible worlds and they are not equally likely to be the world I’m in. Perhaps I will talk about this more another time.

In this figure, the numbers on the side are the a priori likelihood I assign to that world existing. The circles are indistinguishable situations I could find myself in along the time line of that world.

If, on the other hand, I am unable to distinguish between two situations in the same world, then I must consider them equally likely possibilities. In the previous situation the asymmetry between the worlds prevented me from considering the situations equally likely. In this case there is only one relevant world to consider. Since I am unable to distinguish between the situations, I must consider them equally likely as is the case in the following figure.

In the simulation argument all of the situations you could find yourself in occur in the same world just at very different times. The simulation situations occur occur in the future while the non simulation situations occur in the present (‘future’ and ‘present’ assume we are non simulations). So given that being a person in a simulation is indistinguishable from being a person in real life, you should consider yourself equally likely to be any non Sim as any Sim. Thus your likelihood that you are a Sim is the proportion of Sims.

This isn’t actually the best estimate you could make though because you have more information than that you are merely a person. For example, you know that the perceived year is 2010 so you can compare the number of people in 2010 to the number of Sims who think the year is 2010.

This post wasn’t really that interesting in itself, but it’s just a point I wanted to clarify because it is relevant to my next post.

Listening to: The Way I Am – Eminem


February 23, 2010

I’ve been neglecting this young blog recently cuz I’ve been a little busy with work. Also I haven’t been sure what I want to write about.

I have a new plan to abstain from reading my mighty GoogleReader. Initially I thought that GoogleReader was a big time saver by making blog posts come to me rather than having to search them out myself. This was true at the beginning because I didn’t follow many blogs, but over time I added more. Now I find that rather than saving me time, GoogleReader actually wastes more of my time than reading blogs used to. It’s not really wasting time per se since it does allow me to read more than I otherwise would. The problem is more that GoogleReader increased the value of time spent reading blogs for me which led to a greater time investment in blog reading. Funny that I’m complaining about this in a blog post lol.

I imagine the space of knowledge as a tree looking structure. The trunk of the tree holds the fundamental knowledge upon which other knowledge relies.  Then there are big branches for huge topics like Math which later split into thinner branches for the branches of mathematics, etc. These branches can split all the way down to tiny leaves. Occasionally (especially in math), distant branches will become twisted together into one gnarled mass. Of course this tree is growing over time as new facts are learned and more knowledge becomes available to be known.

The problem with blogs is that they are mainly focused on the fresh knowledge in the leaves rather than the established knowledge in the thick branches. They are useful to get a rough idea of what’s going on in different areas and to get an outline of the big branches, but the really important knowledge is in the thick branches. Basically I need to get this knowledge from books. Books ought to be incredibly effective compared with blog posts I would imagine. Blog posts kind of surround an important subject with minor conclusions while books can just blast right through and cover all the really important bits.

Might be a bit annoying to give up cuz I think I read somewhere (prolly a blog post) that checking things like google reader releases a little dopamine in your brain so I might find it hard to break the addiction. Yikes!

Listening to: Brother Ali – Breakin Dawn


January 23, 2010

This song

sounds a lot like

Although apparently this is common knowledge since even youtube commenters know it and youtube commenters are known to be among the dumbest internet users. It was still a new discovery to me. The Doors one is better imo but maybe that’s just cuz i’ve listened to it a lot of times already.

Listening to: Spanish Caravan – The Doors

Story About Being Sick

January 21, 2010

I’m sick at the moment. bah. This is unusual for me because my immune system is normally strong as an ox. In fact, my immune system was so strong that I had perfect attendance in grades 2-6. *brag* And it wasn’t even like I was just going to school sick, I just never got sick.

I remember the last time I got sick before my attendance streak began. It was in 1st grade with Mrs. Lavollo. There was a spelling test scheduled for that day and I had been studying pretty hard all week. Normally I wouldn’t studying for these types of things, that list had been particularly tricky. I remember specifically it had the words “chocolate” and “Mississippi” (lol I had to use auto correct to spell chocolate just now). The problem with my studying was that I had focused entirely on how to spell the word Mississippi and didn’t know how to spell any of the other words. Luckily the day of the test I got sick and the nurse called my dad to come pick me up so I missed the test.

I remember that afternoon I felt bad about missing the test cuz I thought it looks like a transparent plot to miss the test. So I went on our computer and typed up a letter to my teacher telling her that I felt bad missing the test since I had studied so hard. To show her that I had, in fact, studied the words, I typed out the word list in the letter using the spell checker to correct the spellings.

Are We in a Simulation?

January 15, 2010

I wrote a post about this a long time ago but I got a couple things wrong. There are a couple more thoughts I’ve had about it so I thought I’d reintroduce the idea before getting to those ideas.

The main argument is an idea by Oxford philosophy professor Nick Bostrom. I will just speed through the argument, but you can read more about it here. The argument shows that one of the following three statements is likely true.

(*1*) We will never reach the post human stage.

(*2*) Post humans are won’t run simulations.

(*3*) We are living in a simulation now.

The post human stage is just a stage of human development with extremely advanced technology by our current standards. It doesn’t refer to an era after the extinction of humans like the name suggests. Post humans will have extraordinary computing power. Some speculate that they will even build massive computers out of planets.

The simulations these post humans might run could be extremely detailed virtual worlds with conscious beings called Sims. There is some debate on this issue of whether sufficiently detailed simulated beings would be conscious, but seems likely that they would so we can assume that for now. The simulations would be realistic enough that the Sims would be unaware that they are in a simulation. One interesting fact to note is that the speed of time in the simulation doesn’t have to match the rate of time in the environment the simulation is being run in. For example, it is estimated that a planet sized computer could simulated the entire history of the human race up to this point in about 1 second using only 1 millionth of its computing power.

You might ask why these powerful post humans would bother running simulations like this. One reason might be to study them. These ancestor simulations could be used to perform controlled social science experiments. Another use for simulations would be entertainment. The computer game The Sims currently sells millions of copies and the characters are not even very realistic at the moment. It seems plausible that the game would become more popular with added realism.

Since it’s hard for people to get any clues about whether they are in a simulation or in real life, their certainty that they are in a simulation is approximately equal to the fraction of humans living in simulations. Since there are the same number of people that live before the post human era in reality and in simulations, the fraction of people in simulations is:



P=fraction of human civilizations that ever make it to post human stage from our current time

I=fraction of post human civilizations that are interested in running simulations

N=average number of simulations they run assuming they are interested

Once they have the ability to run a simulation it seems likely that they would run many copies. Think how many copies of The Sims are run today. So N is going to be very large. Unless a very small fraction of civilizations make it to the post human stage (*1*) or a very small fraction of post humans are interested in running simulations (*2*), almost all humans are simulations. If almost all humans are simulations and you cannot tell if you are a simulation or not, then you must conclude that you probably are (*3*).

Anecdote From Class

January 13, 2010

A young woman was a student in a class Freud was teaching on his theory on dream interpretation. She was skeptical of his ideas and eager to prove them wrong, but couldn’t think of how to attack them. She fell asleep studying and when she woke up she exclaimed, “Eureka!”

She marched into Freud’s office and confidently declared that she had disproved his theory.

“Vat iz it?”

“While I was napping, I dreamt that I was murdering my own mother with an ax. According to your theory, dreams are all forms of wish-fulfillment which implies that I wish to kill my mother. But I love my mother and certainly do not wish to kill her! So this evidence clearly contradicts your theory!”

“Hmm,” Freud pondered this for a moment and stroked his beard. “You vere already skeptical of des ideas veren’t you?”

“Yea I didn’t buy your theory in class,” she replied.

“Aha I see now what it is. Dis dream vas your subconscious fulfilling your wish to disprove my theory! Once again, my theory is corroborated!”


Listening to: Yankee Bayonet (I Will be Home Then) -The Decemberists