Arjun Badola

Meta updating

Today’s post talks about the podcast I came across on the internet. Title being: “Scott Aaronson on “The theorem that proves rationalists can’t disagree”. So, I would be sharing my view on the podcast and relate it to my experiences.

The podcast talks about Aumann’s agreement theorem, which basically states that two rational people could never agree to disagree.

The speaker takes a lot of assumptions and probability of those assumptions coming together seems very narrow to many people but I think these probability of all these assumptions coming together might be higher than we think. As I myself have experienced it with some people I talk to but realized it only after listening to this podcast. Assumptions:

  1. Two honest & rational people
  2. Common Priors
  3. Different background & experience
  4. People involved must have common knowledge of each other’s opinion.

Now if the above assumptions work then when these two people are having a conversation they will update their beliefs while talking.

I believe 4th point is very crucial here. As without knowing what the other person common’s holds knowledge about would make it very difficult to agree on something.

My Experience

I have a friend, let’s call him Mr. Left, who’s ideology, I am aware of, is towards politically left and I also consider him a honest and rational being, which means I value his opinion. Also, even he holds the same thinking towards me.

You guys might have noticed that whenever there is a debate between left and right political ideology it ends up into very toxic arguments. (At least this is what I have seen)

But I believe if the same thing is discussed by people who could fall under the assumptions of this theorem, the conversation would turn out to be very fruitful.

So, coming back to my example, now I have noticed that whenever I have a conversation with Mr. Left it has always ended with one of us either learning something, or pausing the conversation to continue later with relevant facts/materials required to take forward the conversation.

Now, lets try to fit Mr. Left into the assumptions required by the Anumann’s theorem.

  1. Two honest & rational people: I can say that Mr. Left is honest as I have noticed in several conversations that Mr. Left accepts when he knows he is wrong. Rationality seems subjective to me, so accordingly I consider Mr. Left rational. For example: Mr. Left has agreed on the questionability of survivability of Marxism and I, even though I don’t hold any ideology, being a investor have accepted that exploitations is, most often, one of the reasons businesses succeed.
  2. Common Priors: I believe we might be having same values. Like we both would agree to focus on creativity instead of getting good grades, as it could be our belief that creativity is a good teacher. We both could agree that buying stocks should be viewed as buying parts of a business.
  3. Different background & experiences: I met Mr. Left in my college and before that we both used to stay in completely different places therefore, we definitely have experienced life differently.
  4. People involved must have common knowledge of each other’s opinion: After spending 2 and 1/2 years in college with him I am now aware what knowledge Mr. Left holds and I can have reasonable opinion about what his opinion could be on various topics.

So, for me personally, I think the theorem fits in. Therefore, it could be the reason that it is quite rare that I ever disagree with Mr. Left or vise-versa, even if we do after few hours we end up agreeing with each other.

So, why does this happen?

The creator of the podcast, Julia Galef, has wonderfully explained in this video on why this happens, i.e. why do rational people never agree to disagree.

So, if you have seen that video and understood it then you don’t need to read further.

Julia Galef calls it Meta-Updating and divides the conversation into two levels of updating.

Level 1: Updating

Suppose I and Mr. Left are having a conversation of doing a project on Marxism and I hold an opinion that Mr. Left is rational and much more knowledgeable then me regarding this topic.

So, when I would propose an idea to work on within Marxism and Mr. Left rejects it then I would update my belief of the idea significantly towards Mr. Left’s opinion on why we should not pursue my idea.

And, if I update and agree with Mr. Left then this process would be Level 1 updating.

Level 2: Updating

Suppose I and Mr. Left are having a conversation of doing a project on finance and Mr. Left holds an opinion that I am rational and much more knowledge then him regarding this topic.

So, I would propose an idea to work on within finance and Mr. Left first rejects it and lays down his reasons.

Now I believe that Mr. Left is rational and honest this would divert me slightly from my original position, but I still believe that I might be right as the topic is related to finance. So, I stick to my idea and again propose it to him.

Now, me sticking to my idea and rejecting Mr. Left’s opinion would make Mr. Left reconsider his decision as Mr. Left thinks that I am rational and I hold good knowledge about finance. Therefore, Mr. Left will update his belief and agree to my point of view related to this project.

This type of Updating is Level 2.


Rational people keep on updating their beliefs by using other rational people around them as source of evidence.

Now, just to connect it with investing here is something Prof. Sanjay Bakshi said:

Prof.Sanjay Bakshi