Using CatColab / ModelCollab for productive disagreement

General
1

Sometimes, it is hard to communicate your worldviews to other people, especially if they don’t have the same background, or “don’t speak the same language”.
For instance, from what I heard, the AI Safety community (communities ?) has been experiencing this issue for a while.
Thus, I wonder if tools such as ModelCollab or CatColab could be used to manage disagreement better. In this case, the goal would not be “collaborate to define a common formalized model {of the things you and I care about}”, but rather “give a more precise description of your underlying assumptions”, so that we both know what we are disagreeing on.
Also, for people who care about Bayesian updates, having ideas/hypotheses written down this way might help keep track of the changes/show how you got to your current beliefs.

This idea keeps nagging me, but I’m not sure how much sense it makes.
Any thoughts ?

1 Like
2

I think that this makes a lot of sense, and is along the lines of how I hope people end up using these tools.

Have you heard of https://www.rootclaim.com/? The trouble with rootclaim is that it doesn’t have support for rich non-parametric modeling. For instance, I can’t say something like “I believe this section of the world operates as this stochastic differential equation, and here’s my prior over the parameters, what is then my induced prior over the value of one of the variables ten days into the future.” An ideal version of CatColab would support something like this, I think.

Related is also Introduction. — ProbLog: Probabilistic Programming. I’m hoping someone tells me a nice categorical story for problog at some point, and that there can be some integration of non-parametric models into problog.

1 Like