[jeffdavis]

The problem with degrees of truth is that they aren't composable into an argument. I can chain together a handful of things that are 99% true and arrive at an absurd conclusion because the 1% is the interesting bit.

In this context, the question is whether something is political speech or something else. What is included and excluded from that category will get pretty interesting.

We saw the same thing with compelled union dues in the public sector. Were those for political speech or just routine union activity? SCOTUS said it couldn't tell the difference, so a public institution couldn't compel employees to pay dues.

I guess twitter thinks they can tell the difference, but I suspect there will be a lot of controversy.

[fabianhjr]

> The problem with degrees of truth is that they aren't composable into an argument. I can chain together a handful of things that are 99% true and arrive at an absurd conclusion because the 1% is the interesting bit.

For a general view:

https://www.wikiwand.com/en/Probabilistic_logic

Bayesian side:

https://www.wikiwand.com/en/A_priori_probability

https://www.wikiwand.com/en/Bayesian_inference (Bayesian Updating)

[dariosalvi78]

No, there are formal systems for handling uncertainty, Bayesian inference, fuzzy logic etc. Even informally we do apply similar rules all the time (though we are not free of biases)

[philipov]

> The problem with degrees of truth is that they aren't composable into an argument.

To me, this seems like less a problem with degrees of truth, and more a problem with faith in simplistic arguments.

[kazagistar]

Degrees of truth aren't some option you can choose, it's just an observation about how the world already works, where information is universally not perfect. The point is by acknowledging it, you will be more willing to update bad information and bad conclusions as you get updated information.