>>12218887Consider the height that people on earth have as a random variable.
That random variable has a certain distribution.
Even though that distribution is exact, we don't know it with certainty if, say, we don't have access to the entire population.
Says ? is the true mean. We don't know that ? though, because we don't have access to the entire population.
We want to somehow estimate that ?.
The frequentist will estimate ? with standard stuff, like taking a sample and taking the average of that sample.
The bayesian approach would be to impose objective beliefs about ? before trying to estimate it, for example: "? is 'more likely' to be 1.75m than 1.20m".
This is achieved by treating ? as a random variable and assigning "probabilities" to its various values, thus creating a distribution for it (called the prior), and then using the sample to update the distribution of ?. After doing that you can for example pick the ? with the highest probability to estimate it.
The thing is though that ? is not a random variable, it has a fixed value that we simply don't know. The population is there, and if we could measure every person on earth, we'd know ?.
This means that probabilities assigned to ? are not really probabilities, but objective degrees of certainty that we pretty much made up, and that's where frequentist autistic screeching comes into play.
The bayesian approach "solves this" by simply treats all probabilities as degrees of certainty that can be updated and doesn't give a fuck.
Bayesian approaches are often much better, if the prior is good.
