[YeGoblynQueenne]

>> However, scientist rules that the window was broken by a branch because strong prior knowledge.

Prior knowledge doesn't stop you from forming new hypotheses, otherwise we wouldn't be going anywhere. The way it works is, you make an observation, you form a hypothesis to explain the observations based on prior knowledge and then you test your hypothesis with new observations. Ideally you look for observations that allow you to refute your hypothesis (because refutation is more secure than positive proof). If you can't find evidence to refute your hypothesis, you adopt it, provisionally, until such time as you have sufficient evidence to support it with good certainty, or until you have found evidence against it.

In your example, the initial observation is the broken window. The prior knowledge is that branches keep falling off and scratching the car. The hypothesis is that a branch fell and broke the car window. The second observation that tests the hypothesis is that there is no fallen branch around the car. This observation refutes the hypothesis. Then the scientist is free to go looking for additional observations and form a new hypothesis: the kids playing baseball nearby are the new observation. The next hypothesis is that they broke the window.

This process actually works very well in most situations where we have uncertainty and need an explanation, be it in the sciences or everyday world. I think you assume it's just some idealised version of "science" in the broad, but it's actually the most solid form of empirical reasoning we know of.

[hypnotortoise]

If an offender removes key elements from your observation (or in the example adding a branch beside the scientists car) you will not even consider to form a new hypothesis? That's quite convenient for those kids.