Let's go through your question one bit at a time:
Ive read that the scientific community agrees on that the universe works in a non deterministic way.
This isn't quite correct. The correct version of what you mean here is the following: quantum mechanics* cannot be both deterministic/real and local. By "deterministic/real" we mean that every property of every object has a single well-defined value at all times, and that the behavior of an object in the past, if known to exact precision, can predict the behavior of an object in the future. By "local" we mean that changing something at a distance from an object cannot instantly affect the properties of that object. This conclusion comes from experimental results showing that Bell's Inequality is violated; Bell's Inequality assumes that quantum mechanics is both deterministic and local, and that appears to be the only assumption that we can safely throw out.**
So we can proceed down two roads: either we discard determinism, which means that some things are allowed to have non-well-defined properties, or we can discard locality, which means that objects that are far away from each other can instantly affect each other. When we do the former, we have the standard theory of quantum mechanics, the one involving objects existing as complex wavefunctions and whose properties are probabilistically determined by measurement. When we do the latter, we have what is called Bohmian mechanics, which has had far less theoretical and experimental work devoted to it, but still exists as an alternative, deterministic formulation of quantum mechanics.
However, because we have some kind of structure in the universe I assume (from a super novice perspective) that the randomnes happens within a limited number of possible outcomes.
A lack of structure and the presence of randomness have nothing to do with each other, even if you assume that there are an infinite number of possible outcomes. For example, suppose you played the following game:
- Draw three points, which we will call "vertices".
- Randomly select and draw one point inside the triangle.
- Randomly select one of the vertices.
- Draw a point halfway to the vertex you selected.
- Randomly select one of the vertices again.
- Draw a point halfway from the newest point to this vertex.
- Repeat from step 5.
The result of this game (which, at steps 1 and 2, directed you to randomly select from an uncountably infinite number of possible outcomes) is shown below:
No matter where you place your starting point inside the triangle, you always end up generating Sierpinski's Triangle, which not only has structure, it has infinitely complex structure! Here's a more complete picture of it:
Structure can arise from random processes, even from random processes with an infinite number of possible outcomes.
I also assume that these possible outcomes works on a normal distribution curve.
Not every physical probability distribution is a normal distribution. For example, if you set up a laser near a large flat wall and aim it in random directions many times, recording where the beam hit the wall each time, the distribution of hits on the wall will be a Cauchy distribution, which is definitely not a normal distribution. (The derivation of this fact is known as the "Gull's Lighthouse problem".)
why are particles distributed within a limited amount of possible outcomes and not distributed 100% randomly with no limitation to the number of possible outcomes.
In short, because the laws of physics exist.
Ultimately, if there were no constraints whatsoever on the behavior of the objects in the universe, no rules describing which configurations were more likely to follow others, then you might see the universe adopting totally (from our perspective) unrelated configurations at every instant. But the laws of physics impose constraints on the configurations that the universe can adopt, and also impose a relative weighting on different configurations that make some transitions between configurations more likely than others.
An example of a constraint is the fact that nothing can travel faster than light: a configuration in which I am holding a baseball cannot instantly follow one in which that same baseball was on the other side of the Earth. An example of an assignment of unequal weights to different configurations is present in every interaction. Interactions change the total amount of potential energy in a system based on the system's configuration, and configurations with a lower potential energy are more likely to occur. This is why electrons tend to stay bound in atoms (a configuration with a low electric potential energy) rather than floating free (a configuration with a high electric potential energy). This doesn't mean there's a probability of zero that an electron can spontaneously escape the atom; in quantum mechanics, it can happen, in a phenomenon called tunneling, but this is very unlikely, because the energy of the resulting configuration is high.
Or, why cant an elephant turn pink in the next second?
The particular meaning of the word "can't" is very important here. Usually in physics, we say that something "can't happen" when the probability of it happening is so low that one would have to wait orders of magnitude longer than the lifetime of the universe to reasonably expect it to happen somewhere in the observable universe. Having an elephant turn pink would require rearranging a huge number of pigment molecules in the skin, which would require so many individual instances of tunneling, each of which is already extremely unlikely, that we can safely put it in the "can't happen" category.
*If you're in an experimental regime where classical mechanics is sufficiently accurate, then you can sometimes safely assume that physics is both deterministic and local (a common exception is when Newtonian gravity is involved, as a change in the gravitational influence of an object on its neighbors is assumed to be transmitted instantly). However, this doesn't prevent unpredictable behavior: for example, chaotic systems can behave deterministically and even locally, but are so sensitive to initial conditions that any finite uncertainty in the measurement of initial conditions will eventually lead to an inability to predict the trajectory of the system.
**There are other assumptions behind Bell's Inequality, which led to "loopholes" in experiments claiming a violation of Bell's Inequality, but it is currently believed that all loopholes have been closed, with the first loophole-free experiment being done in 2015 by the group of Dr. Ronald Hanson.