There are no 2-body collisions
The title is intentionally misleading. Of course there are 2-body collisions. But perhaps many fewer than is supposed.
A physicist will name collisions by how many objects collide at once. Two atoms knocking into each other is called a 2-body collision. Three atoms coming together is called a 3-body collision. And so on. Modern physics assumes that we live in a random universe, hence there can be no pattern to atoms colliding, making the 2-body collisions the most common, while 3-body collisions are much, much less common, and 4-or-greater-body collisions can usually be ignored. The thinking is something like this: Let's say you are out walking on quiet roads, and on average you pass a car or person once every 2 minutes, going in either direction. This will be our 2-body collision. I know you do not usually collide with the car or person, but let's just say you get close enough to affect each other, that's all that really matters. How often would you expect to "collide" with two other objects (a car or person), hence making a 3-body collision? Here "collide" means that you all affect each other - maybe the oncoming car needs to slow down to let the car behind you pass first; or the car, the other person, and you are all close enough for your blood pressure to raise slightly because you instinctively know that a stupid misstep at just that moment in time would mean someone likely will be hurt. According to random statistics, we would say that if you come across a car/person on average every 2 minutes, then you should come across two car/person once every 4 minutes. The probabilities are multiplied. You would expect a 3-body interaction to happen once every 8 minutes.
I go through this example in some detail because I have been studying exactly this scenario since April 2020. I walk, walk, walk, all around my house. The roads and bike-paths are not very crowded. After a few months it started to dawn on me that the expected statistics seemed not to hold. Three body collisions were happening just about as often as 2-body, which should not be possible. I decided to get quantitative, and make some measurements. A few walks confirmed the suspicion. Some days the 3-body collisions outnumbered the 2-body. On no days were the 2-body much more frequent than other collisions. If there was a couple walking towards me, I only counted them as 1 object. Same with two cars going in the same direction, since the most common thing in the world is for the person driving behind you to think you are going too slow but be unable to pass on these narrow New England roads.
My son offered his thoughts that living systems by their very nature lead to favor the n>2 collisions. An example might be that we use traffic lights, which bunches cars together; or the heart pumps blood in batches, not continuously. That is a good starting point, since out on the walks I am only counting collisions with other people. But how does this starting point apply to the things a physicist might study, like atoms, nuclei, electrons, etc? My initial thought was to use these inverted statistics to prove or disprove that atoms are alive. But this is weak because what would I do - I would put the gas molecules into a jar, where they would be separated from any complex process, the molecules of gas would be alone, meaningless. Well, what kind of statistics do I imagine would apply then? I am forcing the particles into a state where it makes no difference how they behave, so I bet they will exhibit pure random statistics.
I am reminded of the statistics of stars, how many stars are single stars, like our Sun, and how many are double stars, how many triple. From what we can tell, most stars are multiple systems. A single star, like our Sun, appears to be less common. I do not recall ever reading how that is explained assuming random statistics. You would have to assume that stars are moving around independently of each other. In the past few years we have good evidence of just the opposite - stars move in groups, in rivers, in waves... we are only now collecting and categorizing the types of coordinated movements. This means that we cannot start by assuming stars all move independently and then look for how binary systems would form.
In our plasma experiments often see clear layers, which are very difficult for physicists to explain. Physicists can show that the layers are an allowable solution to known plasma physics; but no one has yet shown how they arise using the Vlasov-Boltzman equation, which is the fundamental starting point for plasma physics which, of course, assumes random statistics, with simple assumptions that 2-body collisions are the most likely, 3-body much less likely, etc. Makes me wonder if these higher levels of macroscopic organization require such statistics like I see on my walks, where 3-body collisions are just as likely as 2-body.