I recently spent some time talking about binding affinities in molecules such as DNA, RNA, and proteins. I said previously that there wasn't any differential binding affinity from one base pair to the next.
That turns out not to be the case. (aka FAIL) There _are_ base stacking interactions that can stabilize DNA and RNA molecules. These means that some sequences will be more likely than others in the real world.
The situation can be explained with an analogy to human language. This is good, because Polanyi worshipers love this analogy. In English, Q is followed by U consistently. That is a complete affinity. We can also talk about an affinity between two classes of letters, those representing consonant sounds and those representing vowel sounds. A string of letters is more likely if it contains alternations between these two classes.
Why is that? Well, in written English you'd be wrong to imagine that the choice of letter order was entirely arbitrary, it obviously isn't. (If it was, we wouldn't be able to compress English text very well, which is obviously not the case.) Written English derives from spoken English, and it is a lot easier to transition from one consonant to another through a vowel sound rather than directly. If you disagree, try the Czech phrase "Strč prst skrz krk" which translates roughly as "stick a finger in your throat".
From here, we can see that Polanyi's analogy was a FAIL to begin with, since the thing he wanted to analogize to, human language, does show constraints based on physical properties of the world and is not an arbitrary symbol system.
And Jenga!? Obviously, the choices you make in this block stacking game are not arbitrary, either. Every player knows that a stack built of "middle" bricks is going to be very unstable.