Commutativity in double interchange semigroups (Q1633195)

The associative operations of elementary arithmetic are also commutative. However, with more advanced mathematics we find that associativity is much more deeply rooted than commutativity. Indeed, this is the basic message of the ``diagram yoga'' of category theory: the relative position of the elements is important, the order in which they are assembled is not. Double categories and semigroups extend this via the ``middle-four'' or ``interchange'' law, essentially a two-dimensional associative law. However, new problems arise. Performing one composition in a diagram can make another impossible, in a way that has no one-dimensional analog. Moreover, there is the surprising Eckmann-Hilton argument that says that, in the presence of units, the interchange law implies commutativity, and even collapses the two composition operations into one. The theory of double monoids is thus a more-or-less vacuous extension of the theory of single monoids. In [J. Homotopy Relat. Struct. 2, No. 2, 217--228 (2007; Zbl 1184.18004)], \textit{J. Kock} showed that, in a large enough diagram, similar things can happen even without units. He showed that, in a four by four grid of elements of a double interchange semigroup, the middle cells of an inside row may be transposed. Imagine a version of the 15 Puzzle in which there are no gaps, but the sixteen pieces are of foam rubber and can be stretched or compressed, so long as they stay rectangular. Some pieces have little freedom: edge pieces cannot leave their home wall, or corner pieces their corner. The four interior pieces, however, can muster enough room to change places by squeezing their neighbors. The second author and \textit{S. Madariaga} showed in [Semigroup Forum 92, No. 2, 335--360 (2016; Zbl 1335.05072)] that there were nine-element diagrams exhibiting such commutativities, and that this number is minimal. This paper considers these matters in the context of operads, and derives some interesting results. Most notably, the authors derive a number of commutativities between ten-element diagrams that, they claim, are not consequences of any nine-element commutativity. A quibble: it is not made completely explicit what it means for one commutativity to follow from another, as both are provable ab initio. Clearly it cannot be understood in the sense in which, for instance, the Hahn-Banach theorem follows from the axiom of choice. I take it to mean that one can (or in this case cannot) be derived from the other by subdividing corresponding cells, without application of the interchange law. This is a very readable paper, and it is surprising to see ``new'' commutativities arising like this from purely syntactic considerations.