Just briefly thinking about this, even simple tile assembly systems can construct structures that would seem to reach arbitrarily high scores on this metric. They can do so with just "the laws of physics and the local environment": molecular crystallization is in some sense capable of Turing universal computation, with the input being the monomers that are present, their quantities, and the temperature. These are not "lattices with some disorder in their structures". They are precisely ordered structures that can be algorithmically defined, and of defined, finite size. The 'molecules only' restriction is a bit unclear to me: do they mean 'covalent bonds only'? If so, the comparisons to structures in cells seems odd.

I would assume here that they are only considering assemblies/molecules found in nature. But in that case, the idea of an automated process trying combinations of components to see if they spontaneously form something complex, seeing that as life, seems unusual. It would be simple to choose components that would form arbitrarily complex assemblies if choosing specific components rather than choosing randomly. You could even make the formed assemblies rather surprisingly dependent upon the full vector of component concentrations as input, in addition to just the possible reaction pathways. I think very few people would consider such systems to be alive.

I would speculate that it might be better to see 'life' as a somewhat arbitrary, culturally-defined concept, without some clear, fundamental, hard line between what is 'alive' and what is not.

I realize that Cronin's work is somewhat controversial, to say the least, and I'm not familiar enough with the controversies, or his work, to discuss it in depth.