The block size limit consensus rule establishes the arbitrary trade-off between utility and system security. Increased block size marginally increases transaction throughput and therefore the resource cost of transaction validation (i.e. processing, storage, and bandwidth). As the cost of validation increases, economic security is adversely impacted by increased centralization risk. As the trade-off is arbitrary, there is no ideal size.
At any block size the system remains non-scalable due to the necessity of confirmation finality. A finite set of transactions must be selected, which implies that others may be excluded. This exclusion is financially motivated by the opportunity cost of not utilizing deployed mining capital, and is the manifestation of non-scalability. This inherent scarcity necessitates a competitive market for confirmation, and finances it in proportion to demand for the money.
Effective transaction carrying capacity, and therefore utility, can be increased by layering. This represents a local and time-limited security compromise, in contrast to the system-wide and persistent security compromise of increasing block size. Either compromise lowers but does not eliminate the utility threshold, which implies the stability property is preserved.
Therefore stability and non-scalability exist at any block size and level of layering.