Most of the better points have already been made (especially Phil on hyperpolarity and Dan txdap on variance), but I’ll add my two cents.
Size matters if there are economies of scale. Economies of scale matter most in head-to-head competition in a relatively stable environment. It also allows one more capital to throw into a project. The costs of scale are that it is more difficult to make rapid changes, or to learn from the experience of others. If, as Robb suggests, we are in a world requiring quick adaptation to asymmetric challenges (economic or security) there is a maximum size beyond which states no longer make sense, and that size may be shrinking.
Instead of unitary states or multiethnic empires, the capital accumulation problem can be handled by federations, common markets, etc. Ethnic (or cultural) autonomy promotes experimentation, but a few limited common institutions allow those experiments–if successful–to be adopted, by consensus, on the larger scale. Diversity allows choices and commonalities allow mutual support. The result looks something like the alliances among airlines (for frequent flier programs and landing rights) or among microchip manufacturers (for R and D).
And has anyone been noticing what’s happening to Belgium lately? It’s not Nigeria, by any means, but it’s getting more difficult to put together a government.
So how about this vision of the future: thousands of microstates (and virtual states) linked into networks of overlapping interest. One state will be in multiple networks, depending on issues. Any pair of states could in competitive networks and in cooperative networks at the same time.
Come to think of it, we’re not so far from that today. Fragmentation within states, coupled with regional and global institutions.
I recommend you take a look at the whole thing.