I had time to think while the boards were down, and a pair of threads popped out.
I don't think that the calculation is ever going to be as simple as "X markerlights per Y points of models," but this is a conversation I'd very much like to have. Depending on the size of the game, the markerlight-dependence of your units (ie. hammerheads need more markerlight support than broadsides because hammerheads have single weapon while broadsides have twin-linked weapons), and your strategy. However, I think it's possible to create a set of rough guidelines, or at least have an enlightening conversation.
Looking at the codex and the math, I think there are a few axioms we can accept to begin with:
• For every markerlight token you want to have on the board, you want at least two BS 3 markerlight.
• Higher quality markerlights (BS 4, BS 5, twin-linked) have commensurately better ratios.
• A railhead wants 3 to 4 markerlights (one or two to boost BS, two to ignore cover).
• Some units don't count as needing markerlight support depending on your list. For example, fire warriors in a mechanized list where they are driving around in devilfish do not need markerlight support - fire warriors in a walking gunline with a cadre fireblade do.
Additionally, while there is not necessarily a clear game size to number of markerlight ratio, there is probably a clear game size to number of markerlight sources ratio, because the larger the game, the more hammer units you are likely to have, and commensurately the more high value targets your opponent will present you with. You're never going to have enough markerlights to light up every target, of course, but you probably want more marker units in large games than in small games, so you have more options to blunt multiple offenses.
So, where do you fall on markerlight theory? How do you figure out how much markerlighting a list needs, and how many marker units at what game sizes?