Oh, that's an interesting suggestion. It would just need to be communicated to players clearly as it might otherwise be difficult to realize.

The code that determines what is a ship class, in regard to how characters can be assigned, can perhaps be improved? It seems to consider the nature of a ship by checking the components by reverse index in the tree of components. This means, for example, that a ship with a Jump Stabilization module on which you add a very small troop component becomes a troop transport.

The most versatile approach would be to have the component added last to determine the nature of the ship; this way, a player can precisely decide what their ship should be considered. If not, then the costliest component should determine it.

Version 2.5.1, in event description:
"stabliise"

Steve is British

Second, Steve clarified later that:

Quote

The program orders the list of potential CMC sites by score and then goes through the list with a 1/3rd chance for each to be selected. Therefore the best option is not always selected and in a system with only a small number of possible sites, none of them may be selected. You are more likely to see CMC appear in a system with more potential options.

So, don't expect deterministic results.
In fact, there's a ~44% chance that the top two sites will be skipped over.
And a ~20% chance that the top four sites will be skipped over.

An angel from heaven comes bearing the word of God Steve! I have been wondering about this for so long...

I am necro-ing this thread to post some additional clarifications that might save someone else (or future me) some head scratching.

First of all, I was mistaken above about the algorithm selecting a system first, then selecting a site in that system. That logic was in VB6, but not in C#.
In C#, all qualifying sites in all systems are ranked by their score (per above), and the selection is made from that list.

Second, Steve clarified later that:

The program orders the list of potential CMC sites by score and then goes through the list with a 1/3rd chance for each to be selected. Therefore the best option is not always selected and in a system with only a small number of possible sites, none of them may be selected. You are more likely to see CMC appear in a system with more potential options.

So, don't expect deterministic results.
In fact, there's a ~44% chance that the top two sites will be skipped over.
And a ~20% chance that the top four sites will be skipped over.