[Illuminated One]

Well, I will try to monitor the usage of special units and spells.
It's very exploitable and better safe than sorry.
Also this would allow a lot more modding options (a spell that changes, depending on how often it was cast, or gets random results, etc).
This is what will keep me for a time, the researcher units are basically finished, except tweaking them.

So I will probably go along these lines (maybe I'll add the opportunity to boost research with a second shape but with a huge upkeep cost - like when you were only recruiting high level mages):

Good research (Arco, Bogarus, Marverni, Pythium, Man, Marignon, Abysia, Bandar)
Turns-----1-----2 to 11---12 to 23----24 to 35---36 to 48---49+
RP/Turn--200------80-----------140-------200-------270------350

Mediocre research (...)
Turns-----1-----2 to 11---12 to 23----24 to 35---36 to 48---49+
RP/Turn--100------60-----------120-------180-------250------340

Bad research (Niefel, Mictlan, Sauro, Pangea)
Turns-----1-----2 to 11---12 to 23----24 to 35---36 to 48---49+
RP/Turn--0------40-----------100-------165-------240------330

Similarily 3 different costs
Expensive (Pythium, MA Ermor, Caelum, Hinnom, Van, Hel, Abysia, Lanka, Bandar)

Mediocre (...)

Cheap (Mictlan, Lanka, Arco, Ulm, Jomon, Man, Agartha)


Magic 3 scales will mean +30 more rp/turn.

As you can see the distribution favors weak nations, however I tried to keep it thematically justified.

Here are my indicators
Good research - highly cultured, strong progressive movement
Bad research - barbaric, very traditionalistic or even backwards

Expensive - established, possibly decadent magocracy, money/status motiviation for research
Cheap - more down to the earth mages, altruistic or patriotic motivation

Are the differences to big/small?
Would be interesting to compare to score graphs of normal games.