1) Keep the old mechanics. As a result, almost all bonus extensions ("farms") would be scrapped due to their insignificance. For example, what are +5 peasants? This building takes a whole day for the town and will never be constructed in a dynamic game, i.e., under normal WGE-standard conditions.
2) Modify the mechanics. As a result, absolute control over growth becomes possible. The cost will be the removal of all growth forecasts, except for the ability to right-click to see the base weekly growth value. This is a plus from a gameplay perspective, but a minus from an interface perspective (the player will have to manually account for growth modifications from fortifications, artifacts, etc.). However, the downside isn't that great—in a dynamic game, growth might not happen at all (end on week 1), or only occur once (end on week 2), making the one-time arrival a more important concept, which is displayed correctly.
I will describe the modified mechanics as I see them:
1) Levels 1-3: 50% (arrival for base building) + 50 / 100 / 150 / 200% (growth from next week: without fort, with fort, Stronghold, Castle
2) Levels 4-5: 50% + 50 / 100 / 100 / 150%
3) Levels 6-7: 100% + 100%
4) Upgrades for levels 1-7: 100% arrival for the upgrade without changing subsequent weekly growth (+ access to alternative creature versions)
5) Bonus extensions ("farms") + 100% growth
6) Fort gives +50% weekly growth to levels 1-5 (available from the start, so the actual weekly growth for these levels starts at 100%)
7) Stronghold gives (50% arrival + 50% growth) to already built creatures of levels 1-3
Castle gives (50% arrival + 50% growth) to already built creatures of levels 1-5
9) External dwellings do not provide +1 to growth
Example №1: peasants (level 1) - base growth 24.
Arrivals: for base dwelling 12 (50%), for upgrade +24 (100%), for Stronghold/Castle +12 each (50%). Total - up to 60.
Growth: for base dwelling 12 (50%), for fort/Stronghold/Castle +12 each (50%), for extension (farms) +24 (100%). Total - up to 72.
Example №2: griffins (level 4) - base growth 6.
Arrivals: for base 3 (50%), for upgrade +6 (100%), for Castle +3 (50%). Total - up to 12.
Growth: for base 3 (50%), for fort/Castle +3 each (50%). Total - up to 9.
Example №3: dragon (level 7) - base growth 1.
Arrivals: for base 1, for upgrade +1. Total - up to 2.
Growth: for base 1. Total - 1.
This system has the following advantages:
1) Lower levels, as usual, have weak and cheap growth, but if desired, they can be boosted to top levels (Stronghold/Castle)
2) Upgrades for 1-5 compete with base buildings (even those of higher levels) - they provide twice the arrival, but lose out on subsequent growth (they do not provide any)
3) Levels 6-7 (tops) stand apart and have a different arrival/growth dynamic: immediate powerful arrival (100%), but further growth is not boosted by fortifications
4) Bonus units from external dwellings do not ruin growth balance
5) Stronghold/Castle are specialized, meaning they are not mandatory universal buildings to buff everything and everyone
6) Extensions provide relevant bonuses (no "+5 peasants" syndrome)
7) Distinction between arrival and subsequent growth (different strategies for maximum arrival vs. maximum growth)
distinction between group and individual construction (for maximum effect from the Stronghold, a full group of dwellings for levels 1-3 is needed, whereas individually it is more effective to boost to higher levels, skipping the lower ones; the same applies to the Castle (levels 1-5 vs. tops)).
As a reminder, all of this will work with reduced town thresholds of 1/3/5/7/9/11 (starting at 4-5), i.e., under conditions where non-linear development is possible dynamically, without stalling while waiting for a level.