Tweaked AI requirements for ascension path perks (solve issues with some galaxy settings)

Tannhäuser Cake

First Lieutenant
Nov 22, 2020
AI empires are currently scripted to not pick the psionic, genetic or mechanical ascension path perks unless they possess 10+ worlds.

The primary effect is that AI empires are prevented from picking an ascension path before they have grown big enough, even if they meet the other requirements and are ethically inclined towards them. In other words, this limits the frequency of AI ascension empires (on top of AI empires being less likely than human empires to pick an ascension path due to the AI's random selection of ascension perks). Considering how strong the ascension path perks are, the 10+ worlds requirement is an artificial dulling of the AI. I suspect this is the intended design and unlikely to change, even if I disagree with it.

A secondary effect, however, is that the 10+ worlds requirement directly ties the frequency of AI ascension empires to the habitable worlds multiplier setting and the number of guaranteed habitable worlds setting, as well as to the number of empires (relative to the galaxy size) and the number of guaranteed habitable worlds.
  1. The latter may actually be in line with the intended design. When the number of empires goes down, the expected number of worlds per empire goes up - and so does the share of ascension empires. This gives a certain stability to the prevalence of ascension empires in the game - if you reduce the number of chances at ascension, the size of each chance goes up at the same time.
  2. The habitable worlds multiplier setting has no such self-balancing effect. If the habitable worlds multiplier goes down, so does the chances of reaching 10 worlds - for every AI empire in the game (barring Void Dwellers). Similarly, if the habitable worlds multiplier is raised to a high number, AI ascension empires should become much more common.
  3. The number of guaranteed habitable worlds setting also has no such self-balancing effect. If you reduce the setting from 2 to 0, the AI will at the same time go from needing 7 new worlds to needing 9.
A direct consequence of the second and third points is that AI ascension empires become much, much rarer in game settings with low habitable worlds multipliers and few or no guaranteed habitable worlds. I play almost exclusively with x0.25 and 0 guaranteed habitable worlds, for game performance and micromanagement reasons, and I cannot recall ever having seen an AI empire ascend psionically, genetically or mechanically. I am very inclined to attribute this to the 10-world rule not taking the habitable worlds settings into account.

I see two alternative solutions, one conservative and one radical.
  • The conservative solution:
    the 10+ worlds requirement for AI empires takes these settings into account by
    1. scaling with the habitable worlds multiplier
      (5 worlds at x0.5, 10 worlds at x1.0, 20 worlds at x2.0 and so on, rounding up so that x0.25 translates to 3 worlds)
    2. subtracting any decrease of guaranteed habitable worlds from the ascension path requirement
      (no change with the default setting of 2 guaranteed worlds, -1 world with 1 guaranteed habitable world, and -2 worlds with 0 guaranteed worlds)
  • The radical solution:
    the 10+ worlds requirement for AI empires is completely removed

    (which not only solves the problem with much less effort, but may also improve AI performance)
Last edited:
  • 3Like

Tannhäuser Cake

First Lieutenant
Nov 22, 2020
I strongly agree with removing the 10-world rule. The AI doesn't need to be weakened like that, specially if one picks the higher difficulties
Especially when considering that
  • the AI picks ascension perks randomly, based on weights
  • the "Engineered Evolution" AP has the normal AI default weight of 10 (meaning that the AI considers it equal to Imperial Prerogative and Transcendent Learning)
  • the "The Flesh is Weak" AP has a default weight of 2 (for materialists it is 20)
  • the "Mind over Matter" AP has a default weight of 2 (for spiritualists it is 20), which comes on top of the already small probability of the AI getting the rare Psionic Theory technology (I believe I read somewhere that the AI does not prioritise rare technologies when choosing what to research)
At least according to the wiki (I don't have the ascension perk file handy and can't verify this right now).


Field Marshal
27 Badges
May 15, 2016
  • Stellaris: Synthetic Dawn
  • Stellaris: Nemesis
  • Stellaris: Necroids
  • Stellaris: Federations
  • Stellaris: Lithoids
  • Stellaris: Ancient Relics
  • Cities: Skylines - Campus
  • Stellaris: Megacorp
  • Cities: Skylines Industries
  • Stellaris: Distant Stars
  • Cities: Skylines - Parklife
  • Stellaris: Apocalypse
  • Stellaris: Humanoids Species Pack
  • Cities: Skylines - Green Cities
  • Ancient Space
  • Cities: Skylines - Mass Transit
  • Stellaris - Path to Destruction bundle
  • Cities: Skylines - Natural Disasters
  • Stellaris: Leviathans Story Pack
  • Stellaris: Digital Anniversary Edition
  • Stellaris: Galaxy Edition
  • Stellaris: Galaxy Edition
  • Stellaris
  • Cities: Skylines - Snowfall
  • Cities: Skylines - After Dark
  • Pillars of Eternity
  • Cities: Skylines
Radical solution seems best.

With some weird settings -- like a crowded galaxy with fewer but larger planets -- the 10-colony restriction makes even less sense.
  • 1