Author Topic: I immediately regret this decision...  (Read 1115 times)

Sphinx111

  • Newbie
  • *
  • Posts: 1
  • Karma: +0/-0
    • View Profile
I immediately regret this decision...
« on: August 12, 2012, 05:37:45 AM »
Was asked to share the screenshots from a recent game where I set the wrong tech scalings for a game against medium AI.

Please not the time difference between the pictures, this is the true meaning of exponential growth.

link: http://imgur.com/a/Uis7T

The settings used were 101 planets spiral galaxy, medium AI researcher... 1.4 gain per level, 1.4 cost per level for tech scaling. This was on standard... I couldn't see any way to kill that many ships despite having slightly superior tech...I achieved a ringworld in 55 minutes, which was then sniped by the 50M scale AI ships :-/

Lovely.

Boonestorm

  • Newbie
  • *
  • Posts: 6
  • Karma: +0/-0
    • View Profile
Re: I immediately regret this decision...
« Reply #1 on: August 13, 2012, 06:21:05 AM »
What you probably forgot was the effect of having more than one planet.
For one planet, the research time should are more or less constant when you
have the same scaling of gain and cost.
However, when you colonize more planets, you get a prefactor
on the research gains, and break the equality..
If you want fast research without breaking the game, I would recommend a ratio of
at least 1.4 to 1.7. Still, this might break down for large games.
(Of course, for high tech levels, exponantial always wins over the prefactor,
but for tech levels < 30 this is not always the case).

Cheers,
Boonestorm.


P.S:
For the more math-oriented:
cost=base_cost * cost_scale_factor^techlevel
output=output_per_lab * no_of_labs * output_scale_factor^techlevel  (calculated as output per second)
research_time=cost/output

This means that for the same scale factors, the output always wins (= the game breaks).
For close cases (almost equal scale factors) it depends on galaxy size and the number
of labs you build. However, as mentioned above, 1.4 and 1.7 should be reasonable
for fast research.