home
Download AI_UO_0147 , the present AI release, 06-02-24.
Release
remarks for version
0.14.7
The
AI is rather peacefull and tries to mind its own
business
till the advent of aircraft. It will then
pounce on
any neighboring enemy.
Best
played
one to one on a 100 % map.
Design
point: 100%
map, 90% land
Changes:
Major speed improvements. Made
better use of the Shinkansen wonder. Improved
the enemy detection algorithm,
improved
city building, faster acquiring of technical advances. Also
some
minor
changes and bug
fixes.
On
the
average, the speed improvements have been eaten up by the
other
improvements, so do not expect much of a change in compute time.
Diplomacy
is
still swiched off.
Known
bugs
None.
Status of the
coding
diplomacy:
switched off
spy
planes:
used, results not exploited
spies:
not
implemented
slaves:
not
implemented
Diver:
not used
Glider:
not
used
conscription:
not implemeted
Hanging
gardens: not exploited
Oracle:
not
exploited
State
improvements: not utilised except for Palace, Great Wall,
Spaceport, Colosseum
Quality
Tests:
tested
against
Civ-S .101 at the design
point
with a 57: 4 result.
As
I was just
about to release this version, Steve uploaded his
new release which really wiped me. So I took the chance and
went
back to the drawing board with quite
impressive
results.
tested
against Civ-S .101 at 100%
map, 30% land
with a 2 : 8 result with 2 games a tie out of 12.
This
is not the
design point and no effort was put
into better performance for this scenario. These test have
just
been included to assure that no bugs have inadvertently
introduced.
tested
against Crystal, Civ-S, Shah, StdAI, Seti on
a 100%
map, 90% land
(6
players).
Originally,
KIAI
was included in the test, but I had to take it out because it crashed
to often to be usable in tournament mode.
All
others,
except Civ-S and StdAi, also crashed but
at an acceptable level.
The
tests were
not very conclusive due to the constant crashes. With KIAI
included, KIAI and STDAI has 2 wins each.
Without
KIAI, the wins were distributed evenly
between StdAI, Seti, AI_UO with 4 wins each.
Again,
this
is not the design point. The outcome is
mainly a result of chance, as
it very much
depends on who your neighbors are and how
aggressive they are.
Early in the game,
some wandering unit may come upon an
undefended city and so kill of competition and at the same
time
gain a city for free.
Maybe
diplomacy is the solution here - try to make peace and hope for
the best.
The
crash disaster most AI's exhibited underlines the importance of an
automated testing tool to reach some reasonable quality level.