1 |
The old hard/brutal didn't really make eco outside of a fixed trigger that was often ignored, they just boil out units. This is a good playstyle vs other AI but humans will reliably just defend, eco up, kill with an excellent attrition ratio and gradually roll it up across the map.
|
1 |
The old hard/brutal didn't really make eco outside of a fixed trigger that was often ignored, they just boil out units. This is a good playstyle vs other AI but humans will reliably just defend, eco up, kill with an excellent attrition ratio and gradually roll it up across the map.
|
2 |
\n
|
2 |
\n
|
3 |
> Overfitted against kgb-13 playstyle?
|
3 |
> Overfitted against kgb-13 playstyle?
|
4 |
\n
|
4 |
\n
|
5 |
certainly a factor, that was my primary source of testing vs raiding and aggression, but it was tested against variations of itself too.
|
5 |
certainly a factor, that was my primary source of testing vs raiding and aggression, but it was tested against variations of itself too.
|
6 |
\n
|
6 |
\n
|
7 |
Investing in eco is more of a versus humans measure, since humans have and usually follow the concept of stabilising a front and focusing on attrition. Bots tend to focus on raiding and attacks over everything else. A focus on eco means that the threat the AI poses continually increases through the game if it isn't continually raided (and most humans don't continually raid ).
|
7 |
Investing in eco is more of a versus humans measure, since humans have and usually follow the concept of stabilising a front and focusing on attrition. Bots tend to focus on raiding and attacks over everything else. A focus on eco means that the threat the AI poses continually increases through the game if it isn't continually raided (and most humans don't continually raid ).
|
8 |
\n
|
8 |
\n
|
9 |
You
can
make
the
AI
competitive
with
its
older
iterations
again
but
changing
the
number
of
preventative
defences
it
makes
per
mex
back
down
to
1
(
currently
2)
and
restoring
its
old
habit
of
turbo-spamming
raiders
all
game,
but
that
doesn't
make
for
a
good
opponent
to
a
human
team.
|
9 |
You
can
make
the
AI
competitive
with
its
older
iterations
again
by
changing
the
number
of
preventative
defences
it
makes
per
mex
back
down
to
1
(
currently
2)
and
restoring
its
old
habit
of
turbo-spamming
raiders
all
game,
but
that
doesn't
make
for
a
good
opponent
to
a
human
team.
|
|
|
10 |
\n
|
|
|
11 |
Obviously it would be preferable that the AI could have more than one set of configurations so its behaviour could vary, but Lamer doesn't seem interested in this approach.
|
10 |
\n
|
12 |
\n
|
11 |
\n
|
13 |
\n
|