Some bugs with AI

[ShuFlngPu]

Ants in general sometimes ignore enemies for no reason.
Ants in general sometimes ignore orders to move for 15-45s before following thru with starting the command ordered.
Doesnt matter if its large or small amounts, game has these issues.
Ants have issues engaging nearby enemies- even if corpses of other enemies litter the ground in the area.
Ants will go socialize with queen, even if in combat or enemies attack thier sisters nearby.

[Mike]

These are deliberate - you're controlling an ant colony, not an army that reacts instantly. You don't put down control points - the marker indicates the end of a trail that the ants will walk back and forth along.

The somewhat sluggishness of them is part of making it feel like a colony or simple creatures rather than a Starcraft army.

[Spyder]

Originally posted by Mike:

These are deliberate - you're controlling an ant colony, not an army that reacts instantly. You don't put down control points - the marker indicates the end of a trail that the ants will walk back and forth along.

The somewhat sluggishness of them is part of making it feel like a colony or simple creatures rather than a Starcraft army.

Well that answers quite alot of things for me. Soooo many times there has been that one ant that just sits there and watches the rest of the ants get killed.

[ShuFlngPu]

Originally posted by Mike:

These are deliberate - you're controlling an ant colony, not an army that reacts instantly. You don't put down control points - the marker indicates the end of a trail that the ants will walk back and forth along.

The somewhat sluggishness of them is part of making it feel like a colony or simple creatures rather than a Starcraft army.

And the ants bypassing attacking enemies despite being in mortal danger and the botten ON for fighting for that group? ...
Some ants even have agression to the point they dont care what they are attacking...
Kinda uhh, falls apart there with your logic.
You also didnt answer why ants would go socialize with queen even while in combat.
(im not sure if this is a rest mechanic or a socialize mechanic because ants do both) regardless, you wouldnt just lay down your arms for a nap or cup of tea mid-combat ant or not. xD

[Buffy]

Socialize?

[Fleringe]

everythime you give an new order around 10-40% of ants close bye walk indeed back to your nest but its something of the game but never saw an ant moving out of combat while it was fighting.
they do indeed somethimes walk back to queen when they kild an enymy ant but only when you gave an new order normally when this aplies.
but thats an part of the game so you should try to calc it with wat your orders are the move in an line never 50 ants togheter always some distance.
even when you order 80 ants that were idle 20 would move around the same time the rest some time behind.
but no mather were they were so if the were in an group togheter or just all over in the nest.

[ShuFlngPu]

Originally posted by Fleringe:

everythime you give an new order around 10-40% of ants close bye walk indeed back to your nest but its something of the game but never saw an ant moving out of combat while it was fighting.
they do indeed somethimes walk back to queen when they kild an enymy ant but only when you gave an new order normally when this aplies.
but thats an part of the game so you should try to calc it with wat your orders are the move in an line never 50 ants togheter always some distance.
even when you order 80 ants that were idle 20 would move around the same time the rest some time behind.
but no mather were they were so if the were in an group togheter or just all over in the nest.

Nah, the ants will just randomly go socalize with queen. Walk to her chamber for whatever reason.
Nor did I give any new order. The trail was a long established one and during the never-ending waves of combat for campaign.
The refusal to attack was the start of Mission 2.1 the beach I think it was.
Ants will legit just walk back to base for no reason even if you give no command.

[Buffy]

They are supposed to go back and forth from the pheromone marker to the queen. This mechanic was inspired on how ants must travel from the start of a pheromone back to the nest in real life. When placing a marker underground, ants will go from the marker back to the queen, but when placing a marker above ground, ants will go from the market back to the underground nest entrance, not all the way back to the queen.

[ShuFlngPu]

lol, people are intentionally ignoring questions now.
I never askeed WHY they go back.

I asked WHY DO THEY STOP COMBAT TO RUN BACK AND FORTH?! lol
ants would NEVER stop combat just to run back to the queen and socialize for tea.

[Buffy]

It must be a bug. Personally, though, Ive never seen an ants, when engaging with enemies, that all of the sudden disengage and run back to the queen.

[Mike]

You might have combat deactivated for some of your groups which will make them only attack when they're being attacked; they won't be agressive. If they leave combat they'll carry on back to their trail duties and only enter combat again if they are directly damaged.

[Friend of Ants]

Originally posted by ShuFlngPu:

lol, people are intentionally ignoring questions now.
I never askeed WHY they go back.

I asked WHY DO THEY STOP COMBAT TO RUN BACK AND FORTH?! lol
ants would NEVER stop combat just to run back to the queen and socialize for tea.

As Mike said, combat return shouldn't actually be happening.

The trail is set up so that if >50% of the group has arrived, ants that have arrived will randomly head home IF they have not detected another job (pick up food / battle). The 15-45 second reaction is caused by the pheromone marker "sucking" ants into it on a timer - the timer used to be set up so that by the time the last ant left, the first one that left would arrive home - thereby making a perfect circuit (assuming nothing happened at the end of the trail). With the Leafcutter update, this timer was sped up (releasing more ants at the start of the command and slowing over time) to make them more reactive. It is possible (in fact much easier) to have them all react at the same time, and all stay at the marker, but we decided to use the gradual flow and auto-return to make it more dangerous to do things far from your nest with small numbers of ants - as well as making the ant responses to commands and swarm movements appear less robotic.

If your ants were in combat (or even near combat) they definitely should not have returned home so you might have found a possible bug. The only other thing that might have happened, as Mike mentioned, is that the "no combat" button was pressed - but if the ants are already in combat, toggling "no combat" doesn't effect them until their current target is dead.

We'll keep an eye out for the problem. We also know the trail system isn't perfect and there are more tweaks we are planning to test. We are considering an override to the "return home at end of trail" situation if the ant deciding to return detects a friendly ant who has detected an enemy (artificially extending their 'scan' radius by sharing info with nearby ants). They already share information in their swarming algorithm to spread out a bit before entering battle.

Also - I have noticed that on the new formicarium challenge, you can end up with ~80-90% of your ants in a big battle near the enemy nest, but then if the opponent slows their output, or if there is a moment where there are no enemies, due to the above algorithm, 30-40% of your ants will all go home at once. This is really quite frustrating, and I think the issue could be mitigated by employing a similar timed trickle to send idle ants home when there are >50% at the end of the trail. Again - non-idle ants should not be going home, I am assuming said 30-40% have all 'scanned' for enemies, found nothing, realised there is excess at the end of the trail, all gone home, and then moments later, the enemies are back but its too late. Scans are currently quite small ~3 tile radius - the larger the scan radius, the greater the CPU cost (the ant communication however, is virtually free).

Edit - if you have any other suggestions, let us know. I get the impression that you would really like to have near-instant responses to commands for the whole group, and ants returning only when necessary. It's far more likely that we will keep adding to the 'trail system' to make it feel more tactile than get rid of it. However, you are not the first person to raise these concerns, and if its a popular enough opinion, it might be possible to provide a "direct response" mode that simplifies the trail mechanics at least in freeplay.

[ShuFlngPu]

I dont mind the trail pherimone system.
I genuinly just had ants in combat run back to the queen.
I had a large stock of ants on that trail guarding my base and legit watched a single ant on the outside go into combat, attack twice then turn around and go home despite its target still being alive- it just up and went home.

I think my main issue is im used to games like starcraft and in todays era am used to AI optimization rather than a more scientific approach to the ants.
Perhaps a lil more explination in the game as to HOW the AI works in the first place might have helped with understanding in the first place... But in general this is just a differnt aproach.

Its just really feels bad when you dont understand why and watch ants die for 'no reason' xD