View Issue Details
| ID | Project | Category | Date Submitted | Last Update | |
|---|---|---|---|---|---|
| 0009465 | AI War 1 / Classic | Suggestion - AI Behavior And Tactics | Sep 3, 2012 8:53 pm | Oct 5, 2012 2:50 pm | |
| Reporter | TechSY730 | Assigned To | keith.lamothe | ||
| Severity | minor | ||||
| Status | resolved | Resolution | fixed | ||
| Product Version | 5.076 | ||||
| Fixed in Version | 5.090 | ||||
| Summary | 0009465: Give Freed AI Ships more possible actions | ||||
| Description | Right now, freed AI ships have only one behavior: take out any stuff around if not retreating, then find a planet to attack, waiting for a good chance to strike if needed, rinse, repeat until dead. Problem is that this is a very predictable behavior, and gives no variety. As mentioned in past blog posts about lessons learned about AI war, 100% the same behavior is too easy to bait. And indeed, threat has become a grind in part because of how samey they act and how trivially easy it is to bait and manipulate them. Thus, some new possible actions for freed ships could come into play. Some new ideas I can think of are -"Unfree" itself and find something else to guard (giving preference to important planets and command stations) -Join special forces -Load itself up into a carrier guardian if one is nearby, regardless of how damaged they are If it had a chance of choosing one of these actions instead of the "wait and attack" action, that could make things a little more dynamic. Also, the "wait and attack" action itself could use a little variety too. EDIT: Feel free to post other ideas of new behavior. If they would require new mechanics or new complicated logic on the part of the AI, it may be worthwhile to post it as a new issue and relate it to this issue. | ||||
| Tags | No tags attached. | ||||
| Internal Weight | New | ||||
| related to | 0007496 | considering | Allow AI to use idle freed ships and special forces ships to aid defense. | |
| related to | 0009396 | resolved | keith.lamothe | After waiting a long time on a wormhole, threat should "reroll" which planet to stalk |
| related to | 0007016 | considering | Reduce the firepower cutoff ratio that "stalking a wormhole" AI ships will wait for before entering | |
| related to | 0009466 | resolved | keith.lamothe | Let threat attack at random sometimes |
|
|
I'm not the only one who should be credited for this idea. Several other people independently came up with this idea too at around the same time. I just can't find out who at the moment. |
|
|
Flocking to the HWs (only if the HWs are under attack) or joining hybrid fleets comes to mind. I also think threat should sometimes just decide to attack no matter what the FP evaluations say. Even if they can't kill a command center or irreplaceable structure, they could easily cause more damage this way. And sometimes they would find that they actually are strong enough. |
|
|
@Faulty Logic See: 0007016 for a variant of this idea. Part of the complexities of this sort of thing is that normal freed ships do not have "groups", each ship makes its own decisions. There is no fleet tactics in terms of "what should I group move with" or "what other ships should I wait for to arrive", or at least not in terms of unit groups. This is why sometimes you sort of see them "stream" in ineffectually. Now there are times where figuring out how to define such a group is obvious (like cross planet waves), but many other times, not so much so. EDIT: Figuring out how to figure out how to "group" units for tactical fleet management decisions from units from disparate sources (aka, not from the same source that purposely produced a group at once) is actually a very tricky thing to do in AIs. |
|
|
Bam, strong support. One of the things I thought was so cool, was when I read about the design process of the game, and the idea that everything the AI did would be more of guidelines, and never actually be predictable. I think the idea of taking that to as high of a rational level as possible would only strengthen that part of AI War, by making it a game that can surprise the unprepared player every single time. |
|
|
Or grindy, due to the difficulty in getting rid of them in an AIP-neutral fashion. (I rarely have success baiting blobs: either they don't move, or I anger the beehive and lose a planet). My last successful tactic was to send in etherjets, and drag 50 to 100 units back through the wormhole to the viral shredding machine. |
|
|
For 5.090: * Since "make the threatballs fish or cut bait" was 0000001 on the second round of the 6.0 poll: ** Threat ships behave as they used to for about 30 minutes after being freed, but after that they are switched to an alternate "Threat Fleet" behavior that is somewhat similar to the new Special Forces mechanic. *** Note: this only happens on Difficulty 7+, as it's not the AI getting anything extra, it's just behaving more intelligently (in theory) with what it has. ** If an AI homeworld or core world is under attack, the threat fleet will rally to defend it. ** Otherwise, if it sees an accessible non-AI planet with a significant human presence that it thinks it can take out, it goes to attack that (it will still pool up at the entry wormhole in some cases, until enough of them are there to pass the threshold, so they don't march in to the grinder one-by-one). ** Otherwise, it picks a planet in AI territory to hang out at until either of the two above conditions are met. ** If a carrier is spontaneously formed from ships that have a significant threat-fleet population, the carrier (and anything it spawns) is also considered threat-fleet. Not exactly what any one request was going for, iirc, but I think this will make that threat at least more interesting. Thanks for the feedback :) |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| Sep 3, 2012 8:53 pm | TechSY730 | New Issue | |
| Sep 3, 2012 8:53 pm | TechSY730 | Description Updated | |
| Sep 3, 2012 8:54 pm | TechSY730 | Relationship added | related to 0007496 |
| Sep 3, 2012 8:54 pm | TechSY730 | Relationship added | related to 0009396 |
| Sep 3, 2012 8:56 pm | TechSY730 | Note Added: 0028169 | |
| Sep 3, 2012 9:01 pm | Faulty Logic | Note Added: 0028170 | |
| Sep 3, 2012 9:09 pm | TechSY730 | Note Added: 0028171 | |
| Sep 3, 2012 9:09 pm | TechSY730 | Relationship added | related to 0007016 |
| Sep 3, 2012 9:12 pm | TechSY730 | Note Edited: 0028171 | |
| Sep 3, 2012 9:14 pm | TechSY730 | Note Edited: 0028171 | |
| Sep 3, 2012 9:19 pm | Lthesaurus | Note Added: 0028175 | |
| Sep 3, 2012 9:22 pm | TechSY730 | Description Updated | |
| Sep 3, 2012 9:23 pm | TechSY730 | Note Edited: 0028171 | |
| Sep 3, 2012 9:23 pm | TechSY730 | Note Edited: 0028171 | |
| Sep 3, 2012 9:25 pm | TechSY730 | Note Edited: 0028169 | |
| Sep 3, 2012 9:25 pm | TechSY730 | Note Edited: 0028169 | |
| Sep 3, 2012 9:34 pm | TechSY730 | Description Updated | |
| Sep 3, 2012 9:36 pm | TechSY730 | Relationship added | related to 0009466 |
| Sep 12, 2012 10:08 am | TechSY730 | Description Updated | |
| Oct 4, 2012 3:30 pm | Draco18s | Note Added: 0028590 | |
| Oct 5, 2012 2:50 pm | keith.lamothe | Internal Weight | => New |
| Oct 5, 2012 2:50 pm | keith.lamothe | Note Added: 0028597 | |
| Oct 5, 2012 2:50 pm | keith.lamothe | Status | new => resolved |
| Oct 5, 2012 2:50 pm | keith.lamothe | Fixed in Version | => 5.090 |
| Oct 5, 2012 2:50 pm | keith.lamothe | Resolution | open => fixed |
| Oct 5, 2012 2:50 pm | keith.lamothe | Assigned To | => keith.lamothe |
