Skirmish Game Design - An Ongoing, Open Discussion - 3-12-24 AI for Enemy Abilities & Powers

[Mad Robot]

Happy New Year everyone, hope you are all doing well and if you're not, I hope you are soon.

I'm working on a scenario that involves tunnels and underground rooms. How concerned should I be that players will be able to recreate what I lay out in the scenario description? I have always assumed that people will make do with whatever they happen to have for terrain. Should I just assume the same will apply in this situation?

I am thinking it shouldn't affect how I write things but I am curious about what you may think.

 

[MedMos]

Some people will get turned off if they don't have the perfect terrain. Others will make do with what's at hand, I guess. Unless it's something very involved and very specific, I wouldn't worry too much about it. A themed scenario is great for inspiring scratch built terrain.

 

[Biggle_Bear]

Like playing warhammer quest, in my opinion. If it really worries you, direct players to places they can find floor tiles or say that they should line the sides of the tunnels with actual rocks or gravel

 

[Biggle_Bear]

But sounds like a cool scenario

 

[Punktaku]

Well, I think that so long as you don’t write that the terrain HAS to look EXACTLY like what you’ve designed, most people will use whatever they have. A description of your idea and a suggested layout or example could reinforce the point they can use what they have.

 

[Biggle_Bear]

Question:

What about a campaign system in which to recruit new characters requires a payment of social currency in order to have a dice roll to find them?

(I'm sure I asked a similar question before and the consensus was that random recruitment sucks due to things like not finding guys armed how the player is capable of modelling)

I'm thinking in terms of a post-apocalyptic setting, so not to make recruitment certain but not to make it wide open to random chance.

In Necromunda terms-

I want a ganger with an autogun (worth 70 credits)

I pay half (using favour points) for a 5 or 6 on a die chance of him turning up and joining because we put the word out.

I can pay an extra 10% for a 4+ roll. Perhaps 30% for a 3+ roll.

Keeping it terms on the character's value to make higher value characters more expensive but attracting a runaway unarmed youth not hard to just pay the extra 30%.

 

[Mad Robot]

Seems like a sensible way to work it. Players could acquire social currency/favor points as they complete missions or specific tasks. If your system is using skills, there could be recruitment skills that would allow for more focused results, so if you're looking to recruit a heavy gunner you don't end up with a medic.

 

[Punktaku]

I may be a weird gamer, but I like the random generators. Maybe it’s because I enjoy the modeling/painting aspect of the hobby the most?

 

[Mad Robot]

Hello folks, another question for your consideration today. I want to devise a way to have AI controlled enemies activate special powers or abilities. I would prefer to not use a flowchart.

My initial thought is to have them triggered by an event...

"Whenever the Cult Leader is engaged in combat for the first time, he will immediately activate his leadership ability on his next turn. All friendly models within a 5" radius receive a +1 combat roll on their next activation."


Any thoughts on this approach? Any other ideas about how to implement this?

 

[Punktaku]

Isn’t a trigger just a simplified flow chart?

Seriously, though. Simple triggers would probably be a workable idea. If X happens they’ll use Special Ability.

 

[Mad Robot]

Isn’t a trigger just a simplified flow chart?

Seriously, though. Simple triggers would probably be a workable idea. If X happens they’ll use Special Ability.

This is why I like you, lol.

A flowchart would theoretically have several processes/decisions/etc. to drudge through and would, IMHO, bog down the game greatly.

These posts are really just a way to ensure I’m not too far off the mark of keeping things simple.

 

[Biggle_Bear]

I think it would work well as you proposed. So long as there is only a couple of special abilities to track.

It could be used to tell story too. For example on turn 5 they activate the missile.

 

[spafe]

depending on how you are wanting to do this in terms of play and what the players know... I'd be a fan of saying 'on a 3+ ZXZZ activates', rather than a certain trigger. That way players cannot 'know' it happens. It might be first time, might be on the 2nd... but they cant bait a power out etc with certainty

 

[MedMos]

Or maybe a progressive scale, like 4+ when x happens, then 3+ next turn etc.

 

[Galtarr]

My gaming group has recently been through a number of against the game campaigns. Blackstone fortress&Cursed City, Frosthaven and X-wing Heroes of the Altari Cluster (HotAC)

One thing we've noticed is that such games can suffer when they have either overly predictable AI, OR overly Random AI.

On the one hand - overly Predictable AI can be easily gamed against - i.e. waiting just outside a known movement range, funneled into shooting alleys etc.. on the other hand overly random enemies can alternate between advancing & retreating or taking actions that don't make any tactical sense (Frosthaven it was purely random based on card drawn)

HotAC and BSF both reacted to the player fighters positions with a small chart whilst still having a couple of options to what they might do. Opponent visible, in range X etc...

So the trigger event might be visible enemy within range X. But still with a chance of roll d6 on 1-4 do A on 5-6 do B. (D20 roll in BSF but similar principle) but you end up rolling for each model every turn.

If they always have the same fixed trigger. It becomes predictable and players will learn how/when to trigger or avoid it. Not necessarily an issue if it's a one off storyline event, or a standing power such as confers benefit to

Obviously you get game design decisions such as what level of complexity/simple & elegance you want. And simplicity does keep the game moving certainly, as much as I really liked the HotAC AI but it was a bit of a trawl through to work out what each ship would do each turn.


For reference Tie fighter AI card for HotAC:

The actions or abilities are done in a list flow chart style so you do the first action that matches the described condition. Note the HotAC campaign was fan created AI campaign/add-on for Xwing and not by games designer.

For reference BSF: AI table on right of card.
From what I remember special abilities were listed under unique actions and would happen on certain rolls when the player was in the correct column.

Ok a much longer post than yesterday if X happens then use special ability on a roll of 3-6. But that also works. I just find AI opponent design fascinating given the games I play and how we always seem to need to tweak them to be a little bit smarter etc.. TLDR balance of chance and a some reflection of battlefield condition/events is best imo.

 

[Mad Robot]

Awesome input! The BSF AI seems really concise and efficient, I will be taking a closer look at that.

It really does depend, to some degree, on the flow of the game one is playing. I am liking the very last bit where an event to have the ability trigger is somewhat tempered by the random d6 roll.

"...if X happens then use special ability on a roll of 3-6."

Seems like a reasonable balance that would still give a sense of unpredictability. It's a daunting task to try and write simple AI that will engage the player and not leave them bored.

Any thoughts on how that might translate for something more deliberate, like a spell?