Proposed initiative system for AI

[Mandragola]

Hi all,

 

I find the alternating activations system for AI problematic. It allows big clumsy planes like marauder destroyers to out-dogfight fast jets if they simply move after them. It seriously punishes a player who brings more expensive planes than their opponent, as they'll have fewer activations, and it makes it very hard to come back from a bad start in which you take a couple of casualties.

 

This system is based on x-wing's pilot scores. The idea is that in the movement phase whichever plane has the lowest initiative score should move first, followed by the next lowest and so on. This could in theory mean one player might move all their planes before the other does, though that might encourage selecting lists with some proper fighters to avoid that result.

 

I haven't been able to test this out due to the pandemic. It would probably mess up the current points system quite a bit, though to be honest the game's points currently seem pretty wrong, charging far more for manoeuvrability than it's really worth. The points might be closer to planes' true value than before.

 

Here are the initiative scores I've ended up with.

 

 

And here's how they were calculated. I basically wasn't sure which characteristics of a plane to use, so I went for all of them. I added up the plane's maximum speed, top manoeuvre and throttle, then subtracted its handling (since a lower value is better). That resulted in some pretty high numbers so I halved them all to get a more reasonable range. And then I decided (fairly arbitrarily) to give all planes with only 2 hull a +1 bonus, to represent the fact that these tend to be the lighter fighter-type planes. There's probably a case for tinkering more with this stuff, like maybe making sure all the aces have one better initiative than the base version of the plane.

 

Overall I'm relatively happy with the result. It looks like the planes activate in roughly the order that they should, based on the fluff as I understand it. But that's a hugely subjective issue.

 

 

What do you think of using a system like this, and how I've gone about it? 

Edited June 7, 2021 by Mandragola

[Beaky Brigade]

I'd definitely like to test it out at some point. At the moment I only have one irregular opponent, and I'm not 100% confident on trying to re-invent the wheel with him, but if it's just switching from You Go I Go to initiative order it might be interesting.

The main immediate piece of feedback is that aces should probably have a higher initiative score than the stock airframe, it would give a buff to the relatively limited powers of an ace which I almost never bother taking.

Alternatively if that would be too powerful (or not give any benefit to max initiative airframes like Lightning), it could count as higher than any aircraft of the same initiative.

So for example if there were 3 aircraft with initiative 7, (2 hostile non aces and 1 friendly ace), the ace could always choose to go last rather than in the middle of the two via hostile aircraft. Does that make sense?

[Arendious]

Does it make any sense to use a straight "pilot skill" score, with a/c handling breaking ties?

 

Since AI doesn't use pilot cards, perhaps assigning a points cost for differing skill levels and Aces?

Something like:

 

Green (PS 2) 0pts

Trained (PS 4) 1pts

Experienced (PS 6) 2pts

Elite (PS 8) 4pts

[General Zodd]

Firstly, we’ll done for all the work, and big thanks for sharing! One interesting wrinkle I’d highlight is that if I were to bring just Thunderbolts, and you were to just bring Dakkajets (I’m assuming most of us have played that particular matchup) the entire Ork force would be activating after me. Now, the T’bolts might have enough stamina and raw firepower to still win through, but I suspect that would result in an Ork victory the vast majority of the time. Is that a problem? I think so? Obviously it would encourage people to bring a mix of Lightnings or Avengers, but potentially drops T’Bolts from regular play entirely, which would be a shame. I don’t have a fix, but just an observation. If you factor in Barracudas as well, then the poor T’bolt only makes sense against Fighta Bombers and actual bombers, but then, they’re not uncommon…

 

The other thing I’d say is that X-Wing originally had 9 layers of initiative and dropped it down to 6. Whether this has improved gameplay, or merely decreased the complexity of game design I’m not sure, but it instinctively feels like there’s potentially too much granularity in what you’ve come up with. The maths you’ve employed is logical and provides this range of numbers. I wonder, following your logic about adding all the numbers together and lighter planes getting a +1, rather than that last tweak, you reduced the initiative level by the number of hull points to a minimum of 1? That would reduce the top level to 6 (if my maths is correct), and bunch up some more planes into the same brackets. That feels better to me, because with X-Wing, having lower initiative actually helps in some ways due to the inflexibility of your dialled manoeuvres and the way bumping works, there is no upside to moving first in AI, irrespective of how much you’ve second guessed your opponent, there’s just too much flexibility within most manoeuvre options. This is compounded by the fact that the planes that move last will have the widest access to manoeuvres and the highest throttle, so they’re most able to react well. This might very well devolve into the PS escalation war we’ve seen at points in the (mostly 1st Ed) X-Wing meta, where if you weren’t bringing the highest PS ships possible you weren’t going to seriously compete.

 

Maybe, in order to not throw out the game balance entirely, we should keep the alternating activation, but within your own force, you have to activate from lowest to highest? That would stop me, while playing The Straggler, or a similar mission, leaving my heavily damaged bomber until last as that’s the most important plane, instead I’d have to activate the predictable lumbering behemoth first. Maybe you could decline to activate once per initiative step if your next plane is (for example) 2 higher than your opponent’s next activation? That’s probably getting far too gritty for a useful set of rules, but thought I’d throw some ideas into the mix and see what responses they spark!

[Beaky Brigade]

What about having activation order based on aircraft class without the granularity of different initiatives for different airframes within those classes?

 

Honestly I like the Keep It Simple Stupid approach of AI, and I'm not sure I'd personally like the book keeping of looking up Initiative values for each aircraft, though I have never played X Wing.

So the game is played you go I go as it is currently but...

Players must activate all bomber Class aircraft before they can activate a fighter or scout class aircraft.

So

Bomber>Fighter>Scout* (I'm putting Scouts last in this option because they are small and nimble, but also because the 2 released are mostly harmless).

Or

Bomber>Fighter and Scout

This would mean that bombers are always going to be prey for fighters to some extent, while still allowing players freedom of force selection (Assuming they don't take a heavy bomber wing without fighter escort). I would assume that bomber only lists will be much rarer than mono aircraft fighter lists which are common.

 

Edited June 7, 2021 by Beaky Brigade

[Mandragola]

Thanks everyone for the feedback. Sorry it's taken me a while to get round to replying. Real life is busy!

The main immediate piece of feedback is that aces should probably have a higher initiative score than the stock airframe, it would give a buff to the relatively limited powers of an ace which I almost never bother taking.

Something like this would make sense. It would also remove the need to calculate aces independently as I could just add one to the base plane's initiative. It could potentially also be part of the 5 point ace upgrade to represent a more skilled pilot.

 

Does it make any sense to use a straight "pilot skill" score, with a/c handling breaking ties?

Since AI doesn't use pilot cards, perhaps assigning a points cost for differing skill levels and Aces?
Something like:

Green (PS 2) 0pts
Trained (PS 4) 1pts
Experienced (PS 6) 2pts
Elite (PS 8) 4pts

This is essentially the approach X-wing takes. It's based only on pilot skill and not at all on the ship's performance. Personally I think both would be relevant. I've based my idea off information we have about the planes because we don't have much (other than the aces as mentioned above) information about pilots. 

 

Firstly, we’ll done for all the work, and big thanks for sharing! One interesting wrinkle I’d highlight is that if I were to bring just Thunderbolts, and you were to just bring Dakkajets (I’m assuming most of us have played that particular matchup) the entire Ork force would be activating after me. Now, the T’bolts might have enough stamina and raw firepower to still win through, but I suspect that would result in an Ork victory the vast majority of the time. Is that a problem? I think so? Obviously it would encourage people to bring a mix of Lightnings or Avengers, but potentially drops T’Bolts from regular play entirely, which would be a shame. I don’t have a fix, but just an observation. If you factor in Barracudas as well, then the poor T’bolt only makes sense against Fighta Bombers and actual bombers, but then, they’re not uncommon…

You're right, that could happen. It's actually the point, kind of. Under a system like this an Imperial player has a reason to bring lightnings instead of vendettas and thunderbolts. They still might want to bring some thunderbolts because they're much better for taking out enemy bombers, thanks to their far greater firepower. Things like valkyries and marauder destroyers will remain excellent at their intended role but they won't outmanoeuvre barracudas. 

 

So for example in a ground attack scenario an Imperial player might bring a bunch of marauder bombers to take out stuff on the ground, plus some lightning fighter escorts. An ork defender might take heavily loaded dakkajets and fighta bommers to try and shoot them down, but they'd be a bit vulnerable to the lightnings.

 

The other thing I’d say is that X-Wing originally had 9 layers of initiative and dropped it down to 6. Whether this has improved gameplay, or merely decreased the complexity of game design I’m not sure, but it instinctively feels like there’s potentially too much granularity in what you’ve come up with. The maths you’ve employed is logical and provides this range of numbers. I wonder, following your logic about adding all the numbers together and lighter planes getting a +1, rather than that last tweak, you reduced the initiative level by the number of hull points to a minimum of 1? That would reduce the top level to 6 (if my maths is correct), and bunch up some more planes into the same brackets. That feels better to me, because with X-Wing, having lower initiative actually helps in some ways due to the inflexibility of your dialled manoeuvres and the way bumping works, there is no upside to moving first in AI, irrespective of how much you’ve second guessed your opponent, there’s just too much flexibility within most manoeuvre options. This is compounded by the fact that the planes that move last will have the widest access to manoeuvres and the highest throttle, so they’re most able to react well. This might very well devolve into the PS escalation war we’ve seen at points in the (mostly 1st Ed) X-Wing meta, where if you weren’t bringing the highest PS ships possible you weren’t going to seriously compete.

Reducing the number of initiative bands is certainly an option. I don't think we should just deduct hull from the total as that would put too many planes on 1. 

 

Instead, you could potentially consider a few categories. Have your heavy bombers like marauders and grot bommers at the bottom. Above that maybe a category for smaller but non-agile things like valkyries, then for heavy fighters like thunderbolts and then one for thoroughbred fighters like lightnings. Aces could go in the band above their base plane's.

 

I think I prefer a more granular system than that. For one thing it means we don't have to argue about how many bands we need or which one each plane should fit in. It also allows for expansion as required, for example with extremely agile Eldar planes. That said, I do think there's a good case for looking at ways to reduce the overall number of bands, because there are so many. I'm not really sure what the best option there would be.

 

I could potentially remove manoeuvres from the calculation. They already help by giving you more options, after all.

 

In reality I think maybe having a lot of bands isn't so much of a problem as it first appears. Not all bands will be represented on the board, after all. If a player has a couple of planes in band 2, one in band 5 and two more in band 8, that's probably not more difficult in play than if we called those same bands 1, 2 and 3. It's not too elegant though.

Maybe, in order to not throw out the game balance entirely, we should keep the alternating activation, but within your own force, you have to activate from lowest to highest? That would stop me, while playing The Straggler, or a similar mission, leaving my heavily damaged bomber until last as that’s the most important plane, instead I’d have to activate the predictable lumbering behemoth first. Maybe you could decline to activate once per initiative step if your next plane is (for example) 2 higher than your opponent’s next activation? That’s probably getting far too gritty for a useful set of rules, but thought I’d throw some ideas into the mix and see what responses they spark!

 

What about having activation order based on aircraft class without the granularity of different initiatives for different airframes within those classes?

 

Honestly I like the Keep It Simple Stupid approach of AI, and I'm not sure I'd personally like the book keeping of looking up Initiative values for each aircraft, though I have never played X Wing.

So the game is played you go I go as it is currently but...

Players must activate all bomber Class aircraft before they can activate a fighter or scout class aircraft.

So

Bomber>Fighter>Scout* (I'm putting Scouts last in this option because they are small and nimble, but also because the 2 released are mostly harmless).

Or

Bomber>Fighter and Scout

This would mean that bombers are always going to be prey for fighters to some extent, while still allowing players freedom of force selection (Assuming they don't take a heavy bomber wing without fighter escort). I would assume that bomber only lists will be much rarer than mono aircraft fighter lists which are common.

 

I think you're both suggesting something similar here, but I'm afraid I disagree. Personally, I find that alternating activations really causes problems in AI, which is why I started this whole thing. I also don't think there's anything you could reasonably describe as game balance right now, so I'm not worried about breaking anything. It's definitely true that points values could be wrong under my system, but points values are already wrong. You pay for manoeuvrability that does almost nothing. A rebalancing of points costs is needed under any system.

 

The big problem with alternating activations is that there's no reason not to bring a whole fleet of tiger sharks, Furies or whatever. If my tiger sharks alternate with your lightnings or dakkajets, I win. I'm trying to design a system where the more agile planes get an advantage.

 

In my most recent game I was doing a troop landing with an imperial fleet against orks at 200 points. I had a couple of marauder destroyers, 3 valkyries, 3 vendettas and Executioner vs a load of dakkajets, fighta bommers and ground defences. I moved my destroyers first at altitude 1, aiming at the ground defences - which couldn't get out of the way. The ork player therefore had to start moving his fighters and that meant I could line up shots from vendettas and executioner on them. The result of that was absolutely brutal, so we called the game long before my valks got near the DZ. The ork fleet was shattered by the end of turn 2.

Edited June 9, 2021 by Mandragola

[Firedrake Cordova]

What about having activation order based on aircraft class without the granularity of different initiatives for different airframes within those classes?

I did wonder if something like the following was workable, in the interests of not injecting another statistic into the game:

• The movement phase is split into the following steps:
  1. Bombers
  2. Fighters
  3. Scouts
• During each of the numbered steps, only aircraft of that type may activate and the next step may not begin until all aircraft of the current type have activated.
• Within each of the steps above, aircraft must be activated in the order dictated by their Handling characteristic (6+ handling going first, 1+ handling going last). The next grouping may not begin until all aircraft in the current grouping have activated.
  • For the purposes of the above, Aces are assumed to have a handling rating of 1 better than is depicted in their profile (i.e. an Ace with a Handling characteristic of 4+ will be treated as having a Handling characteristic of 3+)

From memory, I think this gives a similar outcome in terms of ordering to Mandragola's list (I think Arvus Lighters and Remora Drones come out differently), but may (possibly?) be simpler..?

[Mandragola]

That could work. I'm not sure it's actually simpler in play. Calculating my values takes a little time but once you've got them and assigned a number to each plane I don't think it's all that hard to use. 

 

The problem with picking any one stat, as you've done for handling, is that it'll tend to penalise one faction and benefit another. So you'd end up with all the dakkajets activating before the thunderbolts, for example, which would be horrible. The benefit of including everything is that it tends, so far at least, to be fairly even across factions. Dakkajets' weak handling is compensated by their very high top speed and throttle, for example, but lightnings (equally fast but better handling) beat them.

 

I'm fairly happy with how the numbers have ended up in my system. It feels right that grot bommers and marauder destroyers should activate first, and lightnings and barracudas last. I'm glad that avengers get to go after thunderbolts, as it provides a reason to ever bring an avenger. I do think dakkajets should be after thunderbolts but fighta bommers should be before them.

 

Scouts are an oddity really. Are Arvus lighters the only one? The issue is, the Arvus doesn't look at all agile and it's probably being flown by a not especially expert cargo hauler. It doesn't make much sense to me for it to move after all the fighters. Not a huge issue but just doesn't seem quite right.

[Beaky Brigade]

Remoras are scouts too, and I guess with the Arvus it does depend on the pilot. There isn't really a rationale for Scouts last, so Bombers move, then Fighters and Scouts move would work just as well.

It would be good to try and use some of these ideas in a game, though I may not play again for a few more weeks.

[Mandragola]

I agree this stuff would need testing. Unfortunately I'm not at all in a good place to do it. Between Covid and having a 3 month old baby I don't know when I'll next be able to play.

[Rik Lightstar]

The X-Wing initiative system is a great place to start.

 

 

 

With a 1-9 scale though and the recently announced, coming SOONtm Space Marines and Eldar craft I'd suggest maybe leave yourself with some "Headroom" at the top end of the scale, from a fluff perspective within each bracket of craft there's going to be some rankings of ability based on race.

 

 

 

For example:

 

Eldar

Marine

T'au/Human

Ork

For the combination of design ethos (within the setting), racial aptitude, training & skill of pilots.

 

 

 

Then you have your initiative scores in brackets based on type, for example:

 

Light Fighters 6-9

Heavy Fighters 4-7

Light Bombers 3-6

Heavy Bombers 1-4

Marines on this basis:

 

 

 

A Xyphon an 8 for a Marine Light Fighter

 

A Storm Eagle as a Marine Heavy Fighter/Light Bomber should be a 5/6

 

A Thunderhawk would be a 3

 

 

 

Orks:

 

 

 

DakkaJet 6

 

FightaBomma 4

 

Eavy Bomma 1

 

 

 

Just some thoughts.

 

 

 

Rik

[Rik Lightstar]

Building out from this, you can obviously factor in some exceptions.

 

A T'au Drone (is it the Remora?) may well deserve to be higher ranked than a standard T'au pilot due to being piloted by an AI it doesn't have any issues with squishy bits not liking g-forces for manoeuvres nor does it experience "distractions" or "pain" in a quantifiable fashion, so possibly rate those with the Marine equivalents.

 

Likewise there may be some examples of vehicles that deserve to be dropped by a point or even an entire bracket. A Human or T'au "Civilian craft" should be rated lower than the size of the craft involved as the pilots won't be used to combat or pushing the limits of their hardware.

 

Rik

[Mandragola]

Well, looking at that system Rik I think you'd probably end up with scores pretty similar to what I have. I take your point on leaving the system open for for stuff with higher scores, but then actually there's nothing to stop a plane from having an initiative of 10 or more.

 

The main reason I've gone for a system based on stats is that the fluff is pretty inconsistent, which makes it hard to tell what a plane's performance actually should be. Here we've got some actual data about what our planes can do, so in a sense I'm not really inventing anything - just using what GW have given me.

 

I actually wonder if pilot skill is already "baked in" to the plane's stats. It ought to be, you'd think. So for example are the plane's handling and manoeuvres based on what the pilot is capable of, as well as the airframe? I honestly don't know, but it would make our lives a bit easier if it was. Otherwise you'd need to start adding initiative bonuses and penalties for the various races, at which point we're right back into making subjective calls on whether a Tau or an Ork makes a better pilot.

[Rik Lightstar]

No issue at all with a scoring system that goes above 10, although having a "hard cap" gives you an idea of where things sit in terms of the scale.

 

It's the Dr Who meme again really:

 

"Is three a lot?" 

"Out of 20, no. Out of 4 however...."

 

In the context of 40k with most things being "standardised" as far as who would fly what plane goes, at least in terms of training and experience. I'd say Pilot Skill is certainly already an aspect of the stats on each plane. That doesn't preclude having an initiative bonus for "Aces" though.

 

None of this was meant to be nit-picky, all just some ideas as food for thought as to where it could go........

 

Rik

[Mandragola]

I've now calculated the initiative scores for Marines and Eldar. Here's what I ended up with:

 

 

These are quite interesting scores I think. Eldar are, predictably, the best performers. Marines are, perhaps surprisingly, not all that awesome. In particular the Xiphon is hurt by having handling of only 4+ and not an especially high speed, meaning it ends up with a respectable, but not amazing, initiative of 6. That isn't hugely impressive when the Fire Raptor and Storm eagle are on 5 and the Thunderhawk is on 4 - better than the heavy ork and Navy planes and on a par with tiger sharks. 

 

Overall I think that's actually fine in terms of balance. The Xiphon is more or less equivalent to a Thunderbolt in terms of firepower, and costs around the same, so it makes sense its initiative is the same. It does mean the thing is a bit less super-human than the fluff tends to imply though. Maybe if Marines had something like a Storm Hawk one day it would be more equivalent to a lightning's speed and agility, bumping it up the initiative count.