AI War II: Design Document First Unveiling Is Now!

[NickAragua]

Well, what's the point of AIP? It's basically to determine the level of opposition (in terms of offensive and defensive capability) that a player faces.

The thing is, that there are already about ten other different "gauges". You've got "hacking progress", you've got whatever it is that drives counterattacks for the spire expansion, you've got something that spawns a bunch of jerks in response to you having a champion. The challenge here (in my opinion) isn't "splitting" AIP, it's clearly communicating to the player what AI behaviors are engaged, at what level and in response to what. In AIW, there's quite a bit of it that's kind of nebulous (and maybe intentionally so, but it doesn't help a player to understand why there's suddenly a swarm of large capital ships pounding on their command center).

[PokerChen]

Hmm.. I think AIP itself was too authoritative, and oversimplified how players engaged with the AI. The strategy discussions we used to have, had a lot of focus on keeping a single number below certain thresholds ( for non Fallen Spire) - it really didn't matter what exactly we did or did not do as long as the total sum did not exceed X before we were prepared for it.

It all boiled down to the equivalent of shopping on a budget with AIP reducers being coupons you'd have to fill out surveys for. In that context, splitting the AIP wouldn't have helped at all (turning your money into UN food stamps versus clothing stamps).

Hence I was advocating its complete removal and replacement in the other thread. It's a medium risk operation IMO, but this is one of those risks I would definitely  take a gamble on because reasons.

[eRe4s3r]

The big problem with AIP with stepping behind it is that beyond a certain level (whatever makes MK3 planets to core worlds) it is essentially game over if by that point you haven't killed all the shields and at least 1 homeworld + survived the response. Especially since this applies to ALL worlds, which if you hit the first "stepping" of AIP, can lead to the fastest game over ever, as you could face core ships in actual regular attack waves before you even have more than 1 MK3 ship unlocked.

####

Btw, I hope we get some deeper scripting ability than just XML for modding.. the game begs for LUA integration with hashing applied so that mods can work together in MP.. don't hardcode your campaign core logic if at all possible and allow us to add dialog boxes and if/else triggers + some more logic based scripting (count ships in planet X, do Y if number is Z, display popup box, modal dialog with image AAA and then play sound file whatever*)... it was already what killed modding in many previous Unity games not just by Arcen. This is especially important as the Unity engine is a huge detriment to modding if you don't expose core API elements. (ie, allow us to sideload DLL's with decent documentation) ala KSP.

If you do not plan on adding scripting beyond XML, then at least consider allowing sideloading of DLL's that interact with core game logic and can extend various things, like a script extender (which in turn allowed sideloading DLL's into Skyrim. It would require good documentation and a build guide, but it would also allow for vastly better and more awesome mods, especially if the game gains any mass market traction

Ps.: Though it obviously requires you to add code based triggers for advanced scripting as the one I exemplified above ;P

[Mánagarmr]

Working my way through this document still. Will come with feedback once I've read it.

[Toranth]

Btw, I hope we get some deeper scripting ability than just XML for modding.. the game begs for LUA integration with hashing applied so that mods can work together in MP.. don't hardcode your campaign core logic if at all possible and allow us to add dialog boxes and if/else triggers + some more logic based scripting (count ships in planet X, do Y if number is Z, display popup box, modal dialog with image AAA and then play sound file whatever*)... it was already what killed modding in many previous Unity games not just by Arcen. This is especially important as the Unity engine is a huge detriment to modding if you don't expose core API elements. (ie, allow us to sideload DLL's with decent documentation) ala KSP.

If you do not plan on adding scripting beyond XML, then at least consider allowing sideloading of DLL's that interact with core game logic and can extend various things, like a script extender (which in turn allowed sideloading DLL's into Skyrim. It would require good documentation and a build guide, but it would also allow for vastly better and more awesome mods, especially if the game gains any mass market traction

Sad panda:

No Scripting Abilities In Modding, At Least In The 1.0 Spec
We highly doubt we will do any scripting ability at all (ie like the python scripting that drove much of the AI and so on in Civ 4 or SupCom 1).  That means unfortunately no creating AI logic from whole cloth as part of the modding, by its very nature.

[Mac]

"The players would be able to see this currency of the AI (the "reinforcement currency"), probably?  At least in a tooltip of AIP, most likely."

  I would like the option to turn this off in the game lobby settings, or to have a different way this happens.  I just believe that it would be slightly unrealistic for us to be able to see the AI's budget at all times. (mabey have the currency as some type of intel that the scouts or something similar can discover?)

[tadrinth]

Sad panda

In a different thread, he said that the tutorial is almost entirely data and would therefore be quite customizable by modders.  So, it sounds like hooks for triggering popups will be available in some form. Does that help at all?

[chemical_art]

"The players would be able to see this currency of the AI (the "reinforcement currency"), probably?  At least in a tooltip of AIP, most likely."

  I would like the option to turn this off in the game lobby settings, or to have a different way this happens.  I just believe that it would be slightly unrealistic for us to be able to see the AI's budget at all times. (mabey have the currency as some type of intel that the scouts or something similar can discover?)

I would be fine with the "reinforcement" numbers being something only scouts could discover.

By that measure, "scouts" could be renamed as "spys" in that they can discover unique data.

[eRe4s3r]

Sad panda

In a different thread, he said that the tutorial is almost entirely data and would therefore be quite customizable by modders.  So, it sounds like hooks for triggering popups will be available in some form. Does that help at all?

Modding (With Unity 5 anyway) can only happen with DLL sideloading which is the way to do "external" lua scripting, that's the only way you can do things "beyond" the design scope of a game due to how the Unity API and memory management works...)

We'll have to see how it works in detail anyway... popups would be a good start, but you'd need some kind of data structure to store (read/write/save/load) variables. It doesn't mean changing unit stats isn't nice, but the real modding only starts when we can fundamentally alter AI responses, add complex scripting for background calculations beyond the scope of the game etc and new weapon systems that behave in new ways.

The thing is I feel like AI War 1 already missed the modding train so vastly that it hurts. It's gonna take some really well done documentation and utilities from the get-go to get a real modding community going. And obviously Nexus and NOT JUST STEAMWORKSHOP

Steam Workshop is modding DRM ^^

[GiftGruen]

I am not entirely sure about the squads, particularly the part about ships being reconstructed by engineers on repair: wouldn't that mean you can only construct a squad of fighters as a full squad and not individual units? That would make it seem like you just cut down on the number of ships, when that impression is exactly what you want to counter. Why not make every fighter a squad of its own on creation, but let them consolidate into as few as possible on every move order if the number of squads is too high (an iteration over every squad's number of ships, one division per ship type in the selection, and addition) or individually if they happen to collide with another squad that still has enough space for them? That would mean that you already have 10-squads on your fighters if you just send them all to a rally point (due to collision) or give them orders when they are on their way there (due to move order). And it doesn't seem like a lot more computation than the whole move order itself and all its position offsets etc.

[PokerChen]

Yeah um, you should also read the design doc on how squads are animated. Chris's intention is to give some freedom to squad members so they will look semi independent and reassemble like a bunch of ships (possibly decreasing as the squad takes damage).
The second thing is, you might not know this, but dynamically splitting and merging entities like you suggest is superfluous - because the computer then has to constantly update who is leading and who isn't, decide when to switch over, etc. Note that Classic already has a system where units steal movement and targeting information from others wherever possible, which is similar but minutes the banding logic.

[GiftGruen]

I did read that document, but I did not suggest being able to split squads unless to immediately give over all the squad members to other squads. You do not have to keep track of individual ships as anything else than an integer in the squad data with this approach. Example:

You have the following fighter squads:

 - 1   4/10 squad
 - 3   10/10 squads
 - 16 1/10 squads
 - 8   8/10 squads

When you select these 28 squads and give them a move order, the following happens:

The move command iterates over all squads in the selection (which it would have to do already anyway), and counts 1*4+3*10+16*1+8*8 = 114 ships, but more than 12 squads. It then selects one of the squads (either random order or ideally the one with the least amount of units in them) and gives all its members to other squads and repeats this a total of 28 - 12 = 16 times. With 150 1/10 squads, it would be 150-15=135 passes, or generally O(n).

With squads just being able to replenish lost members with only the use of engineers, you not only paint yourself in a corner lore-wise (we need to pilot our ships manually because AI is against us, but engineers can just assemble a new ship and add it to a group, can they assemble a pilot?), you also lose a lot of the "individual units packed together" feel when you never actually see them being packed together. They just feel like a big unit with comparable stats to many other units when they can just be repaired back to full health, and having a unit that loses permanent health at certain cutoffs (100% -> 67% -> only 70%) without reinforcements also makes them behave a lot more like the individual fighters from Classic: Brought 200 fighters in transports, but 60 of them died in the battle? You now only have 70% of your original firepower even if you brought engineers to repair the ones that survived.

[Cinth]

That's all assuming you can repair to that degree in the field. 

[GiftGruen]

When a squad is being repaired, how are further ships reconstructed in it?  (Presumably they are reconstructed by engineers during repair of the squad to full health -- anything else gets tedious fast for many reasons easily seen in other RTS games.)

I was referring to this part, Cinth, and tried to come up with a different solution. And if you can't repair squads to the point where you would get new ships when they're out in the field, great. If you don't like the consolidation idea (even though it would decrease the number of actual combatants in any follow-up fight, which is the same intent as the one behind the whole squad change), fine. And I can easily overlook the "how does the pilot get there" issue when it's on a controlled planet. It's just not shown, but they can be assumed to travel there from one of your buildings or the planet or whatever.

And only having full squads available to be built is probably also no big deal, if you even plan to do that, even though it feels like the game loses some variety at that point, because they're just more of the big ships, with a unique look and mechanics, sure, but every other big ship has those too.

[Cinth]

I was talking about replacing lost ships outside of your area of control.  There's still some big unknowns about how some of this stuff is going to work out.

Consolidating just feels like it's complicating things for relatively minor gains in the end.