A new test for scenario designers?

[domfluff]

2 minutes ago, markshot said:

That is a good point.  I have been on "Warrior" as I found seeing my own forces was a pain, but I guess "Iron" means you can determine more easily who knows what without reverse checking off of clicking on enemy units?  If your forces are in touch, then broadly speaking they have a similar picture?

I much prefer Iron, since you get a little more information. It's particularly useful for Platoon scouts, since you can see when they leave the LOS of the platoon, and therefore how far ahead they should be.

[Combatintman]

@dbsapp - I used to play Basic Training exclusively, in fact I only made the shift to higher levels to test my stuff in a mode that the majority of people seem to play based on the comments and AARs that get posted on here.  I take your point that it can be tricky, I now play Elite mode exclusively and in many ways it can be less frustrating that Basic Training and maybe the latter should be tweaked as you say.

[chuckdyke]

11 minutes ago, markshot said:

That is a good point. 

Works only during the 'Replay Phase' you know the situational awareness of each unit. The Valentine sees nothing, the Infantry a lot more.

 

Edited May 27, 2021 by chuckdyke

[THH149]

16 hours ago, domfluff said:

As a very recent example - Between Two Fahrbahns in CM: Cold War. That's an entertaining scenario which can be played from either side. It's not massively challenging, and the decision tree is pretty shallow, but it's been a lot of fun to knock around in, and is a well crafted thing. Well worth playing at least twice, from each side.

Scenario design and then selection for publication is a fraught process with many reasons to change many variables, and hopefully the cream rises to the top as it were. I think if designers etc are aware of the different yet demanding perspectives of gamers then they can cater for the wider audience while still delivering a thrilling game that resembles conflict authentically if not wholly realisticly. Part of the appeal of any scenario can be the mastering of it to a level that is beyond merely getting a major victory, eg limiting losses to x etc.

Actually played the Fahrbahns scenario the other day several times and while it gave me a Soviet surrender each time I wasnt happy with losing so many M60s. But I've had to think about why I'd lost those vehicles and what a sound alternative approach may be, and I havent even started on the Soviet side yet, and will need to replay it again and again. That dynamic, going beyond the simple scenario result, makes the scenario worth playing twice for sure and a story to talk about here/other forums.

[Lt Bull]

I really like this line of discussion going on in this thread.  Lots of good points have been raised and provoke thought. I would say discussions like this don't happen enough, let alone is something the majority of players even think about.  I am sure some scenario designers also might not think too much about it either. I have many thoughts on the concept of what it means to make single player scenarios for games like CM where a "CPU opponent" needs to be emulated....somehow.

There are lots of things that have been said that I want to expand on, and there are things that haven't been said that I want to say to invite further discussion.  I might be off on some tangents, but in no particular order....

1) 

On 5/24/2021 at 9:39 AM, THH149 said:

Should designers have a new test of their scenario where the AI needs to win a handy victory against an immobile human defense? Could BFC use that test to qualify a scenario for inclusion in a module?

I appreciate the potential value of a "must pass a test" concept that could be used by scenario designers as a "minimum requirement" benchmark to asses whether the scenario and associated AI plans they have coded are "up to scratch".  The test as you describe it would have merit but only be relevant to a small set of possible scenarios (let alone SP scenarios). Specifically:

• single player (SP) scenarios where the human player against defends against an attacking AI
• human defender is NOT allowed to alter anything about the starting locations of ANY of his units. 

2)

I want to clearly state a thought I have (and why I think it) on the whole concept of designing CM scenarios by any scenario designer. It actually applies to any other game (typically wargames) like CM that pits "one side against another" where either:

a) one side is controlled by a human player and the other by an "AI",  referred to as the "single player (SP) experience"

OR

b) both sides are controlled by human players, referred to as a "multiplayer (MP) experience"

NOTE: I actually object to using the term "AI" as I have as it implies an "intelligence" is involved, and all the connotations we as humans associate with the concept of intelligence eg. reasoning, common sense, awareness, memory etc. I think "CPU opponent" is a better all-encompassing phrase to use as it says nothing about the actual "intelligence", and just describes what it really is trying to be, regardless of how it is being achieved.  However, I have no issue with using the terms "TacAI" and "StratAI" typically used by Battlefront to distinguish between a) the TacAI: what essentially is the hard coded local situational "survival behaviour" that applies to all units in the game (and where Battlefronts "investment in AI" really exists) and b) the StratAI: the realm of what all scenario designers get involved with when they use the Scenario Editor to develop "AI plans" limited to whatever tools Battlefront have given them in the Scenario Editor.

As far as I am concerned, there is a MASSIVE difference between trying to design/create a fun/challenging CM scenario that is:

a) designed to be played as MP

OR

b) designed to be played as SP

OR (even more challenging for the designer)

c) designed to be played both as MP or SP

One of these tasks COMPLETELY ignores and is INDEPENDANT to any knowledge or understanding of an aspect of scenario design that needs to be addressed by the other tasks: that aspect being the utilisation of Scenario Editor tools to create the "AI plans" (essentially "coding the Strat AI" for the scenario) for the SP experience.  For example, requests like the following have no bearing/relevance whatsoever on a scenario designers ability to make the best MP CM scenario:

On 5/24/2021 at 5:08 PM, Glubokii Boy said:

I really wish that we could get some more types of triggers included in the editor and not just be limited to - units in zone etc.....

Having highlighted this fundamental difference between making SP and MP scenarios, this thread probably would be more accurately titled as "A new test for SP scenario designers?".

3)

When you think about it, it certainly is interesting to consider that the ENTIRE evaluation of the "CPU opponent" in any CM "SP experience" is essentially a reflection of how good or bad a scenario designer was at being able to utilise and apply the fixed/limited set of "AI plan" creation tools Battlefront has made available in the Scenario Editor to "emulate" a credible and worthy "CPU opponent" for a particular scenario.  You could say that the TacAI also contributes somewhat to the totality of the opposition of what a player playing a SP scenario is up against, but of course this ubiquitous TacAI (common to all units on both sides)functions completely independently to whatever the scenario objectives may be.

I kind of think of the very much understated task of "coding the CPU opponent" for SP in any "one side against another" computer game like CM to be tantamount to trying to both understand something as complex as how humans would think, behave and respond to certain situations and then trying to emulate and express that via a digital representation of those situations using a bunch of computer code and algorithms, that creates a kind of believable virtual "ghost in the machine" spirit that somehow "takes control of the SP controls" and "thinks, behaves and responds" sufficiently enough to make it "human like".

Broadly speaking, involving oneself in anything that has to do with the actual CODING of the "AI code" component that forms part of the "CPU Opponent" in a SP game (typically involving an in-depth practical understanding of a complex and detailed variety of boolean logic, mathematical algorithms as well as the coding language being used) has ABSOLUTELY NOTHING to do with and is completely divorced from say a CM scenario designer (or for that matter even a game developer) from making  a brilliant scenario (or game) that is designed to be played by two human players heat to head.  I understand that a scenario (or game like CM) purely designed to be played head-to-head MP would still require aspects of "AI coding" (and all that it entails) to have been done to create the "TacAI".  But again, this "TacAI" is fundamentally different to (and serves a different function to) the kind of higher level "StratAI" which is essentially responsible for being the "substitute human player" at the controls of co-ordinating a collection of units to meet a specific scenario (or game) goal/objective, acting as the CPU opponent in a SP scenario (or game).

Consider a CM scenario designer who just wants to make a MP scenario based on some historical battle. Most would already possess a strong enough knowledge of the battle, the map and the OOBs to easily know where to access the resources they need before they jump in to the Scenario Editor. This is where their true passion and drive to create the scenario in the first place stems from.  Many would feel like they are "in their element" getting in to the research and details of the battle because they are naturally interested in the subject matter and can spend countless hours on map and OOB tweaks to get things just right.  This is evident in many scenarios (specifically the maps) I have seen which really is good to see.  If they can get some pre-release playtesting and feedback done (which I know many struggle to get) all well and good, but once the map is done, the OOBs settled and other scenario parameters decided upon (objectives, parameters etc), their work is done.  All it takes now is for two willing CM players to face off against each other on this virtual battlefield that was created.  The scenario designer "threw the party", and the two players "made it happen".

Now consider if the CM scenario designer instead wants to make it a SP scenario.  They would have to essentially do all the things I have explained above (being comfortable and well within their element to do so), but then reach a point where they would need to start "coding the CPU opponent" (for at least one side, if not both) to make the scenario playable as a SP scenario.  This task is not only so different and unrelated to anything they have had to do up till then, but is also a task which is probably the most difficult and challenging to "get right".  They literally are now involving themselves in a very specialised task that not even some dedicated wargame game developers can get right...coding the CPU opponent to "take the wheel" of one to react and behave in a worthwhile, logical and challenging way side so a human can play against it in SP.

I would hope scenario designers definitely know where their real strengths and interests lie before deciding on whether to make SP scenarios, as opposed MP only scenarios.  If they are not interested in learning and dealing with all the shortcomings of what are a very rudimentary and basic set of tools made available to them in the Scenario Editor to create AI plans so that players can have a credible/challenging SP experience against a CPU opponent, there certainly is no shame in just designing the scenario to be played MP, and not SP.

For those scenario designers who do take take on the challenge of designing SP scenarios despite the extra work and extra skillset needed, I definitely hope that Battlefront pay attention to the suggestions for enhancements/improvements in the suite AI plan creation tools in the Scenario Editor. 

Edited May 28, 2021 by Lt Bull

[chuckdyke]

35 minutes ago, Lt Bull said:

I really wish that we could get some more types of triggers included in the editor and not just be limited to - units in zone etc.....

I can't agree more, and I must say I know next to nothing about editing. I don't know what is achievable in macro managing. Can you make a 'hide' for an AFV in the editor and have him move to a hull-down once the enemy armor is closing in?

Here the scouts are 'peeling the orange' so to speak. In my Photo editor you have a 'Macro' in which you can record an action. If you had that in the editor the Tiger could move in 'Hull Down' as a trigger. 

 

Edited May 28, 2021 by chuckdyke

[Combatintman]

@chuckdykeyes you can do this in the editor but it can be tricky if you want your Tiger to go to a hull down terrain spot because you create the AI plan in 2D.  The mechanism to move the tank would be a simple trigger to have the Tiger move from its hide.  So if you want it to move to its fire position from its hide when hordes of T-34s come over the horizon then you would set an enemy armour trigger to trip at a time and space that allows your Tiger to move from its hide and get into the hull down fire position.  On firing the trigger, the Tiger moves.  You can specify an exact action spot for it to go to (your hull down position) but to get it exactly right you need to correlate that position that is obvious in 3D mode to the 2D action spot in the AI editor which is the tricky part and requires testing and adjusting frequently.

[chuckdyke]

2 minutes ago, Combatintman said:

AI editor which is the tricky part and requires testing and adjusting frequently.

Thank you, it would be like my Photo editing program. You do the testing and adjusting frequently but like a video you could record it. Next time you do some project which is similar all you have to do is apply your previous recorded action. I am afraid possibly it takes an upgraded engine to do it properly. Like record it on your 3D map and enter your recording in the 2D editor. That means editing straight in your 3D map. Kind regards. 

[Combatintman]

My favoured method is to use a gravel tile in the 2D editor where I think the Action Spot is and plot the unit's move to it.  I run the test and check LOS from that spot.  If it is, say too far to the left, I move the gravel tile to the left and replot the move in the AI.  Rinse and repeat until you have found the correct spot.  Once it is correct I then replace the gravel tile with whatever terrain is supposed to be there.  As I said, it is fiddly but it can be done.  Of course it doesn't have to be a gravel tile, it can be any type of tile that does not restrict, in this case, vehicle movement but which is distinctive and easily seen in both 2D and 3D modes.

[Lt Bull]

I will add the following....

I often consider the SP experience as simply a substitute for not having a real human opponent to play against when and where you would otherwise want them to be available at your beckon call to play against when you want to play your game.  In this respect the opponent you get in SP is typically a very poor version of the opponent you would otherwise get in MP. The opponent you have when you cant have a "real" opponent.

If you were to consider what exactly it is that a game designer (or in the case of a CM SP scenario designer) is IDEALISTICALLY trying to achieve or emulate when they work on creating a CPU opponent for a CM like wargame (or scenario), I would say that it is a bunch of code that can somehow behaves, reacts and responds to a wide variety of input in much the same way (even indistinguishably) from how a real human would behave, react and respond behind the wheels of the same controls presented with the same wide variety of inputs throughout the course of a game/scenario that has specific victory objectives.

Another way of describing what they "ideally wish they could do", the Holy Grail of "AI coding" if you will, is if they could somehow take the workings of a real human brain (or brains), ideally one that  is "wargame savvy", and somehow express all that collective intelligence and humanness in computer code such that when a game or scenario is played in SP by a real human, this cognitive virtual digital intelligence "comes to life", understanding all the intricacies of the rules and objectives of the game (no different to what the actual player understands) as it takes control of one side of the battlefield against the actual human opponent.  In theory the challenge and experience would be no different to playing a real human, thought without any of the annoying negative things that can sometimes be associated with the MP experience (in the case of PBEM, game progression based on how soon your opponent can post their turn in, delays in your opponent posting his turn in, your opponent abandoning the game for whatever reason).

I know that PRACTICALLY most game designers (or scenario designers) working on emulating a CPU opponent for games like CM get no where near anything that resembles this "idealised" concept of what I think is the Holy Grail "gold standard" for a CPU opponent.  Most are just happy to create something that can be considered "a respectable CPU opponent" with a few obvious shortcomings that the human player nevertheless still finds challenging and worthy of their time, which definitely is still appreciated by many players like myself, even though I know that what I really wish I was playing against was a real thinking human opponent.

Would like to know what others make of all this.

PS: It is worth mentioning that this "theoretical" Holy Grail of CPU opponent coding not only has been reached in a CM-like game, but has been exceeded. This CM-like "wargame" is however a very rudimentary yet deep game that pits one force of varied units against another on a battlefield based on a  set of rules, originally designed to be only played in MP mode head to head by two humans.....chess. Since 1997, the best human chess player has been no match for the best chess CPU opponent.  The current one I believe is called AlphaZero, which has also achieved a similar status for the not so wargame-like game Go.

Edited May 28, 2021 by Lt Bull

[chuckdyke]

5 minutes ago, Combatintman said:

I run the test and check LOS from that spot. 

That means jumping from 2D to 3D and back to 2D if I am correct. Kudos for your patience. 3D is for the testing and 2D for the editing. Nice if you could record the 3D testing like when we replay a turn in WeGo, once you press stop it is saved in the 2D editor. An easy way out is when the Tiger is deemed to be camouflaged and only gives his position away once he fires or moves. Spotting this position from 1km away in hazy conditions with the naked eye is a little unrealistic. 

[Combatintman]

2 minutes ago, chuckdyke said:

That means jumping from 2D to 3D and back to 2D if I am correct. Kudos for your patience. 3D is for the testing and 2D for the editing. Nice if you could record the 3D testing like when we replay a turn in WeGo, once you press stop it is saved in the 2D editor. An easy way out is when the Tiger is deemed to be camouflaged and only gives his position away once he fires or moves. Spotting this position from 1km away in hazy conditions with the naked eye is a little unrealistic. 

You can record - the only way to test something is to save it as a scenario and then 'play' it in scenario author mode so everything that you can do in a normal game, such as take screen shots or videos, you can do in scenario author mode.

[chuckdyke]

2 minutes ago, Combatintman said:

You can record

Thanks, I will copy and paste what you wrote about placing forests and trees to another member. Thanks for your attention. 

[THH149]

On 5/28/2021 at 5:39 PM, Lt Bull said:

"must pass a test"

Very indepth comments that I can't address all facets of now.

One thought I had was that the "Do Nothing and Win" test may or may not require a mandatory pass to qualify a scenario (and there could be many other requirements too).

My preference would be that its a "good to know" result (that the player can by doing nothing).  After all the designer should know what he/she is doing. But it is informative of the challenge to the player.

A couple of requirements: the setup and AI applied is that one that would ship with the scenario (and would need to make sense both in game and military simulation terms). If the player themselves change their set-up then so be it. Obvsiously, its more likely to test the attacker AI plans

It would seem to me that a key advantage of the Do Nothing and Win test is that it helps build better more challenging AI plans. One tension if that challenge is knowing that the scenario could be played multiplayer so a fix can't simply be add more attacker combat power as that would imbalance a multiplayer competition.

[THH149]

On 5/28/2021 at 5:39 PM, Lt Bull said:

this thread probably would be more accurately titled as "A new test for SP scenario designers?".

Yes it indeed could be! A challenging AI makes for a better product and better players, knowing of course that some new players would need to be carefully mentored to develop their skills rather than crash out or quit.