Smart Combat, Unit AI - Why not?

What on earth do you mean? Obviously saying that even top players never micro is a ridiculous statement, so what do you really mean by this?

SC:FA replays don't show much of what a player is really doing, it's like watching a puppet show without seeing the puppeteers, and then saying they don't really do much.

But despite that i suggest you watch a good Air player on a Setons replay, then jump in game and try to get your ASFs to blob up like that, if you can even manage to successfully do it it won't be without prodigous amounts of micro.

What i mean?

1700 vs 1400: he flies his gunships ahead of his interceptors. He doesnt retreat them when enemy interceptors arrive. He loses all gunships.

1200 vs 1500: he stays his commander in the midst of arriving waves of t1 chaff when nothing prevents his retreat. He loses his commander.

Thats what i mean. These replays may be still available on FAF, go watch yourself.

My dear, dear man; i sincerelly appreciate your suggestion that Real Time Strategy should not be about strategy and war, but fun and EXPLOOOOOSIONS!!!!!

For making my day, i will offer you this: Defence Grid. Robots, explosions, and they even not fight back.

Also i would like to remark that if i had you in my yet nonexistent army, i would have arranged immediate meeting between you and the firing squad.

Also many thanks for referral to Zero-K; never would have found it on my own, will try it immediately.

I personally don't like Unit AI, I really feel that the players have to do these kind of things themselves. We can take all aspects of tactics/strategy out of the players hands, but if we start here, where will we stop?

I agree but it's also important to note economic micromanagement can add tedium. Forged Alliance is a great game but suffered from excess economic micro.

http://en.wikipedia.org/wiki/Macromanagement

We will likely see strategy games focused entirely on macro, while AI handles the specifics of base building and battles - but that won't be Planetary Annihilation, slippery slope or not.

A robot can move units back and forth between 2 points - that's why we have patrol commands. A robot can also kite (a lot better than a human), and jinx and strafe. Unit AI should shift the focus from individual engagements to entire battles but it should not substitute focus. Let the computer worry about repetitive move commands, let me worry about managing battles.

Unit AI is only supplementary to actual decision making, e.g. "do I pursue my opponent into a possible ambush, or do I instead coalesce my forces and plan out my next move".

Your concern is real moranic and this post covers the subject in greater detail:
viewtopic.php?f=61&t=46037&start=29

What worry me a bit is if Uber don't like unit AI, they will probably more tend to restrict micro with game mechanisms, like make repairing cost a lot so the player won't need to withdraw the damaged units, make projectiles really fast compare to the unit speeds so the player won't need to dodge them etc, and that will make combat less dynamic and interesting.

My main problem with unit AI is that you're no longer actually playing them. You're telling the AI to go fight some units in the distance. And they will automagically do so. I think that's a bit easy. Let the players fight their battles. We don't want PA to become a War management game. That's the point. IRL soldiers don't move on their own. They have to wait for the commander's word. Also because there is a random terrain gen, the AI can have problems with the landscape. I would not even bother moving rocket launchers on a cliff when the cliff is being shot. The AI might see this as an attack and start microïng the rocket launchers. That's unnecessary and it takes up CPU power.

I think we should let Uber first build their game. Then we can add our own personal suggestions via mods. Let's keep unit AI as a mod so that everyone can be happy.

That's a lot of assumptions about something that you can see for yourself in other games
http://springrts.com/wiki/Games

Here's the post I linked you, all of your specific points were addressed.

But they do. That's entire purpose of military hierarchy, generals dictate strategy, commanders dictate battle and troops dictate engagement. This type of AI isn't new though, in StarCraft when an air unit attacks a ground unit without anti-air that ground unit runs away. Arguably that's more intrusive than what's suggested here which is simply the maximization of unit potential. Increasing the potential value of units has many benefits, besides minimizing the value of micromanagement it allows for a richer and more dynamic balance. Suddenly you can balance around characteristics like agility and projectile speed without balance breaking down as soon as someone with high APM plays.

Indeed, maximizing unit potential only makes balance more dynamic and opens up more possibilities. Without such a system characteristics like projectile speed, acceleration and ROF will either be ignored or insignificant. I should note though, nobody wants to be memorizing stats, units should have self explanatory roles e.g. assault, raid, artillery, skirmishers while maintaining a rich variety of tactics.

I'd say its' not enough if tactics like kiting and projectile dodging is prominent in PA gameplay. Kiting doesn't scale well on multiple battlefields if it have to be done manually as it would put a huge strain on the players attention and micromanagement.

In Forged Alliance kiting and projectile dodging wasn't that prominent because of how hard it is to control larger groups as there was an inherent delay before all units would receive their commands in larger groups and they would also start to bump into each other in the process.

They accomplish different things in my opinion. Smart combat should maximize unit potential without necessitating APM and allow for a more diverse selection of tactics (do I want fast, fragile units with long range: skirmishers; do I want robust units with normal speed and range: assault). The problem with balancing range though for example is that unless a player is actively kiting it provides an advantage of maybe 1 or 2 extra shots - and if 1 or 2 extra shots are impacting battle tactics the game may be excessively engagement focused e.g. micro heavy, low HP to high DPS units. Indeed, if ranged units are balanced around kiting without Smart Combat anyone with high APM will undermine the design philosophy of minimizing micro.

Smart combat as I mentioned has a secondary value of shifting the focus from engagement to battlefield management. Smart projectile-dodging units are then always effective at flanking slow ROF heavy units even when the scale of the conflict is massive and you don't have time to focus on specific unit groups, let alone specific battles. This doesn't make dodging overpowered though, skirmishers will still be weak against normal ROF assault units. This opens you to entire battlefield tactics around having a good mix of units but also drawing your opponent into a position those units can exploit

Excess unit/economic micromanagement are pratfalls I fully expect Uber to handle. I just hope it's not at the expense of mechanics like kite, dodge and strafe which can be used to create intrinsic balance and what make simRTS so deep.

edit: grammar

++

The game could work even without any unit stance, but I think something smarter is important for making the game more macro-oriented and scalable without paying a price on potential unit interactions and diversity.

Indeed, smart combat could be condensed into a separate "Fight" command as it is in Zero-K or a single combat stance.

It doesn't even need any additional UI element, "fight" is basically A-move with more automation, and the most Zero-K players never use the lightbulb button to shut down unit AI anyway, they use hold position instead.

not sure if this would be useful to sorian, given his powess, but this is the zero-k code as an example

Sounds like "play the game for me" command. I want to fight myself
No need to throw arguments at me, I know the discussion already. It's all subjective.

Well, see, there is objectivity to it. The less micro you have to do, the more time you can put into macro. Heck, if we still want micro even, the less unit micro there is, the more time you can put into economy micro. On the other hand, the more micro there is, the less time there is for macro-oriented projects.

Yes, advanced players can overcome this. But even if there is less micro, they will be no less advanced. They'll put their skills to bigger and better things.

And besides. It's not like if they did have unit macros, that you wouldn't be able to control the units manually. That would be silly, there's only so much an AI can do that a human can't

Indeed, why have variable unit ranges if they can only be used by high APM players - or worse imbalance the game when high APM players use them. I would rather have unit variety and smart combat so skill differential reveals itself in strategy not micro.

Pertinent example: Forged Alliance Seraphim T3 Siege Tank
In high level games its short range made it easy to kite with other factions' T3 Bots
In low level games nobody expends the APM to kite and it's the strongest for cost T3 unit

That's impossible to balance for a wide skill differential and burdens competitive players with necessary micro.

viewtopic.php?f=61&t=34222

Zero-K's implementation still allows for superior human control but lowers the APM barrier.