RT v. Turn based.

[F Smith]

Actually, there is no difference, from a programming standpoint.

A 'turn-based' game becomes a 'real-time' game when you loop the turns. You can even delay the loop some, say, give each play 5 minutes to think, then execute each set of orders and move pieces. Programatically, there's no difference.

Or, you can make movement soooo slow on the 'Strategic' map that the units take minutes to move.

[Gregor]

If there was even any debate about which it should be (I say, of course, turn-based), and it were that easy- just make it an option. However, I think Lucasart's Rebellion downfall was trying to make a turn-based game into a real-time game.

[F Smith]

My suggestion is to make it an option for combat, because there's no experience like being pressured by time. Real-time combat is by far more fun, to me, than turn based.

I'm thinking of something like 'Lords of the Realm 2'. 'Macro' or 'strategic' level commands would be made on a slow scale -- armies would slowly move across the landscape. It should take several minutes for an army to make it to the next province. But when 2 armies meet, you would then switch to 'battle' mode or some such, in which you lead your forces in an actual battle.

I suggest we build the code for the macro level now allowing 'real-time' slow movement as an option (since that part of the code will be rather simple), using a stat-based combat system for now. But we plan 'phase 2' or whatever to add a 'real-time' combat option.

I, personally, enjoy real-time games far more than turn-based. I like the 'virtual' experience of time working against you.

[Mog the Orc]

I disagree with Gregor. I think Lucasarts' Rebellion had it perfect.

It wasn't turnbased (no continiually clicking the end turn button) and yet it moved slowly enough (plus it had adequate speed controls) that you didn't feel rushed.

I also like the fact that it would take 3 hours or so to construct that first Star Destroyer.

There were other problems with the game, but I think the timing flow was perfect

[F_Smith]

I'm thinking that no one has done a good 'strategic' scale real-time game. They all are acutally on the 'tactical' scale. For now, the stat-based approach will do fine. I'm actually really hoping that the 'slow move' system will work better than the 'turn-based'. As Mark said, for Multiplayer it would be a riot.

But I'm thinking that eventually, a smaller-scale map for combat (in singleplayer mode, anyway) would be better -- on in which units represented either hundreds or thousands of men, instead of entire armies. You could then command the battle on the 'tactical' scale (but not man-for-man, like AoE or Starcraft).

[Glak]

I am against it for a few reasons:

Probably the worst is multiplayer. If you are fighting on ten fronts you will be at a serious disadvantage. Also you would have to be able to control all of your armies from one zoom level, otherwise you would have to switch back and forth between battles.

It would also be too slow. Slow real time games are slower than turn based. Some people favor slow games but I like a typical battle (whether turn based or real time) to last five minutes or less. The faster each battle goes the more you can have.

Also the work involved would be too much. Getting the units balanced would be extremely hard, esepcially when you factor in that someone could be using a different set of units than his enemy. Real time games are probably the hardest to design and balance, that is why most of them are trash. Time spent getting a mediocre RTS could be spent perfecting other areas.

I don't see any benefits. This is not because of a bias against RT combat, after all my favorite game is Starcraft.

[F_Smith]

I respect your feelings, but I think you're referring to 'tactical' scale Real-time games.

I'm talking about a scale more like 'Caesar3'.

1) Fighting on multiple fronts, in my opinion, *should* be a killer. That is one of the oldest no no's in the book. (Just ask Hitler!).

2) Actual combat would not be affected on the 'strategic' scale. There would be no 'control' issues, or 'time' issues. Battles would be quick and automatic, relying on stats and the unit mix you chose to send.

3) I believe most RT games are lousy because the designers are only interested in the game engine and pretty graphics, and spend no time/thought on the actual game rules. AoE is the perfect example of this (a chariot of 2 horses and 2 men uses less food than one villager?) . . . altho it is a *beautiful* game engine, and the graphics are first rate!

[Mark_Everson]

Real-Time slow movement would be fun for multi-player too.

As the combat spec works now (and it's Very limited) for an ancient battle your armies would each generally be in a single square. When they engage each other they fight it out depending on army composition and strenghts, and other factors like terain and random factors. Are you saying you'd go into a completely separate combat mode? Or are you picturing, for example, modern combat that's spread over a much wider front, still being controlled on the main map?

[Jason K]

F Smith:
Accutally, there is a difference from a programming standpoint. You can't just loop the turns... it won't work, and it would be way too slow to be constantly generating all that data, especially if little has changed(which would be very often). Instead, every time the game goes through the main loop you must poll the interface to find out what the user has done.

That aside, there are many other differences. RT games must be designed differently, because time is everything. Reaction time being part of that, which is why RT games are often called 'click fests' because those who excecute their strategies faster have an advantage. This I think is one of the main reasons strategic games are mostly TB, because each player has equal opportunity.

I'm a fan of RT games myself(I play them much more than I do TB), but I don't think it would work well in this case.

------------------
- Jason Kozak
Project Lead of Stellar Civ.
"It's not a new paint job on your battleship... it's diplomacy done right"

[Mark_Everson]

Hi Jason:

In principal I think the threading should be capable of both taking in keypresses and running the graphics, especially if they're not cutting edge. I think our problem is adding a much larger AI and game model load into the process also. We'll need to see.

My main point in this post isn't that however. Let me say right up front that I personally have no use for RTS games, at least as they're currently done. I believe that RTS games currently available are more Real-Time Tactics games. What I would propose (not for the first relese of Clash BTW) is a real RTS game where Strategic decisions must be made quickly against a resourceful opponent. Providing we can do well on the AI This would torque the usual RTS genre into a regieme where the person that can formulate good strategy quickly, rather than play the interface like a piano, would have the advantage. This seems like an intriguing concept to me. To pull it off you need good AI so that you can trust your orders to be executed well.

-Mark

[Jason K]

I'm just speaking form personal experience. Currently I'm working on my own Turn-based strategic (Stellar Civilizations in the Alt Civs section) and speaking from my personal experience there is a difference, but maybe I'm just thinking to deep. It is true that a RT game does similar logic to TB games except at a continous pace. I agree with both of you on the Tactical aspect of RT games currently. It would be nice to see a true RTS. If unit movement is made slow enough it should work well. An idea though... I'd like to be able to set movements, and then excecute them seperatly. I've always wanted to be able to set up accuratly timed manuvers, but the current crop of RT games don't facilitate them.

------------------
- Jason Kozak
Project Lead of Stellar Civ.
"It's not a new paint job on your battleship... it's diplomacy done right"

[Mark_Everson]

Jason:

Timing of attacks and synchronization of actions is in the plan. Also things like don't attack unless you think you have better than 2:1 odds. I Hope we'll be able to pull it off Here's a snippet from the web page outlining the overall goal.

In the military area, what is needed is a football-play-type diagram. For instance, "on this front attempt to break through here and here with this many armored divisions, go this far and then hook (shown by arrows indicating motion drawn by player on map) and surround and crush this resulting pocket of the enemy".

Of course you can micromanage military more than this if you want.

And a few snippets from the commenting of the Attack class (so far little of it is implemented)

Attack keeps track of the planned attack on a particular square or series of squares. It records units earmarked for the attack (attackers[])and allows for a variety of timing constraints and other conditions (for instance synchronization) for the attack.

The idea being you could fire-and-forget your entire attack until something happens. Alternatively the player could ask the AI (General Zorg or whoever) to formulate an attack on the hated Bladzians. If you like its suggestion you can just say "go ahead" or you could modify it on the football-play diagram.

It all hinges on doing a Superlative job with the AI. We are already on our way with the map AI, and I have a setup partly implemented that will allow the AI to simulate an attack based on its knowledge (guess) of enemy dispositions and strength. If the attack is stupid (based on current information) then the AI won't make it.

Wish us Luck, we're gonna need some of it

-Mark

[F_Smith]

I share Marks thoughts down the line.

Current 'RTS' games are 'tactical' in scale. (but yes, they do have 'strategy', for any hard-core RTSers out there!).

But my favorite type of wargame is 'Strategic' in scale. And no one, to my knowledge, has done a 'real-time' game on a 'strategic' scale. This is a market niche we can probably exploit, esp most of the biggest selling wargames have been 'strategic' in scale.

And the idea is to make the interface/unit controls not reward the fastest mouse! And I'm sure we can do that, esp by slowing down unit movement.

[F_Smith]

P.S. -- Jason, you just run the 'turn' logic once per thread loop, only executing orders and determining sprite positions. It makes no difference, from a programming standpoint, whether that loop happens when you hit a button or after a give period of time.

[Glak]

This is a bit off topic.

I read an article once why heavy automation would ruin an RTS. It would be possible to preplan an attack. It would be impossible to preplan a defense. Thus whoever set up his attack first would win, simply because while his attack was managing itself he could be doing other things. Meanwhile the defender would have to analyze the attack and counter manually. This would be too hard. If he managed to survive the attack would have gained a significant advantage because he had free time during the attack. In RTS games the main resource is concentration (mental cycles, clicks, key strokes, screen space, etc..), automation messes this up by making certain actions cheaper than they should be.

However with a bit of effort you can set up nice attacks. In Starcraft for example you can have 12 BC all yamato simultaneously at different targets and then command them to go home. It is possible through a technique called cloning. It requires that little wireframe window so other games aren't likely to have something similar. First you tell them all to yamato, then you deselect one, tell them to yamato another target, etc.. until they are all targeting seperate targets. Then just select them all and shift-queue them all home. It is really scary if something like that happens to you.