Page 1 of 1

What makes a good challenge?

Posted: 19 Jul 2012, 15:02
by NoQ
After making some single-player challenges in my map thread, i'm getting a bit confused. Just how to make the challenge interesting?

In most of the challenges i tried (or made), the enemy gets weaker over time. Once you are able to survive the early rush and hold your ground for a while, and then manage to control just one more oil derrick, you already won. Simply because you were already equal, and now you have one more derrick, which both makes you stronger and makes your enemy weaker.

This is a simple monotonic curve: you get stronger with every tick, sit there and do nothing but defense (with tanks or towers) most of the time, and then destroy everything in the last 5 minutes.

Most of the challenges around boil down to: survive the early rush, hold your ground for as long as necessary to come up with the winning design (its components may be different), destroy everything with the winning weapon.

That's what i have noticed. And i sort of find it boring. Am i the only one who feels like that? (:

How to break this pattern? How to make the challenge non-monotonic? How to, for example, make timing frames in which you are stronger and then other frames in which your opponent is stronger and make it vital for the player to push in certain frames but then fall back, receiving only a certain energy advantage for this attack, once the enemy is ready to strike back? What other sorts of things will make a single player challenge interesting?

This theory doesn't apply to easy challenges (eg. back to basics), where you can easily beat the challenge in under 10 minutes, and then compete for the time record.

Re: What makes a good challenge?

Posted: 19 Jul 2012, 17:48
by Iluvalar
That's a good question. But isn't that the exact same in multiplayer ?

Have a map with multiple plausible frontiers. Have the player need to expand is border closer to other sources of trouble. Make the "center" oil more attractive than an enemy base. So the player will expand that way.

It would seem complicated, but maybe you could do a map that is easier at the ring #1 for a player (or some player) but would become symetrical at the ring #2 and more difficult at the ring #3 ?

In NRS there is a system I called the derrick taxe that make each derrick less efficient than the previous one. It's a nice feature that make the game resist a couple more minutes the collapse.

I also found myself losing a 2 hours game in NRS lately. The simplified tree allowed me to tech the AI to research correctly up to that extreme.

Also, I found that an AI that win over another and take it base is a excellent event to change the game pattern. Maybe you could make a map where some AI are stronger and some are actually on a bad spot. I guess it would make better challenges.

Re: What makes a good challenge?

Posted: 19 Jul 2012, 19:30
by Shadow Wolf TJC
How's about a map where the AI starts with a LasSat Command Post, and has outposts that are accessible only by hover and VTOL, and a main ground-accessible base that's protected by long-range artillery?

Or what about a map where both players start with VTOL technology and Cyborg Transports, but you need to use your VTOLs to cross a chasm in order to have access to the external oil?

Re: What makes a good challenge?

Posted: 19 Jul 2012, 19:35
by Andrie
@Shadow Wolf TJC so you need transport, I like your idea.

Re: What makes a good challenge?

Posted: 19 Jul 2012, 19:37
by Andrie
What about a challenge where you can only use Cyborgs.

Re: What makes a good challenge?

Posted: 19 Jul 2012, 20:25
by NoQ
Please don't decrease this thread to technical issues or technical ideas. I do admit that lassat could actually spice things up though, it's a nice idea if balanced correctly (: As well as the ability to choose your starting location by building your base in a certain spot or using a transport to land on a hostile continent at the very start of the game. As for borg-only, we can't set structure limits in the challenge files so far, i think. But these are pretty-much off-topic. The real question is what to do next.

Ilu's suggestions were quite helpful, i think. It seems to make sense to put many weak AI enemies rather than a few strong AIs, with their base put close together and relatively far from you, so that killing one of the AIs essentially meant giving its oils to other enemies rather than to you. This allows keeping the power balance while the player progresses. Then, some tight balance of oils on your base and in the open area is necessary to force you to expand with some risk rather than don't expand at all or always expand safely.

:hmm: i'll have a look at what i can do with this.

Another idea might be to have an allied AI that inevitably dies, buying you some time at the start. But once it dies you can no longer hold your fast expand and need to fall back, keeping only the power you managed to gather with you.

Re: What makes a good challenge?

Posted: 20 Jul 2012, 01:49
by milo christiansen
With 3.x you can specify custom rules and AI files for challenges which should make it much easier to skew things in the AIs favor :wink:

Re: What makes a good challenge?

Posted: 20 Jul 2012, 13:10
by zydonk
Talking ski here. To see a ski game purely in terms of challenge might be too narrow a view. It's RTStrategy, after all. I think it all boils down to the map, rather than the balance(ie the mod). The AI will always play a linear game; that is, it will always be either winning or losing. It cannot hold ground or manoeuvre its opponents (though Troman's Aivolution came pretty close to that). It has a number of options, though. It attacks the weakest enemies. If is destroys an enemy's asset(s), it will then concentrate its attacks on that enemy. But in all cases, it will withdraw if its losses go above a certain level. It will then not attack again until it has strengthened its forces through research. It will often, though not always, defend its oil - more usually when it is close to its base.

Thus, if there is more than one AI player, the human player may not be the object of any AI's attack, and it is a strategy available to the human player to manipulate the various AIs so that they fight among themselves. Given that WZ plays best with strength against strength - which is how Pumpkin originally envisaged it - the basic challenge facing the human player is to avoid being overwhelmed through the one great advantage available to all the AI players: while you twiddle the mouse as you plan your next move, the AIs will have run through many thousands of decisions. Again, when the AI decides on a tactic, it gets followed though with merciless precision. There is nothing quite as disconcerting as the remorseless drive of heavy tracks through your defenses and base.

So, the human player is always on a knife edge between advancing and defending. You must always strengthen against strengthening enemies, and always move with caution.

What kind of map suits this gameplay best? Test the original Pumpkin maps on v1.10 set to hard. You will find that these maps concentrate on exposure, where the need to take and hold a point will in turn expose you to danger. Rolling Hills is a very good example. You can defend the immediate environs of your base, but pretty soon you find that the area is too small. Once you try to take control of the next circle (see Iluvalar above), you find you run the risk of overstretching your resources. Getting more oil (the next circle) exposes you even further. A good custom map to test is the little four-hander called River Canyon - still one of the best custom maps ever made for WZ. It is hopelessly unbalanced, but the whole game of securing choke points is here raised to as high an art as WZ can offer.

The challenge here lies in the play of strategies. It is often a matter of patience and perseverance. The secret here, I think, is to understand that it is not a question simply of competition - an experienced player will win most games as a matter of course - but something more like play between friends. The more you allow the AIs play their parts, the more enjoyment you get from the game. You should like it when the AIs put up a good fight, build effective defenses, make good use of the research lists, put you under pressure. It's a game, after all.