Flash of Steel header image 2

Three Moves Ahead Episode 20

July 7th, 2009 by Troy Goodfellow · 21 Comments · Design, Podcast, Three Moves Ahead


The panel returns to game design territory with a discussion of the problem of runaway winners in strategy games. Can you design around it? Should you? How do you keep a game interesting for everyone?

Also, a progress report on our Dominions 3 game.

Listen here.
RSS here.
Subscribe on iTunes.


21 Comments so far ↓

  • MazeMazure

    Great episode again! Nice insights. Also enjoyed the progress report on Dominions 3–but now want to hear the other side of how Bruce sees the game developing. Hey, he should do a separate recording of 15 minutes alone! That’s the least he should do when missing out on his co-host responsibilities! ;) Kind of a raw Tom and Bruce unfolding…

  • Rythe

    I’m also interesting in seeing what Bruce thinks of the Dominions 3 game’s progress and the sorta coalition against the perceived inevitable (his victory). Also, there’s now way, way too much cheerfulness in the podcast.

    The runaway victory is something that has always annoyed me about strategy games. It’s like a battle of attrition in reverse and underscores a flaw in most of them – the weaker force cannot win. Of course, this assumes that the stronger doesn’t make any gross errors and that we’re not playing Risk.

    The question is whether or not there is a place for the finer tactical methods of Guerilla Warfare, traps, and terrain manipulation in grand strategy games? It’s hard to even pull off an affective bait and switch because losing the sacrificial piece often hurts you more, comparatively, than anything you can hope to gain. Even time is your enemy in such a situation. In the real world, victory for the weaker is often a question of whether they can put the stronger in an unfavorable position for long enough and/or if they can put a slick trap in place. Both are simply not options in the digitized versions.

    I think one of the biggest problems is modeling how cumbersome large forces can be. Supply lines – one of the obvious targets for a weaker, often defending, force – are often ignored because they’re simply tedious and not fun. Tedious and not fun are probably the greatest factors in why larger forces are usually just as efficient, swift, and easy to support as smaller ones in games.

    The second big problem is stealth. Larger forces are often just as discreet as smaller ones. You can’t send your elite squad into a war torn city and reasonably hope that they can get behind enemy lines, cause carnage, and then disappear. Even if you do pull off a flank, it simply doesn’t matter 90% of the time.

    So can you use a computer to model these sorts of things and keep the implementation sleek enough that you don’t loose more than you gain? I’d say yes, so why hasn’t it been done besides for considerations of effort? I mean, if half the triple A titles switched the money put into modeling and interface work with the money they put into graphics, than we’d have those games.

  • Chris Nahr

    There are still strategy games with AAA budgets? Where? :p

    Otherwise I agree, the fundemantal problem is the total lack of supply and moral/cohesion modelling in most games. If a big army is just as effective in every respect as a small one, then of course the bigger army is going to win every time.

  • George (Stuttgart) Brof

    Chess is the king of strategy games. You can play it tactical, or very strategic, or both, or you can change your style according to your standing in the game. It is an RTS even when you do not play with the clock (time and tempo matters in chess).

    Hey Julian, Troy or Tom, you should try Chess With Friends on the IPhone (email chess), my handle is “brof” Would love to see an invite from you guys :-)


  • spelk

    Always enjoy the show folks, always lots of food for though and mental mastication afterwards.

    About “snowballin”, I was trying to accommodate the idea in all gaming, but feel it has a very different context when applied to player versus player (or team versus team) competitive play over single player versus an AI. For competitive play the limiting factor constantly melting the snowball has to be applied even handed over all players, whereas with AI play, you can tweak the “melter” specifically at the mechanics that are likely to start and maintain the hefty roll. In PvP it has to be seen to balance with all styles of play and indeed spoiler mechanics may enhance and prolong the enjoyment of the game, but in an AI game balance only has to be maintained in terms of the experience, carrying the player through a campaign, and providing enough challenge to make the players success feel rewarding.

    I may sound like a bit of a broken record here, but the one game that instantly jumps into my mind as having a snowball “melter” in place is AI War: Fleet Command. A 2d space RTS where you are pitted against an AI, singularly or co-operatively, and your goal is to conquer the AI Headquarters. The game uses your actions against you, at the start of the game you have a measure of the aggressiveness of the AI (its called AI Progress), and as you start to build, explore and expand, the more planetary systems you take over from the AI, the higher the AI Progress gets, and subsequently the more aggressive the AI becomes. So it starts to send more waves at you, fleets with different types of specialised ships, as well as building up its own defenses. If you adopt a “conquer everything” approach, you will quickly find out that the AI Progress has shot through the roof, and you’re being annihilated wholesale.

    In order to win the game, you must pick and choose your interplanetary paths and targets, aiming to find the data centers and advanced research labs so you can steal the AI’s tech and have a chance at weakening the AI’s progress level bringing it back down a notch or two. So rather than amassing a huge fleet, and snowballing over everything, you have to create precision strike fleets cutting away at the AI like a surgeon, barricading and blocking your flank, whilst sniffing out key targets and moving forward to ultimately make the final push onto their H.Q.

    You need defensive tactics, to protect yourself, but you can’t simply collect and grow, and because the AI shifts gear as you make your attacks, you never get to the snowball rolling point. I should mention that the resource needed to upgrade your ships to better tech is the only resource that is finite, per planet, so it drives the game forward towards exploration and tactical conquest, that coupled with ship numbers capped, means even your low tech ships are still useful mid-game. Anyway, the AI part of the game is the interesting draw for me, and the developer is quite open to discussing his AI, have a look[1].

    I’m a big fan of Endwar myself, so count me in with my hand held up high, for a podcast devoted to the game. Even without the novelty of the voice recognition commands mechanism, the game plays out at a pace where tactics are not relying on twitch and higher strategies can be adopted.. but ultimately looks good at the unit level point of view. Tom, I think the “space bucks” in the game are referred to as Command Points (CP).

    Finally, Go was mentioned, and I wanted to express (with shame) my complete lack of ability to play Go. I think its because I’ve been reared on the likes of Chess, with pieces having clear roles and moves, and the board having an initial setup and a clear direction of play. With Go, the pieces are all the same, the board is an empty playground without meaningful clues, the territory grabbing patterns don’t stand out to me, my mind has to fish about calculating liberties and laboriously trying to put into place a strategy based on visual clues I’m just not seeing. Still, the game fascinates me. How can such a simple mechanic produce such a complex and seemingly impenetrable game? No doubt even Go’s pass mechanic is in place to prevent snowballing games from carrying on too long.

    [1] http://christophermpark.blogspot.com/2009/06/designing-emergent-ai-part-1.html

    Only one link this week Troy, apologies for triggering the spam mechanism last week and having to bother you to sort it out for me.

  • Dectilon

    Dominions 3 really relies on diplomacy to offset one players advantage over another, and there are numerous ways to dick someone over without actually starting a war with them. For example, there’s a random event about a cyclone causing destruction in your province, but there is also a spell you can cast on an enemy province that will generate the same message for him. Was it you, or just random? ;)

    Of what you’ve said so far about the game I’d say your assumption is basically correct unless you do dogpile, although that might not help either. After turn 1 you should be able to capture one province per turn, preferably more. Whether or not it’s reckles to send your pretender into combat has everything to do with which pretender you’re using, or if it’s awake at all. A mage pretender won’t do much early on since it has no spells, and it could easily get killed by a stray arrow. The Gorgon pretender insta-kills anyone that attacks it in melee, meaning it can take on entire provinces alone right from turn 2 (it’s unwise to attack blindly right at turn 1 since 30 knights could be waiting for you).

    Recruiting crappy units in every province isn’t the way to go about it either. Everything costs upkeep, so only get units that complement your castle-built forces (like archers for teams that lack them). It’s best to have a second castle before turn 10 if possible, and that costs a lot of cash.

    Also remember that once you’re able to capture provinces with impunity it’s time to build a force that can fight a force like yours with impunity. That means summoned creatures. If you forget to summon creatures I can guarantee that Bruce will show up on each of your doorsteps with an army that dwarfs yours in power.

    If you didn’t buy a rainbow pretender (that is a mage pretender with a picks in most, if not all, magic paths) then you should probably start out researching site searching spells so you can have some gem income for when you need to start massing summons.

  • Jon Shafer

    I think it was Julian who said that when a city in Civ 4 isn’t producing anything the production is lost, but that’s actually not true – it gets added up and does apply towards whatever you pick next.


  • howard

    I think run away victories are part of the abstraction of RTS games. You loose the stretching of supply lines, the isolation of forces and you generally don’t have the unrest of local populations to contend with.

    Most systems that are there to prevent runaway victories always seem a bit too contrived and clunky in games, entirely too overt and overbearing. And the natural increase in management that arises from massive games like GalCiv2 really just make the game take a bit longer rather than really preventing the steam rolling of your opponent.

  • Aristander

    Dectilon has given you good advice take it. The strength of the monkies is in its special summons and a strong bless. Bruce has taken a nation of strong giants but his units can be swarmed in battle. In dominions battles can be won by the smaller force if you break his morale.
    Dominions 3 is hands down the best strategy game I have ever played. This is coming from an old timer who started with Third Reich before computers and played the original civ and moo.

  • Dhaeron

    I’ve followed your experiences with Dom3 interested, and i hope you’ll get into the game (say, after fifty games or so :D) I’ve been playing the series since Dom2, and i’ve at least one multiplayer game of Dom3 going on ever since i bought it, even if it’s just a PBEM game with a buddy. So, if you grow to like it, it can last for a very long time.
    I wanted to comment on the observation that resources don’t get carried over. This is (obviously) by design, and it is not a bad decision. The most important point in limiting resources on a per turn basis is that this limits the number of troops you can recruit in a single province in a single turn. If those resources could be preserved you might build up thousand of resource points in some backwoods little castle, and when an invading army aproaches, you could blow all the gold you get from thirty provinces in a single castle. To prevent that, resources place a limit on how many troops can be recruited in a single castle at once.
    Because of that, balancing gold and resources is also an important stategic decision, though it’s one you make before the game starts. When you design your dominion for your nation, you should estimate how much gold you’ll use in relation to your resources, and adjust the order/productivity scales accordingly. Nations such as Vanheim have troops that are extremely expensive but require very low resources, while nations such as Ulm might actually have a resource to gold ratio of more than 1. (For vanheim it can be something like .1) And all that is actually a fairly minor part of dominions gameplay, since the real fun kicks off with the various magic strategies.

  • Katy

    Just thought I’d point out the Game Design Concepts e-course/blog spent some time discussing feedback loops (e.g. snowballing) today. (Search for “Feedback Loops” about halfway through the entry).


  • Andrew

    I’d be curious if you guys are going to do a second or third game of Dom3. The game is the definition of Min/Max in a lot of ways that aren’t immediately obvious. Try reading some of the strat guides by Baalz on the Sharpnel Games Dom 3 forums for priceless examples. GL!

  • CitizenParker

    Romance of the Three Kingdoms IX had an interesting anti-snowball feature – when you were far enough ahead (in any scenario featuring many different states), the other players would get the option of forming an “Anti- Coalition.”

    Anyone who joined the coalition would get increased honor and prestige for doing damage to the target of the coalition, but if the coalition failed (ie, the targent wasn’t significantly set back), then they would lose a fair amount of their gains.

    The actual effectiveness of this in preventing runaway victories varied depending on the aggressiveness of the members of the coalition. Oftentimes the coalition will end up having a bunch of members that won’t attack in any situation regardless, so it’s pretty ineffective. But the thought was nice.

    Oh, and no dedicated End War podcast, please. I love Tom to death, but I think we already have 4 or 5 podcasts’ worth of material from him on this subject throughout the past few episodes.

  • CitizenParker

    Erg, that should read above as “Anti-*YourNameHere*-Coalition. I should have known better than to use angle brackets.

  • solomani

    Kind of surprised there was no mention of Total War. That has extreme run-away victories which are laboriously long. They did a good jog with Rome with the civil war. Wasn’t perfect but certainly more interesting than the standard fair.

    For the wargame players out there. Every played World in Flames? A strategic level wargame of all of world war II across the whole globe. I have played it 3 times to completion and each time has taken 2 years of real time.

    That’s a game where at about 43/44 you can tell who is going to win but then its a grinding down of the enemy that takes 12 MONTHS of real time.

  • Alan Au

    I used to play a lot of 3-player games, and the main problem was that the first player out ended up having to sit around waiting for the other two to wrap it up. Sometimes the losing player would get to play king-maker, which frustrated the leading players and made them feel powerless.

    We ended up setting up a system where we allowed players to negotiate partial victories, the idea being that faster resolution meant we could play more games. This had the side-effect of letting the 2nd and 3rd place players track performance over time.

    This also solved the problem of having the top two players stuck in a stalemate that would have dragged on for hours. Now they could both negotiate for a shared victory and not have to worry about grinding it out until the very end.

    The extra bonus was that we could use the system to truncate games in order to break for food or catch a movie, etc.

  • Erez

    Is it just me, or is Julian a bit bitter here?

  • Blueprint for a War Machine | RobZacny.com

    […] the total number of men presently available for military service. This is one of the ways that EU3 prevents runaway victories. Unless your nation is exceptionally populous and wealthy, you cannot use giant armies to steamroll […]

  • Solace

    Demigod actually has a pretty interesting anti-snowball scheme. There’s one thing that’s not obvious to a new player, which then gets shouted at the start of every match in case you have one on your team: “Do not buy more minions!” Aside from your character’s personal summons, all minions give the opponents money/xp when they’re killed. Although more minions do help you out, they’re basically short term gain for long term loss, and it’s a big strategy to decide when to go straight from no minion upgrades, to maxed out minions.

  • Anders

    Why can’t I download this episode? It just opens a webpage and plays automatically…

  • Troy

    Right click and save file to location.