Thursday, March 3, 2011

Designing upgrades

In my last post I talked a bit about CiV design decisions.  Ziggyny questioned my stance, at least in part basing his criticism around the idea that upgrades that come later in a game don't have to be as good as earlier ones - in fact they can cost more and be worse and still be good to take.  His primary example was Starcraft 2, which uses that exact model.  For simplicity, SC2 upgrades look like this:

Cost: 100
+1 damage on all attacks.

Attack2 (requires Attack1)
Cost: 200
+1 damage on all attacks.

Attack3 (requires Attack2)
Cost: 300
+1 damage on all attacks.

Now since the player in SC2 generally starts with a small army and gets bigger as they go along the upgrades make a lot of sense.  The later upgrades are still very much worth it because they affect so many units.  Of course they aren't as necessary as the early ones; you might skip the later levels of upgades.  This system works fine.  To extend this to CiV mechanics though, imagine that there was a set of upgrades just like the ones above, but there were also two other sets of upgrades available.

Cost: 100
-1 damage on all damage taken

Defend2 (requires Defend1)
Cost: 200
-1 damage on all damage taken.  Heal 1 damage every 2 seconds.

Defend3 (requires Defend2)
Cost: 300
-1 damage on all damage taken.  Heal 1 damage every 2 seconds.  -20% damage taken.

Cost: 100
+20% movement speed.

Speed2 (requires Speed1)
Cost: 200
+20% movement speed.  +20% attack speed. 

Speed3 (requires Speed2)
Cost: 300
+20% movement speed.  +20% attack speed.  +20% training rate.

Nobody is ever going to take Attack2 or Attack3.  They cost a lot, just like the other upgrades, but they aren't even remotely comparable in power.  On a really huge map, if you were doing a mission for an hour, you might run out of other upgrades to take and eventually break down and take Attack2 or Attack3.  Until you get to that point where all the useful things are taken though you are never going to be interested.  The Defend and Speed trees above are roughly analogous to the Science, Culture, XP, Food and Production building tracks in CiV.  The more modern buildings have really powerful effects and large costs to go with them.  However, their powerful effects are so good that you definitely want to acquire and build them as soon as possible.  You can't get them all, so you need to figure out which ones are best in which cities, but unlocking a new building is important - you probably want to build them right away.

If you have a building track that looks a lot like the Attack tree it is going to be absolutely ignored.  The later buildings will never get built since you won't run out of actual good things to do so having them there just isn't relevant.  Of course, you might not really know this in your first game and you might well build them.  I don't think 'trap for newbies' is a particularly good niche though.  Lots of games get by on very shallow gameplay where after a few tries a good player will know exactly what to do in every game because so many of the options are so terrible.  The best games have a number of options all of which will be selected by good players depending on game state.  This means that games will take many different forms and lots of strategies will be feasible.  

In the unmodded game of Civ, for example, the only buildings that are relevant are universities and colosseums.  Nothing else matters.  If you are playing optimally you will never under any circumstance build more than 3/4 of the buildings in the game because they simply aren't good.  This leads to incredibly repetitive play and poor replayability.  A big focus in my modding has been to make sure that lots of different things are reasonable.  Sometimes I need more money so I build some banks.  Sometimes I need more border expansion so I build temples.  Sometimes I need a powerful army so I build barracks.  I know I have done something right when figuring out the optimal choices is difficult and I know that small changes in the game would have changed my choice.


  1. The thing is, your made up example isn't the same thing at all. Attack2 and Defend2 and Speed2 all say different things but at their very core they do essentially the same thing. When your units battle your opponents units they do so more efficiently than before. Which is better and which is worse depends on the different valuations on the different stats. It's like in WoW, where Attack2 is strength and Defend2 is mastery and Speed2 is haste. Depending on which class/spec you are you may want to buy a different one first.

    But in Civ gold and science and shields and happiness and culture don't all do the same thing. You tend to be able to convert between them at terrible rates in some directions but if your goal is to bribe a city-state then building more universities isn't going to do squat, for example. It's not intrinsically wrong to buy a bank just because it 'looks worse' than a university if what you want is more money. All the universities in the world aren't going to get you more money (and in fact are going to cost you more money in upkeep cost if you keep building them).

    It's like in WoW for tanking when you compare threat stats to mitigation stats to max health. It seems like the str/mastery/haste comparison for a DPSer in that you have stats that make you better and it feels like there should be a way to weight them. But there isn't. For a DPS spec it's easy to choose between 50 strength or 50 hit or 65 haste as a bracer enchant. You take the one that most improves your damage. There is a definitive right answer. (Probably strength!) But for tanking, do you take 50 dodge or 50 hit or 65 haste or 50 strength or 40 stamina? Here it depends. If I need more threat you probably take the strength. If I need to be able to interrupt a boss you probably take the hit. Healer mana an issue? Dodge is possibly good. Stamina is the worst by far in terms of stat point efficiency (the enchant isn't even from this expansion) and yet it's the one I probably want in general. Until this week I was using haste!

    You wouldn't be able to walk up to me and tell me I was wrong no matter which one I was using. It's not like DPS, where you can point out that strength just does more than mastery does. You can plug those into a spreadsheet and spit out the 'right' answer. For tanking the stats do fundamentally different things, and they do them in drastically different ways. The stamina enchant clear scales the worst. From an item point standpoint I should be getting 75 stamina, not 40, when compared to 50 of another stat. I should be getting 98 instead of 40 when compared to the 65 haste. And yet because they do fundamentally different things if I value the thing it does highly (and I do) then I should take it even if it 'looks worse'.

    Heck, even when it comes to the easily definable stats for DPSing I had to use a glove enchant from TBC in Wrath when I was DPSing, even though it certainly looked worse. Because despite other enchants having bigger numbers or things tacked on they just weren't as good.

  2. You are precisely right that even if the Bank was +1% Gold for 10,000 Production cost it might sometimes be right. Maybe you are playing a game that you intend to last 5,000 turns and at the end you want to have the maximum amount of money for buying city states. That is possible. I am not arguing that a game state does not exist where the bank would be worth building, I am arguing that that game state is so unlikely that you will actually never see it come up unless you specifically set out to play a game with the intent of using banks.

    Imagine this scenario in your bracer enchant example above: The only stamina bracer enchant is +1 stamina. Nothing else on the list can replicate that. Someday you might not care about damage, you might not care about healing, or avoidance, or anything else, but only about stamina. Therefore +1 stamina is a fine value for a stamina enchant. No! It isn't! It is useless and no one will ever take it. That makes it a nonchoice and thus a waste of time from a development perspective. Adding things in that only a few people will use is fine, adding things in that nobody (who is actually trying to play well) will use is a waste. We do not have an infinite number of buildings or enchants, just a few, so having useless ones just means we have no choice.

    I think your perspective on 'if I need money, I don't want universities, because they cost me money' is also not reflective of CiV at all. You *always* need more science. You are trying to win the game and every single game winning plan involves racing to endgame techs as fast as possible. Even if you are at the endgame techs and thus have run out of techs that are useful to win the game the Bank is useless because the game will be over in 10 turns! The bank will never have time to recoup its cost. Remember all the resources in CiV feed on one another. I may not be able to directly convert science to cash or food to money on a whim but the more of anything I have the more of everything I get. Adding a small amount of money is much worse than adding a large amount of anything because those other things get me policies, border growth, closer to endgame techs, bigger/better military, more buildings, and more money.

    I think I should throw a question at you to see where your argument goes. Do you think that the Bank in particular is balanced at 25% Gold and no other benefit, or are you just arguing that it is theoretically possible to have a later building in a tech tree that is worse than an earlier one and have that be ok? In addition, if I know that the Bank at 33% and +2 Gold will be regularly taken by players who are looking to boost their Gold output but not so much taken by players who already have sufficient Gold (from their perspective) then isn't that better design than a 25% bank that is never built?

    I can tell you from the perspective of a player who has played a *boatload* of CiV and done extensive modding that the Bank as it is in the base game is trash. It is a non choice. Is it possible that a later game building could be right to build even though it sucks compared to its prereqs? Yes. Does that make it a good design decision? No. Even from a fun perspective it is bad to have buildings that are terrible. It is exciting to get new options and have real choices in style that don't boil down to 'good' and 'stupid'. It is good to see people use different builds and try new things and still be able to be successful - this adds tremendously to replayability.

  3. I played a fair bit and did build banks. I did skip universities in some cities because of their upkeep cost. I may not have been playing theoretically optimally but I was having fun and the banks allowed me to keep playing. Money was a big problem for me.

    Maybe I'm stupid, I don't know. But I didn't think it was a non choice at all when I was playing. Would I have built the buffed bank too? Of course. Players always want stuff to be more powerful.

    My biggest issue was you claimed the developers were doing a bad job and made tons of terrible design decisions without justifying it. And now the justification you've given actually makes it seem like universities are a terrible design decision too, doesn't it? If you're always building it in every city no matter what because more science is awesome then shouldn't it be nerfed?

    So yes, I think bank at 25% and no other benefit is a fair building and one I personally built. I am also arguing the theoretical point as well because it was the entire crux of your initial post lambasting the developers.

    And yes, if a building is actually never built and a buffed version is sometimes built that is certainly a better design. But you didn't make that argument in the first place and unless I was abysmally bad at the game and my play style should be thrown out I don't believe that to be the case here in practice.

    And for the record, back in classic raiding I would have traded any amount of any stat for 1 stamina. Even in TBC I was willing to trade oodles of most things for stamina and agility. (Greens of the monkey over purples, anyone?) It's not as extreme now since with only 10 people and strict enrage timers tank DPS is more relevant than it used to be, but you better believe as soon as I don't need to be hit capped to interrupt anymore that the 50 hit rating is going off my bracers again.

  4. Clearly CiV was and is balanced such that people can play suboptimally against an easy AI and beat it. If you play on moderate difficulty levels and build banks you will still be able to win. However, the measure of a balanced game is not if the player can make subpar decisions, have fun, and beat a dumb AI. When I talk about balance I emphasize complex decision making even at the most brutal difficulty levels and the appearance of many competing strategies at those same difficulty levels. Banks do not appear in those strategies because they are crap. In SC2 I can beat the AI doing all kinds of dumb stuff. It is only when I dial up the difficulty to max that I suddenly notice that I have to play well and that certain strategies are simply not viable. Same goes here. Banks are bad. If you were handily winning while making lots of banks then you were playing on a easy difficulty setting (for you). If you wanted a real challenge you could have dialed up the difficult to where you would have lost a lot and then you would find ways to win - and you would end up dropping banks.

    I can give you a sense perhaps of why banks are so terrible. You need a lot of all the resources. The only one you can totally ignore and still dominate is culture. It turns out that the way to generate a lot of science is through science buildings. Building more science buildings drastically increases your science output as you have almost no source of science outside that. Production is similar but not as extreme - production buildings increase production quite a lot but not as much as science buildings by far. Money is totally different - very little of your money comes from buildings. Thus if you build a university in each city it increases your science output by 50%. If you build a Bank in each city it increases your gold from cities by 25% - which probably means 5-10% more overall Gold. Now it is true that Science buildings have upkeeps and Gold buildings do not, and also that doubling your Science and doubling your Gold incomes is not the same thing, but you can see why Gold buildings are considered terrible - they simply don't affect the vast majority of your Gold generation and thus have a minimal impact on your overall cash flow.

    I don't think your playstyle should be thrown out - but balancing around people playing a mediocre game against an easy AI is not especially useful I think. Would you be pleased if SC2 was balanced to the extent that players who play on normal difficulty find it okay but every PVP match was roach vs. roach because at the top levels nothing could be roaches? I don't think you were terrible at CiV but if you really think Banks were good buildings then it is clear you weren't playing at the top levels. Balance should be such that top players use a variety of strategies. I should note that even up to two weeks ago, prior to the latest patch, the optimal strategy involved building Colosseums, Wats(University), Monuments and absolutely nothing else in each city. You presumably have the city build *something* once those buildings are done but it is completely irrelevant what it is. This is a real mess because no strategy that involves building other buildings can compete and it isn't even close.

    Your comments about classic raiding are really hyperbole. Imagine you could have had 10,000 Str on your bracer or 1 sta. You could drop half a dozen dpsers with the extra dps (and threat!) that 10,000 Str gave you add half a dozen healers. I guarantee you half a dozen healers > 1 stamina. I do agree that there is no reasonable itemization formula they could have used that would have made stamina not the best choice. If you just mean that you take the stamina item/enchant over the notstamina item/enchant then I agree.