The Missing Combos - Filling the Gaps in BTS

[TheMeInTeam]

I have to say I enjoy TMIT vociferously insisting that no one could know whether IND/PHI is overpowered, while making 100% guarantees about what sorts of testing Firaxis did and did not do. Not that I disagree on either point, but that's quite the epistemological asymmetry.

Hmmm, well let's put it this way:

Firaxis basically told us outright they didn't playtest features in BTS. "we've not won an AP victory" or some such. While I was not part of the civ IV design team in any phase, I still have some pretty strong suspicions...would you instead accept 99.99%? That's a better 99.99% you'll ever have in civ by the way.

Put simply, Firaxis does not do comprehensive tests of balance of features against each other. It doesn't take Sherlock Holmes, people with Firaxis, or a Blizzard employee to figure that out. There are plenty of things that I do not claim knowledge of whether Firaxis tested or not. However, when it comes to a comprehensive game-to-game statistical balance analysis of traits to see if any combination is overpowering (or some other way that would offer definitive proof), I would bet the entirety of my possessions that they didn't do one in a heartbeat if I would get even a 50% return beyond the bet. The balance of SPAWNS in this game is so off kilter that they can easily outstrip any trait; good players point this out often. For sake, look at the balance between top and bottom tier existing civs, it's a joke. Are you trying to tell us that IND/PHI would consistently out-perform getting rushed by immortals? SKIRMISHERS (no strat resource needed, hits early, and if you don't have your own strat resource or inca, you're dead)? IND can't even out-perform the benefits of a wonder resource, and yet even years after release and with tons of experienced players saying otherwise we still see nooooooobies blather on about what is "obviously overpowered".

Yes. I'm calling them out for not bothering with comprehensive balance testing dating from before civ IV was conceived all the way until now in civ V. I'm calling out you too, if you somehow think the trait combo is OP when compared against contemporary alternatives. Rolling Gandhi with stone + marble in mapfinder isn't illegal in most HoF gauntlets. Using Inca at all is banned. What was overpowered again? Even HoF didn't do anything to prove Inca OP, it was just "overused", despite its inferiority on faster game speeds + high difficulties.

You think Firaxis possibly tested this crap? HAAHAHAHAHAHAA. Let's take it one step further: do they even know HOW? Show me a Firaxis release where each of the game options were roughly comparable to others on a typical map, and you're probably showing me civ II. It's been that long, and back then it was achieved by making the civs identical.

There's one more thing that pushes the margin $@#% close to 100%: the state of MP on the games' release. Forget setting up a rigorous statistical data analysis; there's hard evidence to the public that the raw data couldn't possibly exist! That is, of course, unless you believe some conspiracy like Firaxis releasing a version different from what they're working on. Yeah right.

Maybe the statements are a tad less asymmetrical than you indicate. What I've put above is a LOT more than Firaxis has ever told anybody about their balance methodology. In other words, right now my case is stated a lot more strongly.

To that end, I've created a small mod: Field Test: Phi/Ind Leaders with two Phi/Ind leaders. Shah Jahan I has been added to India, and a new civilization, Palmyra, lead by Zenobia has also been added. Palmyra has no UB or UU, so players wanting to test the alleged power of the Phi/Ind trait combo without the potential bias of UBs or UUs may do so. Shah Jahan was added so people who want to play with a UU and UB can. Other than the changes mentioned above, gameplay is unchanged.

How is this "field test" being controlled? If you don't have good methodology, the only thing you're accomplishing is having people try a few new things for fun. Here is some immediate bias in your field test that needs to be addressed:

1. Players have different skill levels. Even players at the same difficulty are not really comparable. I can beat deity, but I can't do it like ABCF or USun or Duckweed and pull out 700 AD rifles (how many people @ firaxis do you think could pull 700 AD rifles on deity?)

2. Map spawn balance. Say you give the IND/PHI a whirl and roll plains cow + 1 FP. Then you play an existing civ...say someone pretty good like Mehmed or a top tier guy like Darius and give them 2 wet corn + 2-3 gems on riverside. Or give plain old Gandhi marble and/or stone. You think IND/PHI is going to compete with that? No chance. So what do you do? When you have this much "noise", you have no choice but to control it via game spam. So..............................who's up for 30+ games of this trait combo and several others...more or less depending on just "how strong" this IND/PHI combo is? Anyone willing to bite? If you play as fast as me, this might only take you a few weeks of nonstop play. Most players don't play as fast as me, or even 1/3 as fast.

3. By what metric are you measuring "overpowered"? Player ability to win the game? It would have to be.

4. You can't remove UU/UB from trait balance and still call it "balance". Maybe Firaxis can do that, but nobody with sense will believe it. A merely above average trait set can become truly devastating, even top-tier by slapping in top unique units. A good trait set can be hampered by miserable UU/UB combos that don't add much. Even starting techs play a small role. Are you going to start comparing traits all on a civ without UU/UB? If you're not, your "field test" is ignoring important considerations that drastically affect balance in this game. Without skirmishers, the Mali are just ok-to-good in MP. With them, they're a top tier civ...you're extremely hard for anyone to rush and are a serious rush/choke threat yourself. Take those away, or prats away from Rome (replacing with some garbage like ballista elephant w/o ivory, SEAL, or Panzer), and their trait sets leave them looking average tier in SP and Rome's case probably unused/rarely used in MP. Sure you want to ignore OTHER trait UU/UB in this analysis? You can't.

So how do we test this comprehensively? Simply put, we don't have the resources to do a responsible study to determine balance of traits. That's the department of game design that pretends to care about balance. For a forum, the payback (proving or disproving a point) to posters is almost never going to be high enough to give enough noise-free data to be conclusive, and so people won't test it as it needs to be tested to prove anything.

This reality is also part of the reason I'm confident enough to bet my entire worth Firaxis didn't do it as part of their balancing of traits + decision to not include certain traits (unless you want to make a case for CHA/CRE and ORG/PRO being OP too lol). What probably happened is that someone or a couple ran wonder spam back in 'nilla testing and it "seemed" too good. Besides, what's the point in bothering with "balance" when you're making a DELIBERATE decision to leave AI THAT DON'T EVEN TRY TO WIN? What a joke. Asymmetrical my foot.

 

[ABCDPuppies]

Hi TMIT. First of all, I was kind of just lightly poking fun at the tone of your rhetoric . No harm meant.

But as to the serious point, I wasn't so much questioning your claim that Firaxis didn't comprehensively test trait-combo balance -- that sounds about right, and you certainly know about it than me -- as I was puzzled why you thought no one could in principle know whether PHI/IND was overpowered, given that Mods exist which allow you to play that combo and any other with any leader, civ, etc.

For what it's worth, my own suspicion is that PHI/IND is probably not too overpowered -- certainly not if by that you mean it would give a decisive advantage over any other trait/UU/UB combo, regardless of the map and other contingent factors. But then again, that is probably too strict a conception of what being overpowered amounts too. After all, Praetorians won't give a decisive advantage (or any advantage for that matter) on map without iron. But it is still an open question whether Praets are overpowered.

So PHI/IND might be overpowered in a weaker sense -- not that it always gives a huge advantage, but that it gives an advantage at a rate, and to a degree, that is significantly greater than any other trait combo.

Of course, it wouldn't follow from the combo being overpowered in this sense that it shouldn't be included in the game. As you say, you would just have to balance it with other civ factors. Actually, come to think of it I quite like the idea of a builder leader who has kind of a crap UU/UB.

 

[ParadigmShifter]

Do you really have Adult Baby/Cross Dressing puppies?

 

[TheMeInTeam]

as I was puzzled why you thought no one could in principle know whether PHI/IND was overpowered, given that Mods exist which allow you to play that combo and any other with any leader, civ, etc.

Because actually KNOWING/PROVING something to be overpowering on an average basis is quite difficult. In order to iron out the noise, you need not only to play the mod allowing it, but also to have a ridiculously large sampling that is properly analyzed. I'm saying that nobody I'm aware of has done that; it's essentially beyond the capability of mods unless they have a built-in method to automatically send game data back to them in large #'s and a way to process it. What modder does that ?

After all, Praetorians won't give a decisive advantage (or on any advantage for that matter) on map without iron. But it is still an open question whether Praets are overpowered.

It's still an open question whether fast workers, samurai, panzers, and ballista elephants are overpowered too.

For what it's worth, my own suspicion is that PHI/IND is probably not too overpowered -- certainly not if by that you mean it would give a decisive advantage over any other trait/UU/UB combo, regardless of the map and other contingent factors.

Defining OP itself is indeed tricky. At minimum, you would have to show that this trait combo, on average, outperforms other trait combos across a large # of games. The only relevant metric I can conceive is victory; good luck getting statistically significant differences there while controlling as many factors as possible.

So PHI/IND might be overpowered in a weaker sense -- not that it always gives a huge advantage, but that it gives an advantage at a rate, and to a degree, that is significantly greater than any other trait combo.

So might AGG/PRO, or more realistically say FIN/ORG or SPI/CRE or IND/FIN in tandem with strong UUs. To even show it gives an advantage at ANY rate would take evidence. There exists none.

Of course, it wouldn't follow from the combo being overpowered in this sense that it shouldn't be included in the game. As you say, you would just have to balance it with other civ factors. Actually, come to think of it I quite like the idea of a builder leader who has kind of a crap UU/UB.

Germany and USA pack such leaders...they're rarely considered anything but average or less .

 

[ABCDPuppies]

Well yeah, no one's rigorously proven that PHI/IND is overpowered. But then, no one has proven that Firaxis didn't test this either.

My point was that both assumptions seem more or less equally plausible, on the sort of rough and ready evidence we have -- more or less anecdotal evidence from our own playthroughs and the reports of others in the former case, and inductive inference based on Firaxis's statements and track record in the latter -- although again, that depends on what you mean by 'overpowered'. And more broadly, that there seems clearly to be no difference in the kind or degree of evidence available that would warrant the claim that one could in principle know the latter but not the former. But that is a rather abstract point.

 

[A_Hamster]

Because actually KNOWING/PROVING something to be overpowering on an average basis is quite difficult. In order to iron out the noise, you need not only to play the mod allowing it, but also to have a ridiculously large sampling that is properly analyzed. I'm saying that nobody I'm aware of has done that; it's essentially beyond the capability of mods unless they have a built-in method to automatically send game data back to them in large #'s and a way to process it. What modder does that ?

Okay, what factors are in our control that would reduce noise? I had not thought things out as far as you have, but I thought if a hosted game like the NC series, with all players playing on the same map with a worldbuilder save of their preferred level should give some useful imput. With the Full of Resources mapscript, one can control starting resources, so the bonuses in the BFC would be the same for everyone (e.g. two food, 1 Wine, Copper).

Ideally, some players would play the same start position as a Phi/Ind leader, but using Louis XIV instead as a control for Culture Victory. However, from your previous two posts, it seems there is nothing meaningful that can be done.

 

[Kid R]

The Ind/phi thing came up a couple of months ago (http://forums.civfanatics.com/showthread.php?t=402285)

Here is the quote by Soren Johnson which mentions the reason it was left out.

That was different in Civilization IV. The civics, the unique units, the trade balance all came to be what they are through extensive testing. I remember a pre-release Civilization IV game where Aeson had got 40 or 50 great people, and I thought: “You can't have half the tech tree from great people. I need to adjust the algorithm a bit.” I suppose another good example is that there is no industrious and philosophical civilization. There are 28 trait combinations and 26 leaders, so there are two missing. One of the two is industrious-philosophical. That used to be Rome. Friedrich and a few other people, probably Aeson, made a strong case that philosophical and industrious is just too dangerous. Multiplicative math: If you're industrious then you can build more wonders, and if you can build more wonders you get more great people, which then is multiplied by your philosophical trait. They basically had proven that the Romans were too powerful.

Rewinding my memory to before the Civ 5 farce I'm prepared to believe the devs tried hard and in good faith to make a balanced game with all the decisions they made on Civ 4. All this bitterness will eat you up inside people

 

[babar]

I don't doubt their good faith, but I'm not sure if they always did try hard enough, or perhaps didn't have enough money to spend. Plenty of stuff (see TMIT), including the IND/PHI explanation, just doesn't make sense. On the whole they did a great job though, and I for one am not bitter at all .

 

[Optical]

I still think it's not OP, but I'll let you know in a month or 2 once I've got my personal mod up and running.

 

[G-Max]

phind is cool for Ancient or Classical starts, but prag absolutely murders all other combos in Renaissance (one of these days, I'm going to try Toku + Ottomans + Medieval start, pop Liberalism for Gunpowder, and roflstomp everyone). Ind/Org (Roosevelt), which combines half-price Forges with half-price Factories, rules Industrial and later periods, while Exp/Imp (Joao) similarly pairs cheap settlers with cheap workers for a fabulous early land-grab and Cre/Phi (Pericles) combines half-price Libraries with half-price Universities for epic science and culture output. Agg/Char, of course, is just cruel in any period.