Jump to content

Attack stats


Adeimantos
 Share

Recommended Posts

I've noticed that in 0ad, unlike most RTS games, the range units have higher attack than the melee units on average. I'm wondering what the reasoning behind that is, what are the pros and cons of making Melee units stronger in attack rather than health or vice versa than range units? And why do javelins have such high attack?

Edited by Adeimantos
  • Like 2
Link to comment
Share on other sites

This approach generally means that melee units become tanks and ranged units damage dealers. It's functional but not ideal for representing ancient warfare. In most cases ranged units would skirmish before melee combat started or in peripheral locations while melee units would do the main clash and most damage. But that's hard to represent in an classic RTS setting.

The ratio of attack/defense capabilities of a unit is one of the factors that will affect how it plays in game. A melee unit with high attack and low health will do loads of damage if in possition to attack, but will also die easily, especially to ranged units (unless it has high ranged/pierce armor and/or speed).

Heavy javelins make sense to have high damage ( low attack rate for balance/realism). Light javelins not so much, while pilum could slow down enemies or reduce their armor, but to my knowledge that's not currently possible in the game's engine (and roman foot are just swordsmen in game).

  • Like 2
Link to comment
Share on other sites

 I suppose it is good for battlefield tactics to revolve around taking out the enemies range units while protecting yours, makes more possible maneuvers than just having two Melee forces clash.  Hoplites as tanks seems reasonable enough to me but I think it would be good if some Melee units had more attack and less armor. Maybe there should be an Axman. I also would like a heavy cavalry with more attack.

In my mod I've  slowed down the javelin fire rate more, but given them less pre-attack delay than other range units, to give a little more of the effect of them having only a few javelins and having to collect more after that.  It'll be nice if they come out with special attack types  so a unit could throw a javelin and then switch to sword, etc.

  • Like 3
Link to comment
Share on other sites

Yes it can work that way. A mix units with varied attributes (melee units with high and low attack/hp, same for ranged) can also work. Or ranged units having low dps but being cheaper. For the weapon switching thing you could try using the "pack" mechanic of siege weapons or the upgrade system (both far from ideal though). 

Link to comment
Share on other sites

Take into account that ranged units often miss; they do less average damage per attack than the number stated.  It seems to me that melee units do typically kill things significantly faster than ranged units, if they are already in melee range.

An issue with low HP high damage melee units is that melee infantry in 0ad are also slow.  With low HP they would have a lot more difficulty walking up to attack a ranged unit.  Perhaps melee infantry should receive a "charge" ability or upgrade that lets them charge/sprint at the enemy over short distances.

But do consider realism.  An armored hoplite is certainly going to be slower than an unarmored guy.  The historic advantage of hoplites against skirmishers was that the skirmishers just couldn't do enough damage against their armor, and also once the hoplites met with the enemy, the skirmishers had to stop shooting because of friendly fire.

Edited by causative
  • Like 3
Link to comment
Share on other sites

11 hours ago, Adeimantos said:

I've noticed that in 0ad, unlike most RTS games, the range units have higher attack than the melee units on average. I'm wondering what the reasoning behind that is, what are the pros and cons of making Melee units stronger in attack rather than health or vice versa than range units? And why do javelins have such high attack?

Keep in mind ranged attacks behave differently than melee attacks. The difference is not only the range. Melee attacks are instant hits, ranged attacks delayed attacks affected by projectile speed and gravity. Furthermore, unlike most RTS games, ranged attacks do not necessarily hit where you're aiming at, not even when the target is not moving. In 0 A.D. ranged attacks have a probability function,  a spread (basically a standard deviation), which randomly distributes the projectiles over an area around; whether they hit their target or not is a matter of chance.

Despite their higher nominal damage, they are not necessarily able to kill quicker than the lower but more reliable melee attacks. Furthermore, ranged units have typically significantly less health and armour, meaning they die much quicker than melee units.

As for why javelinists have a relatively high attack, it's because they have a much shorter range than archers.

  • Like 1
Link to comment
Share on other sites

As others said, ranged units have higher damage and lower hp than melee units for convention. Instead of using a hard counter to represent the fact that melee units, once they reach a ranged unarmored unit, are able to kill them in a close quarter fight,  they simply have high armor and health as they can absorb up some damage with armor and shield.

About the movement, it is realistic to have armored units with slow movement ( despite someone stated that it isn't necessarily true ) and ranged with more dexterity. Still for convention some ranged units move slower than expected to give melee untis the time to reach them.

If ranged units had same hp of melee units and more damage ( as they do now ), would just be a more efficient unit in any situation rather than a weak unit with a high dps doing their best while protected by heavier melee infantry.

Also, ranged units have high damage with a chance to miss depending on their accuracy (spread) and projectile speed. This means that their real damage is, most of times, lower than the one shown on the structure tree as they can miss 1 hit of 3.

Accuracy is something realistic but I am not a real fan as it simply too many other factors like units moving back and forth resulting in an exploitable dance. I guess that, when formations will be fixed, spread will be a lesser factor as units within formations are displayed in row and columns very close each other.

Link to comment
Share on other sites

3 hours ago, Grugnas said:

Accuracy is something realistic but I am not a real fan as it simply too many other factors like units moving back and forth resulting in an exploitable dance. I guess that, when formations will be fixed, spread will be a lesser factor as units within formations are displayed in row and columns very close each other.

Accuracy is also bad towards balance as it adds randomness, especially in small engagements where more shots completely miss. A programmer could correct me here, but perhaps it's removal could also reduce lag in large battles?

Link to comment
Share on other sites

4 minutes ago, Prodigal Son said:

Accuracy is also bad towards balance as it adds randomness, especially in small engagements where more shots completely miss. A programmer could correct me here, but perhaps it's removal could also reduce lag in large battles?

Actually I think the ranged attack random spread is a great feature. It adds realism, complexity, and balance. Also, the randomness affects the position, which means ranged attacks can hit units they are not aiming at. This makes archers better vs closely packed masses than they would be in 1:1 fights.

As far as I understood it, lag has primarily to do with the pathfinder, not with ranged attacks.

  • Like 3
Link to comment
Share on other sites

3 minutes ago, Nescio said:

Actually I think the ranged attack random spread is a great feature. It adds realism, complexity, and balance. Also, the randomness affects the position, which means ranged attacks can hit units they are not aiming at. This makes archers better vs closely packed masses than they would be in 1:1 fights.

As far as I understood it, lag has primarily to do with the pathfinder, not with ranged attacks.

It adds realism and complexity for sure. I also agree with your last point on archers. But about balance I disagree. To balance an encounter as much as possible you would want the same results everytime it happens and there's always the chance of very bad luck messing things up completely. Starcraft II, that takes multiplayer more seriously than any other RTS I know of, has removed randomness/variation in attack damage and miss chances while it's predecessors had them. It's not a "must-remove" feature for me, there are good points for keeping it, but it's something to consider in case we want things as balanced as possible.

I know that the pathfinder is the main issue, just asking, as a clueless person on the subject, since I can imagine hundreds of accuracy calculations on top of everything else making large battles a bit heavier for the engine.

Link to comment
Share on other sites

Lag is due the units path continuous recompute, i guess.

Well, I kinda agree with the fact that spread breaks balance as introduces randomness but imho it is necessary for a realistic return. With formations and projectiles hitting the unit near the target within formation would also prevent the exploitable dance ( with dance i mean the repetitive movement back and forth of an unit in order to move away from the thrower trajectory), i hope.

In previous alpha there was a bug where a missed unit would still take halved damage, which, despite it was a bug, I didn't dislike it at all. There is already such a mechanism in many RPG games, where a projectile can hit for its full damage, half hit ( it is considered like a scratch. I am sorry for my lack of english skill and grammar ) or simply miss for no damage at all. Having Fully hit and "scratchy" hit conditions would partially reduce randomness without removing too much realism avoiding such dance. The con   is that would be weird to see an unit being damage by a projectile at few meters of distance from it.

  • Like 1
Link to comment
Share on other sites

Randomness can be accounted for. It's just a normal distribution. Spread does not unbalance anything.

Ranged units have the advantage of outranging others and are thus able to kill melee units before those could attack. Why I think spread actually improves balance can best be shown with an example. In AoK a single longbowman would not have any chance vs a paladin. However, thanks to their high fire rate and long range (in TC they could even outrange castles), a body of 40 longbowmen could finish off an equal number of paladins with minimal losses, just by concentrating fire on the closest foe; one shot times 40 arrows meant instant death to any paladin. In 0 A.D. this is unlikely to succeed, partly because spread ensures ranged attacks are no guaranteed hits.

Balance means something should work both for one vs one and for many vs many. Anything that closes the gap between is an improvement. Random spread does this, therefore it actually improves balance.

  • Like 1
Link to comment
Share on other sites

Randomness could be accounted for in theory, while making balancing calculations more complex. But in practice, nothing prevents the game from generating 3 or 4 unexpected misses in a row, or having a game win/loss defining last hit miss. Or both your archers missing in what would "statisticly" kill the approaching horseman and result in getting the archers killed. Imbalanced situations messing with the cost effectiveness of units will happen and some of them, even if rare will be extreme.

@Nescio I fail to get how 40 AOK longbowmen would take off 40 paladins with minimal losses, unless superiorly microed. Same with the rest of the example. Could you explain better?

@Grugnas Warcraft III has a solution to the half-hit display. Missiles don't stray sideways, and a "miss" label floats over the targeted unit. I guess not ideal for 0 A.D. though.

 

 

 

Edited by Prodigal Son
Link to comment
Share on other sites

4 minutes ago, Prodigal Son said:

Randomness could be accounted for in theory, while making balancing calculations more complex. But in practice, nothing prevents the game from generating 3 or 4 unexpected misses in a row, or having a game win/loss defining last hit miss. Or both your archers missing in what would "statisticly" kill the approaching horseman and result in getting the archers killed. Imbalanced situations messing with the cost effectiveness of units will happen and some of them, even if rare will be extreme.

The complexity of a calculation is irrelevant. And a victory at a game level does not depend on a single unit hit or miss.

6 minutes ago, Prodigal Son said:

@Nescio I fail to get how 40 AOK longbowmen would take off 40 paladins with minimal losses, unless superiorly microed. Same with the rest of the example. Could you explain better?

It is probably not supposed to succeed in AI vs AI, however, I did manage to slaughter my brother's paladins without losing any of my 40 E. longbowmen (in TC they can have 6+6 range). By the way, AoK's trebuchet also had a randomness affecting its accuracy to keep it somewhat balanced.

  • Like 1
Link to comment
Share on other sites

2 minutes ago, Nescio said:

The complexity of a calculation is irrelevant. And a victory at a game level does not depend on a single unit hit or miss.

I can mostly agree on the first part, ease of balancing isn't the most important factor here. I disagree on the second though, since be it a rare occasion or through a snowball effect it can still happen. Anyway I think the rest of my point still stands.

6 minutes ago, Nescio said:

It is probably not supposed to succeed in AI vs AI, however, I did manage to slaughter my brother's paladins without losing any of my 40 E. longbowmen (in TC they can have 6+6 range). By the way, AoK's trebuchet also had a randomness affecting its accuracy to keep it somewhat balanced.

Well, if it's not taking in account equal levels of player control I doubt it's a strong point. It  also seems strange to me that the stronger and faster unit would lose 40-0 under whatever circumstances, excluding path-blocked paladins, their controlling player just walking them around or maybe extreme difference in tech upgrades  (btw there's the thumb ring tech which gives perfect accuracy to archers - which like trebuchets, they normally don't have). But what I'm mostly interested in and I failed to phrase properly, is where/how do we draw the line of single unit power vs critical mass effect, to decide if this upside of spread vs full accuracy makes it worth it over it's drawbacks? Unless we fully understand it's advantage I'll tend to believe that SC2 designers know something more on this.

Link to comment
Share on other sites

1 hour ago, Prodigal Son said:

I can mostly agree on the first part, ease of balancing isn't the most important factor here. I disagree on the second though, since be it a rare occasion or through a snowball effect it can still happen. Anyway I think the rest of my point still stands.

Well, if it's not taking in account equal levels of player control I doubt it's a strong point. It  also seems strange to me that the stronger and faster unit would lose 40-0 under whatever circumstances, excluding path-blocked paladins, their controlling player just walking them around or maybe extreme difference in tech upgrades  (btw there's the thumb ring tech which gives perfect accuracy to archers - which like trebuchets, they normally don't have). But what I'm mostly interested in and I failed to phrase properly, is where/how do we draw the line of single unit power vs critical mass effect, to decide if this upside of spread vs full accuracy makes it worth it over it's drawbacks? Unless we fully understand it's advantage I'll tend to believe that SC2 designers know something more on this.

The example I gave is unusual, yes, however, it's not impossible. 40 longbowmen could win against cavalry because a human player can command them to concentrate fire and kill off the enemy one by one. It won't work with 10 vs 10, simply because 10 arrows are not enough to kill any knight at once, thus requiring more than one shot. What I was trying to show is that when you increase numbers, the game balance could occassionally work out differently than intended. Long range units benefit more from large numbers than melee units (who block each other's path). Having some random spread affecting projectiles decreases the advantage large numbers of archers have, therefore it improves balance, in my opinion.

As for balance, you could micro-balance on paper, e.g. 1 archer will always win vs 1 pikeman, 1 pikeman will will always win vs 1 horseman, and 1 horseman will always win vs 1 archer. However, 1 vs 1 balance doesn't necessarily translate to many vs many battles, nor to balance at a game level.

The only way to properly balance the game, not in theory but in practice, would be as follows. Run e.g. 100 Athens vs Britons 1 AI vs 1 AI battles on different maps and sizes. Note down how often each faction wins. Repeat the experiment for all other combinations of factions. Also run a couple of 2 vs 2, 3 vs 3, and 4 vs 4 AI games to check if you get different results. If a game is balanced than every faction should win about as frequently as any other.

Link to comment
Share on other sites

I like range units missing, and that does balance their attack some but up close they don't miss much, right? And when in groups Melee units get in each others' way, reducing their attack compared to range units.  Anyway I like 0 ad's system overall, but javelins should be reduced since range is less important than attack especially when supporting Melee units. To mix it up I'm working on making an axeman unit who will have high attack and move fast, with low armor, it'll be good against sword and siege units probably. 

Randomness doesn't unbalance things when armies are larger

Link to comment
Share on other sites

20 hours ago, Nescio said:

The example I gave is unusual, yes, however, it's not impossible. 40 longbowmen could win against cavalry because a human player can command them to concentrate fire and kill off the enemy one by one. It won't work with 10 vs 10, simply because 10 arrows are not enough to kill any knight at once, thus requiring more than one shot. What I was trying to show is that when you increase numbers, the game balance could occassionally work out differently than intended. Long range units benefit more from large numbers than melee units (who block each other's path). Having some random spread affecting projectiles decreases the advantage large numbers of archers have, therefore it improves balance, in my opinion.

As for balance, you could micro-balance on paper, e.g. 1 archer will always win vs 1 pikeman, 1 pikeman will will always win vs 1 horseman, and 1 horseman will always win vs 1 archer. However, 1 vs 1 balance doesn't necessarily translate to many vs many battles, nor to balance at a game level.

Your example isn't impossible, but imo for reasons I've already described, is true only under extremely specific circumstances. That said your overall point is valid overall and one I've missed so far, thanks for that.

Still, it's hard to evaluate if it's superior or not over the other approach, cause issues with randomness/spread I've mentioned above are also valid. Another issue with spread that came to me through this discussion is that while it can reduce ranged unit effectiveness in big fights, it does even more so in small fights, where there are less units to hit, so even more attacks go completely wasted.

Looking at AOK, where the last tier of armor upgrades gives one additional point of pierce armor, it think the designers look for a solution to mass-ranged superiority by using it, so spread isn't the only possible way.

20 hours ago, Nescio said:

The only way to properly balance the game, not in theory but in practice, would be as follows. Run e.g. 100 Athens vs Britons 1 AI vs 1 AI battles on different maps and sizes. Note down how often each faction wins. Repeat the experiment for all other combinations of factions. Also run a couple of 2 vs 2, 3 vs 3, and 4 vs 4 AI games to check if you get different results. If a game is balanced than every faction should win about as frequently as any other.

This would only somewhat balance AI fights under selected circumstances. It doesn't take account of every possible unit class vs each other, something that the AI won't do. Also computers play very differently to humans in general. Good old spreadsheet balancing and extensive manual testings, preferably in multiplayer seem far better to me.

 

Edited by Prodigal Son
Link to comment
Share on other sites

17 hours ago, Prodigal Son said:

Your example isn't impossible, but imo for reasons I've already described, is true only under extremely specific circumstances. That said your overall point is valid overall and one I've missed so far, thanks for that.

Still, it's hard to evaluate if it's superior or not over the other approach, cause issues with randomness/spread I've mentioned above are also valid. Another issue with spread that came to me through this discussion is that while it can reduce ranged unit effectiveness in big fights, it does even more so in small fights, where there are less units to hit, so even more attacks go completely wasted.

Looking at AOK, where the last tier of armor upgrades gives one additional point of pierce armor, it think the designers look for a solution to mass-ranged superiority by using it, so spread isn't the only possible way.

This would only somewhat balance AI fights under selected circumstances. It doesn't take account of every possible unit class vs each other, something that the AI won't do. Also computers play very differently to humans in general. Good old spreadsheet balancing and extensive manual testings, preferably in multiplayer seem far better to me.

Let's try to put it differently. Single combat, tactics, and strategy are three different things. Likewise, balance exists at different levels:

  • micro-level: one unit vs one unit duel
  • meso-level: many units vs many units fights
  • macro-level: a full game

The example I gave is just that, an example. Its purpose is to indicate that something at micro-level does not necessarily translate to the same result at meso-level (one knight will win against one longbowman, but 1,000 longbowmen will massacre 1,000 knights). Likewise, victory at meso-level does necessarily translate to victory at macro-level. It's perfectly possible to win the battle but lose the war.

Spoiler

In the game my example game from, killing paladins with my longbowmen was merely amusing. Five minutes later, while I was busy elsewhere hunting down trebuchets with my light cavalry, my brother sent in a dozen siege onagers and annihilated my 40 longbowmen within seconds. Half an hour later the bulk of my army was guarding my trebuchets bombarding some enemy fortications; simultaneously two groups of 40 enemy paladins, without any siege weapons, razed my castles and main town centre. Losing my base was inconvenient, but I skillfully managed to evade defeat and survived for perhaps another hour.

I believe the map we played was Great Britain; I (Britons) started out in Scotland, he (Franks) in England. Having lots of water allowed me to flee to Ireland initially and later to those small coastal strips of Norway and France.

Anyway, the reason why I ultimately lost the game (macro-level) was not because I lost a specific fight (meso-level), nor because paladins are the best soldiers (micro-level) but unavailable to the Britons. I lost, at least partially, because I was too slow with raiding. My brother already controlled about half of the land (and thus resources) when I had less than third. The game lasted for perhaps three or four hours, but the first half an hour, without any real fighting, proved to be decisive.

Ultimately we want 0 A.D. to be a balanced game (macro-level). Balancing individual units on paper (micro-level) is a sensible approach to start with, useful but insufficient to get the desired result. Units do not exist in isolation, they interact with each other. So after you've tweaked individual units (micro-level), follow it up by generating e.g. 100 units vs 100 units battles (meso-level) in the scenario editor to check if you still get the desired results, not just for pure unit types, but also for mixed armies. Also, observing multiplayer games can be useful; if a certain unit is never used, there is probably something wrong with it. However, all this is not enough.

As I wrote, we want 0 A.D. to be a balanced game, thus at macro-level. The only way to properly get there is by lots of playtesting. Because the 0 A.D. multiplayer community is not that large, finding someone to play dozens of test games with can be hard. Furthermore, every human is different, which means that human players just add more variables, making it harder to get macro-level balance. Generating many AI vs AI games is both more realistic and more useful. E.g. choose five sufficiently different maps; there are seven different map sizes; generate three games for each of them; 7*5*3*1=105 games. Do that for every combination of factions, to ensure no civilization is better than any other. This way 0 A.D. can be properly balanced (macro-level). Ideally. (I don't really expect anyone to actually generate thousands of AI vs AI games).

  • Like 1
Link to comment
Share on other sites

On 3/5/2018 at 3:55 PM, Adeimantos said:

I like range units missing, and that does balance their attack some but up close they don't miss much, right?

I haven't tested to see exactly how it works, but if I imagine/remember it correctly they should miss less indeed. If I'm correct, spread should be working with angles so the further away a unit is, the harder it would be to hit it. 

@Nescio You make some good points, but I still believe AI tests as the main focus aren't good enough or the best way to attempt RTS balance. When I say spreadsheet balancing and multiplayer testing, I'm not talking only about unit class vs unit class balancing. 

I'll try to describe the entire logic I'm using, though on top of my head I'll probably miss a few things. I'm pretty confident that you're aware of much of it, but it might be useful as a whole to other people interested as well.

Tech-tree balancing:

  • First there is the (very) early game. Here possible different unit interactions are preferably low and units across different civs are more or less cloned in stats or at least in function. Exceptions are possible, like a civ starting with ranged infantry instead of melee. Those make it harder, so I tend to avoid them, especially if many civs are present. Then relatively good balance is doable and essential,  this is the most important part of the game to have as balanced possible. Early game advantage can snowball, and uncounterable situations more likely due to less options. Best is to start with a single available melee unit class, not very good at offense and speed so that very early rushing, while doable, is not potentially devastating. If the game is going with civ bonuses those may cause some imbalance, but they should be carefully selected not to be really powerful, while any particular civ shouldn't be strong or weak in a majority of phases, unless those strengths or weaknesses are really mild. If a civ can potentially rush with one more soldier or with it's troops moving slightly faster, the other civ should be able to reach the next tech tier a little faster, or have some minor advantange upon reaching it.
  • Then as the game progresses, the techtrees gradually allow the training of more unit classes. Every new unit class available for a civ should be counter-able at the point it can attack the opposing base by any other civ, provided that both players are similarly skilled and the responding player has made the right choices. Likewise, upgrade and structure availability should be layered along those principles.
  • With the full techtree unlocked, the number of possible imbalances is (rather) high, depending on design choices. Still approximate balance through carefull unit availability is doable and essential. For example, if a civ doesn't have heavy ranged siege due to historical concerns, they could have archers with fire arrows so that they can strike a heavily fortified possition without sacrificing extremely higher numbers of units or resources than the defender while trying to break through.

Unit & Overall Balancing:

  • Clear unit roles. Preferably no two unit classes should play very similarly within any civ. This makes the game more interesting and every unit usefull at any state of the game if the circumstances are right, while giving more countering options. If we're going the historical way like in 0 A.D. those better be heavily inspired by real unit attributes.
  • Individual unit balancing according to it's cost, power etc. I'm using an edited version of a formula I found at GPWiki (which had a really nice article on RTS design, though it's sadly gone, even from internet archives, for some years now). It allows for easy balance changes and comparison of unit attributes. Each unit has it's stats adjusted according to it's level (L). Exact modifiers per stat can change. Say units prove too expensive, you can adjust the resource cost modifier to something lower universally. If kill rates make the game too fast, dps can be reduced universally. Units deviate in some of the attributes to get different roles, but usually no more than 20-30% from the standard of their level.

    Pop Cost: 1xL
    Resource Cost: 100xL
    Train Time: 10xL
    Hp: 100xL
    Dps: 10xL
    Armor: 5xL
    Movement: Average
    Range: Melee
    Other: (Could be damage bonuses, vision, skills, etc)

  • Testing follows: Units in 1v1, many vs many and mixed unit combinations in the editor. AI testing checking whatever's included in the AI's menu. Multiplayer testing, both in normal games and arranged situations to catch things that the AI can't help you with, test everything in real multiplayer conditions and get observations by other people.

It's hard to achieve perfect balance, in fact impossible with assymetric civ design. But depending on design and balancing choices and how much time you spend on testing, acceptable results are possible. I'm pretty much content with results I've had in the past using more or less what I describe above.

Edited by Prodigal Son
Link to comment
Share on other sites

3 hours ago, Prodigal Son said:

@Nescio You make some good points, but I still believe AI tests as the main focus aren't good enough or the best way to attempt RTS balance. When I say spreadsheet balancing and multiplayer testing, I'm not talking only about unit class vs unit class balancing. 

I'll try to describe the entire logic I'm using, though on top of my head I'll probably miss a few things. I'm pretty confident that you're aware of much of it, but it might be useful as a whole to other people interested as well.

There are many different procedures to balance a game. The specific approach someone uses when designing is irrelevant. However, to figure out if a game is balanced, not in theory, but in practice, lots of playtesting is necessary. Multiplayer and singleplayer games can be helpful, but AI vs AI games should certainly not be overlooked, and because the AI is constant, these test games are actually more reliable. Furthermore, one or two test games are not enough; large numbers are required to cancel out “luck”.

To clarify, I'm not advocating balance by trial and error. All I'm saying is that no matter how perfect the design (initially), playtesting (afterwards) is the only way to show a game is balanced at a game level.

3 hours ago, Prodigal Son said:
  • Pop Cost: 1xL
    Resource Cost: 100xL
    Train Time: 10xL
    Hp: 100xL
    Dps: 10xL
    Armor: 5xL
    Movement: Average
    Range: Melee
    Other: (Could be damage bonuses, vision, skills, etc)

Although rules of the thumb can occassionally be useful, there are no guarantees. Also, if a unit X has twice as much health, double attack damage, and increased armour compared to unit Y, X is not merely twice as strong, but actually about four times as effective.

3 hours ago, Prodigal Son said:
  • Testing follows: Units in 1v1, many vs many and mixed unit combinations in the editor. AI testing checking whatever's included in the AI's menu. Multiplayer testing, both in normal games and arranged situations to catch things that the AI can't help you with, test everything in real multiplayer conditions and get observations by other people.

No disagreement here, all this is useful and sensible.

3 hours ago, Prodigal Son said:

It's hard to achieve perfect balance, in fact impossible with assymetric civ design. But depending on design and balancing choices and how much time you spend on testing, acceptable results are possible. I'm pretty much content with results I've had in the past using more or less what I describe above.

Here you're mistaken. Symmetry can make things easier, but it's not mandatory. An assymmetric design can be balanced (e.g. AoM). Also, a game can be balanced (macro-level) even if individual units etc. (micro-level) are not, and vice versa. “Overpowered” (I dislike that term, it's used far too often unappropiately) features do not necessarily amplify each other, they can cancel out as well, or have no larger consequences.

Although tweaking individual stats can be highly enjoyable, often they hardly influence the outcome of a game. What counts is balance at a game level. How you get there and what your design is doesn't really matter.

  • Like 1
Link to comment
Share on other sites

9 hours ago, Nescio said:

There are many different procedures to balance a game. The specific approach someone uses when designing is irrelevant. However, to figure out if a game is balanced, not in theory, but in practice, lots of playtesting is necessary. Multiplayer and singleplayer games can be helpful, but AI vs AI games should certainly not be overlooked, and because the AI is constant, these test games are actually more reliable. Furthermore, one or two test games are not enough; large numbers are required to cancel out “luck”.

To clarify, I'm not advocating balance by trial and error. All I'm saying is that no matter how perfect the design (initially), playtesting (afterwards) is the only way to show a game is balanced at a game level.

I didn't say my way of balancing is the one and true way, I said it's my way. That alone doesn't make it relevant or irrelevant. I just got tired of recycling arguments as we both did in the previous posts and tried to lead the discussion to something more productive. You did this by keeping to defend your initially rather poor "longbow vs paladin" example while constantly editing it along the way, and then your "AI balancing as the best method" thesis, for which you avoided replying on what I proposed as it's weaknesses, instead you tried to find weaknesses only in my counterarguements, while making a lot of assumptions. I'm guilty as well, by often focusing on minor details, at times not expressing very clearly and continuing on and on already messed up parts of the conversation. So I thought, let's discuss on something more coherent, while it might also be more interesting for others as well. I'm also welcoming on learning my approach's faults.

Since we're still at it as it seems, AI balancing has it's uses, but it's vastly inferior to player vs player. The downsides of player vs player are indeed the ones you've mentioned, possible difference in player skill and hardship in finding many people available for it. But If you manage to find players of similar, preferably high skill and run many tests, that's the way to go. AI won't use many possible strategies, nor all possible units/unit combinations, won't tech reactively and will generally play very differently to humans, leaving huge parts of the possibilities out of consideration while giving us results, no matter how many tests you run. Human players will keep finding new strategies and tactics/micro, which will require additional balancing. The AI can't help with that either. The only thing it will be really more valid on is computer vs computer balance.

9 hours ago, Nescio said:

Although rules of the thumb can occassionally be useful, there are no guarantees. Also, if a unit X has twice as much health, double attack damage, and increased armour compared to unit Y, X is not merely twice as strong, but actually about four times as effective.

What I failed to note is that I also keep track of the percentage changes for each field of each unit's stats, so in the end all units have those changes similar overall. Then, not all stats changed as a percentage have the same percent effect in unit strength. Here is where experience in working with such formulas and in game tests for possible imbalances come helpful. If someone finds this part interesting I can further expand.

9 hours ago, Nescio said:

Here you're mistaken. Symmetry can make things easier, but it's not mandatory. An assymmetric design can be balanced (e.g. AoM). Also, a game can be balanced (macro-level) even if individual units etc. (micro-level) are not, and vice versa. “Overpowered” (I dislike that term, it's used far too often unappropiately) features do not necessarily amplify each other, they can cancel out as well, or have no larger consequences.

Although tweaking individual stats can be highly enjoyable, often they hardly influence the outcome of a game. What counts is balance at a game level. How you get there and what your design is doesn't really matter.

I didn't claim symmetry is mandatory and I'm not using symmetry, but varied volumes of assymetry, depending on game phase and specific design (say each of my mods or proposals for different games). I thought that would be pretty clear by reading my last post. 

What I'm saying though is that it's hard (or maybe impossible) to reach perfect balance with assymetry. We can reach something acceptable, as AOM did before the titans expansion and many games with various levels of assymetry have. No game with huge assymetry and more than 3-4 civs comes to mind though (saying this because I consider we largely/partly have 0 A.D. balance in mind during this discussion).

As described in my previous post, I'm not just tweaking individual stats. Your indeed important point of balance at game level comes of as very incomplete and counter productive to me, due to the dismissive way you approach it. We need a process to build each phase of gameplay as balanced as well. Not every possible encounter needs to be balanced (nor it can be), but each decision by a player needs to be counterable by the other player. Who wins in the end (and having a balanced ratio among civs) consists of huge numbers of variables that need consideration. I've found my approach largely successful at that. So far you haven't really pointed on why it's wrong or irrelevant.

Link to comment
Share on other sites

This discussion is heading into the wrong direction. I think I understand your position but I'm not sure you see my point of view. My intention is not to win this argument or to prove you wrong. Actually I believe we broadly agree on many things; the difference is where we put the emphasis.

Again, an example is nothing more than merely an example. Also, balancing is not either/or; how you work towards it is totally up to you. All I'm trying to make clear is that ultimately the only thing what really matters is to get a balance at a game level.

Link to comment
Share on other sites

8 minutes ago, Nescio said:

This discussion is heading into the wrong direction. I think I understand your position but I'm not sure you see my point of view. My intention is not to win this argument or to prove you wrong. Actually I believe we broadly agree on many things; the difference is where we put the emphasis.

Again, an example is nothing more than merely an example. Also, balancing is not either/or; how you work towards it is totally up to you. All I'm trying to make clear is that ultimately the only thing what really matters is to get a balance at a game level.

We did that almost from the start by going far from the thread's original purpose. I believe though it was an interesting discussion overall with valid points made by both of us and other people. 

I don't think I'm missing your overall point, with which I largely agree, I just happen to disagree with much of "your approach" towards it and some of your comments on "my approach". Winning an argument by all means is counter-productive. Winning an argument to propose something useful is fine though. I make such discussions for the shake of personal/mutual/general improvement, though everyone's ego can get in the way to some degree.

Balance at game level is the end goal, but discussing ways to get there is also very useful and interesting to me.

Link to comment
Share on other sites

1 hour ago, Prodigal Son said:

We did that almost from the start by going far from the thread's original purpose. I believe though it was an interesting discussion overall with valid points made by both of us and other people.

True, we certainly went off-topic. Maybe someone should split out our discussion and move it into a new thread. Or perhaps just rename this one to e.g. “attack stats and balance”.

1 hour ago, Prodigal Son said:

Winning an argument by all means is counter-productive. Winning an argument to propose something useful is fine though. I make such discussions for the shake of personal/mutual/general improvement, though everyone's ego can get in the way to some degree.

No hard feelings, great, we are on the same page then. Which means we can continue discussing :)

1 hour ago, Prodigal Son said:

I don't think I'm missing your overall point, with which I largely agree, I just happen to disagree with much of "your approach" towards it and some of your comments on "my approach".

Actually my opinion is that no approach is intrinsically better than any other, which is why I can be dismissive about favouring any specific approach. An “approach” implies a starting point. Balancing by trial-and-error would be an approach, but not necessarily the best. To me, lots of AI vs AI playtesting is not an approach; it's a critical (later) stage in development, the most reliable way of testing if a game is balanced. And yes, I still believe the specific approach chosen is irrelevant.

1 hour ago, Prodigal Son said:

Balance at game level is the end goal, but discussing ways to get there is also very useful and interesting to me.

Yes, I agree, interesting discussions can be enjoyable. However, as I stated earlier, I think the way to get there is actually unimportant. I tend to consider “balance at a game level” as an equilibrium. I had wanted to avoid that term, because it's quite technical, however, it's also precise, which is why I decide to introduce it now nonetheless. As you perhaps already know, an equilibrium is “a state in which opposing forces or influences are balanced”. Also, your distinction between “acceptable balance” and “perfect balance” is understandable but artificial; something is either in balance or it is not; one equilibrium is not more perfect than another. Furthermore, it's perfectly possible to have a balanced situation from asymmetrical influences and unbalanced forces, because unequal things can still cancel each other out.

 

 

Now, let's return to 0 A.D. It is often claimed javelinists are “overpowered” in A22. This can be considered as problematic at a low level (if you can win by exclusively training cavalry javelinists, then why would you want to train anything else?). However, it is not that important at a game level; although not every faction has a cavalry javelinist, every one has a javelinist (cavalry or infantry) and a cavalry (javelinist or not) unit available in the village phase.

A far greater problem is that 0 A.D.'s battering rams are not only good at razing structures, they are also capable of massacring large numbers of soldiers. An easy way to cripple an opponent is by sending in half a dozen or more rams, unescorted, against the enemy centre. Once it's destroyed all other structures will be lost as well, because they are no longer connected to a territory root (snowball effect).

Rams have a very high pierce armour and most attacks are pierce, allowing them to survive quite long; swordsmen have a high hack attack. Not all factions can train rams, which is problematic, and not everyone has access to swordsmen; factions which have both (e.g. iber) are clearly advantaged vs factions which have neither.

A way to solve this would be as follows:

  • enable rams for all factions
  • prevent rams from killing humans by inserting into its melee attack: 
          <RestrictedClasses datatype="tokens">Soldier Support</RestrictedClasses>
  • give fortresses a territory root to prevent losing your entire base when your centre is lost (adapana, Ishtar gate, monuments, pillars, and wonders already have a territory root)
  • enable constructing city walls in neutral territory for all factions (including spart)
Link to comment
Share on other sites

 Share

×
×
  • Create New...