Why Winning Should'nt Make Winning Easier

Way to simplify the problem to make it look like there is no problem. How about we uncover it so that it can't be simplified like this?

See, the person with the advantage gains more relative advantage exponentially instead of geometrically. As in, 1,2,4,8,16 instead of 1,2,3,4,5... And no, this is not direct analogy, but more close to what's actually going on.

Let's say, the base value at a certain given point is 10 for both players. Then a fight occurs and player A wins. Player B base value might drop to 8 whereas other player base value will increase to 12. This is because it takes more effort to recover from loss whereas the other player made it a lot easier for himself to expand for example.

Yes, advantage is great & necessarily thing. But when you have this feedback loop, it means that games will end a lot faster than they are intended. How do we know this? Because most of what PA has to offer remains completely untouched before you can tell who wins the game. That's why it's referred to as "feedback loop".

Though you might have a point with maps that have multiple planets. Which unfortunately does not help us with one planet maps or situations where the game has ended before any significant interplanetary game play would occur.

what is wrong with the feedback loop?
all it does is better reinforce the importance of foresight, you're not playing the game in the present, you're playing it in the present and future

Ok, so I just finished a little experiment with the AI, and I will post videos later as soon as I can figure out how to fiddle with their framerate.

I had the AI play itself in 3 different circumstances to demonstrate my point on a symmetrical metal planet with 50 mex. I changed the metal production of mex to various amounts, which the AI does not account for, so it played as normal, building mex when it could even though they sapped metal.

1.) Commander produced 500 metal, basic mex produce -15 metal and adv. mex produce -30.

2.)Commander produced 250 metal, mex produce none.

3.)Everything normal.


After several different games, this was the result. In #1, games lasted about 45 minutes. One player would get ahead, make progress, destroy mex, and then get beaten back because their opponent actually had metal now. This back and forth oscillation kept going for a while, until the size of the oscillation was so large that it completely overwhelmed the opponent.

In #2, the game lasted about 35 minutes. It was very static, and remained essentially tied until 25 minutes in. One AI won a favorable battle and came out with more units, and this ball of units kept slowly growing and forcing its way into the other AI's base, eventually winning once it had literally 200 sniper bots.

In #3, the game lasted 20 minutes, but was decided in 8. One AI got a little bit ahead, and continued to box in the other player until the mass income was 284 to 23.

Obviously, this wouldn't actually work, but it demonstrates that vanilla PA is highly chaotic, with a slight variation in initial conditions totally dictating the outcome. Ideally, we would want the game-play to start out similar to 1, and progress towards 3.

But how?

I think maybe, a combination of better, more mass efficient defensive strategies, reclaimable wreckage, and finally some way to slightly decouple territory gain from mass gain would do it. Maybe slowly increasing the mass production of the commander over the game, or some clever tinkering with changing the mass production rates of mex over time along with t2 mex doing something else.

Maybe even giving an inefficient energy->metal conversion structure.

All I know is that we really need to think of something, and it needs to be done in a way so that the winning player cannot take advantage of it to push themselves even farther ahead.

The looser needs a real chance to come back, and not just via snipes, otherwise games will be very predictable and decided very early on.

I described what is wrong; the fact that matches have pretty much the same dynamics in them as most basic RTS games out there, because the more interesting features of PA won't get any action.

And feedback loop does not increase the importance of foresight. Winning those early fights should be important, but right now, it's pretty much all that counts. If you're failing on all of them, you obviously shouldn't win, but right now losing one fight means you're less likely to win the next one as well.

Because the feedback loop actually reduces the importance of foresight. All you need to do is focus on getting that first early advantage and winning that first trade. Once you do that, everything gets easier and easier.

It doesn't matter if you have these genius plans about intentionally ignoring his right flank so you can later build a teleporter there and smash his unprotected factories, because his first bomber sniped your fabber and killed a mex while yours flew too close to his com and blew up. He has twice as much Eco and twice as many tanks because you have been loosing slowly ever since.

When have you played a (good) game that gets easier as you go along? Think of Mario games, strategy game campaigns or even the halo games. The first levels are easy, and they get progressively harder as you get closer to finally winning, culminating in an epic boss battle that takes 15 tries, multiple visits to gamefaqs for help, and two broken controllers.

As it stands, PA is the opposite of this. The first level is the hardest, and it only gets easier from there.

is this not how fights work?

That's one way of how fights can work. But it practically simplifies the whole game into a very narrow concept of what it's supposed to be. Fights can be about who wins the first tradeoffs with unit groups or about who has the best strategy accompanied with the fights. When you have exponential feedback loop like this, it never gets to the point where those other decisions would matter. It just stays around who wins the earliest fights. Chances for comebacks are mitigated.

And I argue that they're not supposed to be like this because there's plenty of content that remains unused for this reason alone. I am very certain that devs would have not implemented all that if they don't want it to be used. You can argue back against this by saying that they get to be used in other scenarios than 1vs1 with same starting planet. But this scenario is universally the most preferred one, I would argue you back.

@stevenrs11
There are two separate things you are raising. Contested game length is too short, and there is too much snowball. Let's address the latter first.

Snowball
Relevance of the data you are using
When players have equal skill level, their metal income will generally be equal together, and when players do not have equal skill level, the metal will not be. There is nothing wrong with player skill being a key determinant in the outcome of a match, and infact this trait is desirable if not necessary. However, this does not mean higher skilled players are guaranteed wins as i will explain later on, only that they are naturally favoured in one or more aspects due to their own player actions having meaning.

When you watch snowball matches with unequal metal income, you are most of the time observing the result of unequal skill matching. It isn’t because of metal income that someone is winning, it’s because they are outplaying in all other areas that they are able to get this advantage. You are mixing up the effect with the cause. This information problem is enhanced by the current lack of ladder transparency that does not allow replay watchers to understand the relative skill levels of the players entering the match.

If I was to watch a pro chess player crush an amateur player, should I conclude that chess has too much snowball? Or even with two amateur players playing each other. Would I conclude that the 4 move checkmate thing is imba and gives too much snowball for white simply because neither player knows how to use all the tools readily available to them?

I have had plenty of epic games with players of equal or greater skill that know what they are doing. Some of them with the goal of purposefully allowing the opponent to get a huge metal advantage, only to crush him with a unit timing, sometimes with teleporters in his base. Doesn't always work,

It is inaccurate to say that epic comebacks are rare or uncommon let alone non-existent. They happen often, and I would cite the most recent example from memory being the PAG vs Promethean matches in clanwars yesterday night. Though important, simply having more early metal income is not as significant as it may at first seem.

Mechanics design
There are two classes of mechanics that are usually used to address issues as identified by the designer. I'll give examples from other games, and then an example of how it is already significantly implemented into PA. Maybe it will give you some ideas about how to make use of everything that is available in PA so that you can have better games.

Fixed - These are artificial mechanics that serve to level out skill differences by reducing the relative impact of any variable mechanics that lend to snowball.

1. Unearned income
   • High natural, unearned income in LoL as opposed to Dota
   • In PA we have commander starting income that is relatively high at 30/3000. Pre-10 minutes it makes up a significant share of your total income, and by 10 minutes it still usually makes up 15-30%.
2. Number of easily obtainable and defendable resource points
   • Natural expansions in an RTS like starcraft
   • In PA, we have relatively high amount of easily defensible 'natural expansion' metal on ladder maps. In addition to the reclaimable trees, players get a free ride to 100-130 metal income and generally keep this fixed base for the rest of the game
3. Power of static defense
   • Towers in MOBA games, whether they be static in power like in dota or variable in power like in LoL
   • In PA, we have very strong turrets and walls. Most importantly, they are cheap, which allow players to hold off large swarms of units without needing either large swarms of units themselves or large economy time investment

Variable - Player driven come back mechanics are not always snowball. They do not rely on directly controlling the snowball like a fixed mechanic in order to address the problem, and instead add dynamism to gameplay.

1. Player skill come back
   • In CS:Go players can pull clutch multi kills to get back into a round
   • In PA you can use good unit control, engagement, positioning, composition, map awareness, and harass. These include multi prong attacks, inferno drops, dox harass, bomber harass, mines, as well as system objective prioritisations such as annihilaser or halley.
2. Force multiplier and cost/benefit come backs
   • In starcraft, these include forcefields, psi storms, banelings, mines, ravens, mules, fungals, muta. For example, landing a good psi storm can kill alot of enemy units, or good forcefields can drastically reduce incoming damage.
   • In PA, these are wall/turrets, anchors, combat commander, infernos, unit composition counters, surgical bombers, and to some extent t2 mex.
3. All-ins and objective snipe come backs
   • Not many games have these to as large an extent as PA. The best examples i can think of are starcraft all-ins, worker snipes, or lane/throne sniping in dota (less viable if at all). Afaik, there are no objective snipes in LoL, only stuff that reduces individual player impact on the game.
   • In PA, there are an incredible number of different ways to still win when behind, with each way requiring a totally different type of investment for the other player to defend against. These include boombots, sxx, inferno drop, bomber snipe, halley, annihilaser, teleporters, anchors, nukes, holkins.
     
     Without scouting, you cannot blindly defend against all of these, creating a natural state where the economically advantaged player is never resting on their laurels.

League of Legends
League of legends is popular not because it is a good game with good mechanics, but because it is a casualised game that appeals to the lowest common denominator. It implements fixed mechanics and design processes that artificially lower the skill cap while at the same time increasing the skill floor, so that everyone is experiencing the game within a narrower bandwidth. The tradeoff for this being lower player determination. For example, LoL has a very high ratio between unearned income and earned income (creep kills) in comparison to a game like dota.

Notice however that this does not create differences in game length between the two, and infact the greater snowball in dota that comes from the emphasis on earned income is actually a desirable trait that lends itself to gameplay quality, both spectating and participating.

LoL hides this weakness behind an overabundance of heroes, giving the illusion that the game is deeper than it really is. LoL also makes up for the lack of dynamism by purposefully and artificially changing the "this hero is op" metagame on a periodic basis.

I'm not saying whether this makes it a good or bad game, only that LoL's popularity has not much to do with real gameplay depth. It's a casualised cashcow designed ontop of a skinner box that supports their f2p model. Any game that has a large base always plays heavily on some psychological aspect of human behaviour.

The esports aspect is simply the organic outcome of having a huge casual playerbase within a competitive multiplayer game. Esports is not a significant design goal of LoL, but rather just another element that supports the growth of the casualised playerbase. If this segment makes a profit, great, but it could also function as a cost center and still provide net benefit.

Gamelength
With game length, context is everything. Less snowball does not mean longer games, nor do longer games mean more interesting ones. Snowball is not the main lever you pull to affect this variable. It is affected by

1. The genre of the game, the expectations of the playerbase, and it's repeatability. Would you say that the game length of tic tac toe is too short in comparison to LoL? Can you even draw a meaningful comparison here?
2. The nature of the game as an issue of artistic judgement by the designer. Staleness vs pacing of interesting events.
3. The manner in which the game is played 1v1, 2v2 etc. and map size.
4. The skill level of the players involved and the pace at which they accomplish tasks.
5. The relative strength of strategies available to players (rush vs macro)

Could PA use with new units, new balance, new mechanics that support more ‘epic’ games? In some ways yes. PTE addresses some big problems with the current stable build and you should check it out .

Should PA directly curtail relevance and importance of player actions in the same hamfisted way that LoL does? No.

That's a great post though there's one concern I have regarding your rhetoric. The given example in LoL was not referred to as "this is what we should aim at", but rather, it was an example of how, for example, LoL dealt with snowball issues. You seem to take too much focus on the given example because LoL, as you said, is highly oriented towards casual players and that's something you would hate to see for PA to happen. Too much emphasis on this given that OP presented LoL as a description, not prescription.

In short; your post could have had much better fruit if you didn't cater around LoL.

And the "variables" you presented are ideal for your case because they're predetermined and give the impression that these cannot be changed anymore, because either they are determined by genre, developers (where you assume that status quo = what developers wanted, something that I would question in my previous comments) and the skill of the two players along with extrinsic condition of map layout (which determines most of relative strength between strategies).

You do not actually present anything to support the claim "snowball is not the main lever". And when you present these variables, it seems really that you're trying to shoehorn this very specific sense of how this problem can (or more specifically) can't be solved by implying that they would have to change the genre of PA, disappoint original artistic visions (assuming they match right now), disappoint original developer choices that supposedly led to this (assuming this is what they wanted to achieve with them) or more clever map design. And player skill can't be affected directly at all, so no go there.

You had good thing going in the initial analysis part, but honestly, it feels like your conclusion was completely distorted by the LoL example and resulted into extremely biased representation of the problem. Have you read the suggestions in this thread?

I want wreckage back! yes!

Not sure if this would work, or maybe try the exact opposite, just something to through into the mix.

In Dawn of War 1 the 'metal' points would gradually produce less metal. amongst other things this encouraged expansion, and also stopped it becoming a slug-fest as your army inevitably got smaller.

should resources be limitless or could you use up a metal point a la starcraft minerals?

Give us at least flat (pathable) wreackage please

eh?

I addressed that specifically to the OP. It's pretty clear what's going on here in that regard. OP dislikes snowball and high degree of player determination. He cites LoL as an example of a game done right without understanding why they do it the way they do, what the costs of that method are, and why it doesn't really apply to games which seek to somewhat intellectually stimulating.

You don't give examples without also giving direction, and it's quite obvious that he's looking for more than just adding wreckage. He wants advantage gained from skill to matter less by arbitrarily adding mechanics to level differences between players.

Also don't think you understand why i separated the concept of game length from the concept of snowball. Yes, more snowball means shorter games than normal and vice versa. But these are just descriptions of change. What actually defines that normal, average, game length?

Nominal game length is subject to artistic judgment as well as all of the other factors that i listed, and maybe some other ones that i did not think of. OP cannot simply say that because a game goes for 15 minutes, it is too short. Really as simple as that. Snowball is not the main lever, in exactly the same way you do not mow your lawn with a pair of scissors. It is entirely inappropriate.

People always have this idealistic thinking of something along the lines of, oh i want this epic battle for 1 hour and i'm going to do this, and he's going to do this, and i'm going to do this, and it's going to be so awesome because it's in my head.

Well yea, great, but why does it need to be over 1 hour? What if we were able to get that same ammount of epic and fit it into 20 minutes? What about players who don't have that much time, or the players who don't want to be mentally exhausted after every game? What if the game length and pacing isn't appropriate given the circumstances, the genre, and the way the game is designed to be played?

Would you think 1 hour of the same tic tac toe game is epic?

There is also another major difference between MOBAs and a regicide style RTS like PA : snipes. A player with lower metal can often put all of their energy into a snipe if the other player makes a mistake or simply doesn't focus sufficiently on a specific area. Case in point the last 1v1 I played my opponent stopped building naval forces towards the late game and let my ships move in so that he could finish rushing out an SSX. I'd seen all of his base except the orbital launcher so didn't know to expect it, I had no anchors or orbital radar. Unfortunately for him he then put his commander in an astraeus and flew it over my narwhals, but his snipe would have worked.

I'm not saying that snipes remove the need for any other modifications, just that it's an aspect that's overlooked if one player is thought to be "clearly in the lead".

Reintroducing reclaiming would be great. I especially like the suggestion that most low armour units should leave rubble, with maybe only heavier T2 units and ships leaving actual wreckage.

Final point : winning should make winning easier, if only slightly. If we take a non-esport example, in football being a goal ahead makes things easier, you don't have to be aggressive and thus you can take less risks. Some people seem to be suggesting being ahead in PA is more like playing football in a game where the other team loses a player every time you score a goal, but I don't think the lead is that huge. A bit of raiding and you can slow down their building and cause them to stall harder because they've built too many factories. The main thing we want to avoid is rubber banding of a sort between players, making life easier for a player because they're losing is just frustrating for the better player and just frustratingly increases the required skill gap for one person to win.

@elodea

I get this feeling LoL isn't your favorite game, heh. Its not mine either, and I used it as an example of 'snowball' control because it is very heavy handed, therefore very obvious. Also because it doesn't just limit the snowball; it encourages it later.

For your points on game length, I see what you are saying. The thing is, in games that last under 20 minutes, you rarely see significant use of more than half of the unit roster. The real problem though, is that games often end far before the commander explodes. That, no matter how you look at it, is not good.

The reason I did my tests with the AI is to avoid player skill confounds. Two brutal AIs playing against each other should not result in one AI getting absolutely crushed for 3/4s of the game.

Do you have to let wreckage effect pathfinding? Make it so units can move through them like they are not even there. Win/win I like the way building wreckage is now.. the way you have to move around it but you can destroy it. I'm just talking about moving through it for unit wreckage.

Ofc not. For buildings I also agree. When I made the screenshot for FA I had the mantis walk straight through big and tall looking structure wreckage before. I was like "oh wtf that was a thing" xD

Even though we have snipes, it's unfortunate that they are considered the only way of a comeback in PA. It also requires either a rather good knowledge of the ENTIRE PA unit rooster to develop one which isn't countered by default.

This is just fine when you play above a certain skill level, but if you play around the skill floor, players expect to be able to push back in a symmetric scenario as well. And that is something PA doesn't allow you. Once you are cornered in one domain, that domain is lost for you as you usually need an force of equal size to cause any significant losses. If you go up with 10 Dox against 20 Dox, your 10 Dox will be shredded, while you take out 3-5 Dox at most, even less if the defending player rotates his units.

And then there is also the issue that if the action has reached the border of your base, you are usually getting steamrolled. Once the enemy has reached a critical mass to break your perimeter defense, any interior defenses are futile as well and your commander itself is the second next thing which would be capable of resisting in any way.

Deathballs are still a thing in PA, there's no reason in denying that. The are not invulnerable, but they are almost impossible to slow down in any way.

Rubber banding or diminishing returns aren't a bad thing in general. That's most important when it comes to assaults on bases, when you want to make sure that the effective penetration depth of an assault can be no more than linear proportional with the army size. Diminishing returns can prevent a critical mass from happening, and thereby allow for a second round.

It's not as if the defending player could make up for the losses either way, but it's essential for the morale to give him a chance of surviving without abolishing his base entirely.

This would be a different case if PA would properly promote proxy bases in chokepoints, so the enemy would be forced to take these down one by one, first. But since you can attack from almost every single vector, proxy bases for defensive purposes just don't make any sense. Even less so, if you consider that all orbital units completely IGNORE any land control with their pin point accuracy on orbital insertion maneuvers.

That's one way of interpreting it. But OP did not say "it was done right in LoL", that is, unless you can provide me with something that would convince me otherwise. I did not see where OP used LoL example as a direct prescription.

And snowball does not equal "high degree of player determination", it just affects it's distribution along the time table. The way it works right now is that most of significant player determination occurs, as OP described, before a single projectile would hit any commanders. At the point they do so, it's usually obvious who wins.

Ever seen this picture?



Now you have. Poisoning the well of OP is not really adding up to any argument - you're just making assumptions that OP has motives to casualize PA extensively like LoL. When someone is pointing at the moon, you shouldn't look at the finger but the moon itself.

Yes, OP can argue that by mentioning that half of the unit roster (that can be deemed effective) remains unused for this reason in most games. You're also assuming that 15 minutes is the nominal game length that was envisioned in the artistic impression - something that you cannot back up without quoting devs themselves.

The "same amount of epic" is not included right now in the 20 minutes because a lot of the unit roster remains unused. It's really simple as that.

Sure, if it's in space and you have death stars as well as celestial bodies that can be tossed around, units that can be blasted across the star system to another planet and so forth. However, if we're talking about classic tic-tac-toe, then we're talking about a game where first four moves will decide if it's a draw or X will win. So if you manage to turn this game into 1 hour long match, at least one of the players should be heavily handicapped mentally. So this isn't quite clever example at all. More over - if we have current average length of 20 minutes which would be turned to 60 mins (again - nobody said it should be 1 hour, just let's assume this), this would relatively not translate to 1 hour long tic-tac-toe games, but rather, triple the current average time. So if it would be 30 seconds, it would be then 90 seconds. But the additional 60 seconds don't provide you with anything new to play with either, unlike in PA.

This is certainly one of the best forum entries I've read since I joined a few months ago. All hail the King!