Why Winning Should'nt Make Winning Easier

Ok, long post that a was a while in the making. Please excuse my ramblings.

I was watching some PA replays and noticed something very disconcerting. For 1v1 games, I could almost always guess who was going to win in the first 5 minutes of the game, by looking at a single number- mass income. Generally, one person will lead by about 15-30% and either increase or maintain that lead for the rest of the game. Furthermore, the higher ranked the players where, the shorter the games lasted. Many high level games lasted only 15 minutes, sometimes less!

This is not good. ​

Sometimes, though, mass income will be closer. If the incomes are within 5-10% of each other, these games go on much longer(~30 min) and are FAR, FAR more exciting to both play and watch. The whole scope of the unit roster comes into play, and we get to see what PA should be and almost is.

Even between players with very similar skill levels, only very few of their matches will go on longer, so its not just caused by skill difference. Something else causes most games to 'end' way before any commanders have even lost health.

Why? Because in PA, winning a little bit makes winning a little bit more much easier. This positive feedback loop results in games where if one person gets just a little bit ahead, especially in the beginning, they will win.

We generally call this 'snowballing', and controlling the snowball is what makes competitive games fun.

Lets look at some games that I think control the snowball very well, and see how they do it.

League of Legends:​

Arguably the most successful current esport, LoL uses very obvious ways to limit snowballing. A player that dies more and more will reward less and less gold to his killer, slowing down the rate at which the killer can increase their power. Conversely, killing that spree-killer rewards a large chunk of gold, giving the loosing player a chance to catch up.

Inner turrets are more powerful than outer turrets, and closer to their friendly spawn point. As a team looses turrets, it becomes harder for the opposing team to make more progress.

There are more subtle methods, too. Items that are more efficient gold per stat cost more, so to get the better item, a player must 'float' a larger amount of gold. The player that is behind can spend their gold right away on less expensive but less efficient items, increasing their power relative to the player who is ahead, limiting snowballing.

LoL does not just limit snowballing, though. It limits it in the early game, but actually encourages it later in the game. Player respawn timers are a very good example. Early on, they are very short to limit the damage an early death does to a player. Later, though, they get very long, so a small victory can lead to a series of larger victories through snowballing, ending the game. The 'inhibitor turrets' are another example- by killing one, the team without the inhibitor is put at a disadvantage, making it harder for them to defend other areas of their base.

What does this mean for PA?​

Right now, "winning" in PA is an unstable equilibrium from start to finish. Its like a ball on top of hill- any push to either side starts in rolling, and more and more effort is required from the looser to come back. In games where both players are evenly matched, the chances of this happening are very low and just get lower. There are no epic comebacks, very few back and forth games, only one player getting out-expanded and crushed.

To fix this, we need to make PA have a very stable equilibrium in the beginning, so small variations from 'perfect tie' naturally return. A small mass deficit should be dampened so that it cannot go on and dictate the entire game.

If this was the only thing we did, though, games would never end. As the game goes on, it needs to become more unstable. Things like nukes, SXX, unit cannons and powerful, expensive advanced units promote instability, so we already have this to some degree.

Later, Ill post some videos showing examples of this. Two brutal AIs fighting each other *should* result in crazy epic games with tons of units everywhere because neither can win, but instead on AI gets lucky, a bit ahead, and snowballs early.

Ill also post some ideas I have for solutions, and I encourage you to do the same.

add costs to attacking

The metal income is very unstable though. It's easy to effectively reduce your enemies production just by some harassment.

forgive me for lack of RTS knowledge, but could you explain the picture and how it adds costs to attacking?

I agree, that would help for sure. Would it be enough? I'm not sure. The mass from reclaiming would need to be a enough mass to give the defender a real advantage, even though they have less actual income.

I think this would need to go alongside changes that make defending significantly more metal efficient than attacking. Having units behind walls is definitely more efficient, but that can be used offensively as well.

An overhaul to static defensive structures, however, could help I think.

This goes both ways, though. You might say that the loosing player with less territorial control can defend easier, but if they control so much less territory that this starts becoming a thing, they are so far behind in mass that it doesn't make enough of a difference.

Reclaimable unit wreckage from a battle. Send units to be massacred on your opponent's doorstep, they get a bunch of stuff to kill you with.

The picture shows wreckage of mantis t1 heavy assault bots on the map four corners in Supreme Commander: Forged Alliance.
The "cost" of attacking is an indirect one: You pay for your tanks as usual, but losing tanks on enemy grounds costs you more, because the wreckages in FA could be reclaimed to get back a substantial amount of metal, I think 60%? So if you move in 100 tanks and lose them your opponent can reclaim them and make 60 tanks to attack back on top of the tanks he can make anyway.
So if you are attacked and suffer damage from it you don't fall behind as much because you can reclaim and attack back based on that.
A correct balance of metal/energy/wreckage payback yields a lot of dynamic back and forth over the wreckage fields.

Pretty sure colin's talking about reclaiming. In SupCom and TA, unit wreckage was a great source of mass for a struggling economy, and so if you fought off an enemy attack in your base, you got to reclaim all the metal and energy from the dead units to rebuild your army. In PA, units don't leave wreckage anymore, and reclaiming is far less useful than in its predecessors, which means that if you get attacked and raided early game, you're permanently set back as you try to recover what you lost.

EDIT: Apparently you only get metal from the dead units. Still, it's better than PA's system, where reclaiming is a drain on your energy.

You only reclaimed metal, not energy. Energy in FA was less of a limiting factor due to cheaper energy generation.

Wreckage also provides a stablizing feedback by effectively creating new walls at the defender's doorstep. This was a real dynamic in TA.

Except from trees. =P

Yeah but the amounts were tiny. Also trees are just there by default.

yupp with the unitammount of PA however it becomes a nightmare ... wreckgeneration needs tweaking on how heavily damaged a unit was by the time of destruction ...
at least for bots there should be a higher percentage of being heavily damaged so they rather turn into ruble more often than blocking husks
or for low armor/hp units in general

Has Uber said anything on the topic of wreckage after they removed them?

I agree, this is one thing I dislike about CS:GO too, there is no buffer.

In counter strike the team who wins gets more money, and anyone who survived last round keeps their weapons from the previous round. This means if you win last round you have an advantage next round, which can build up to where it'll be pistols vs assault rifles, or snipers.

Anyway wreckage is supposed to break pathing, that's part of the idea. Wreckage is made to stop winning a battle from making winning the game too easy. It gives a buffer so the losing player can gain back economy. TA, and SupCom both had it, and for good reason.

Ew, nova. Nah, I'm not saying it's impossible to come back, I'm only saying it's a bit odd to have no real buffer. =P

This problem really needs to be dealt with. Unfortunately I can't come up with a magical answer. You want to cut the feedback loop - the endgame snowball exists already with the nukes, halleys, death ray & unit cannon. Just that it never gets that far.

On the other hand, you want that acquiring more MEX would always be one of the priorities but on the other hand, right now it's pretty much the same priority as destroying your opponent, because that's what follows from it.

The only other system besides wreckage i can think of would be some sort of supply system that slows units down/ makes them weaker the further away they travel from their base or supply buildings, and that's like, a whole new mechanic, too much work I'd think.

Kinda like the fuel system that air units had in supcom, but for everything.

I'm all for wreckage back.

I do not think there is a problem, 1v1 is poor because of the one-planet maps, not because the person with the advantage has an advantage.