Dawn of War II Reborn - Artillery's "Atlas"

Cross-posted on Gamasutra

There once was a game…

A few months ago, I learned the talented shoutcaster Sean "Day[9]" Plott was hired by the rising indie studio Artillery. While unexpected, I wasn't surprised, given his extensive knowledge of Starcraft. I saw it as an interesting development, but initially I didn't get too excited about it.

Curiosity would later pull me back to learn more about Artillery's Project Atlas. That's when I did get too excited, as shall soon be apparent. They weren't just making any RTS. Without knowing it, they were creating the spiritual successor to one of the most underrated RTSes ever made - Dawn of War II. Not only did Atlas share a lot in common with its ill-fated predecessor, but all of the challenges they now face (as highlighted in their blog) were ones that DOW2 not only tackled, but overcame in a spectacular fashion.

So I thought to myself “man, it would be awesome to work on a game like that!” My mind raced with the possibilities, analyzing and breaking down all the things I loved so much about DOW2. Then a crazy thought popped into my head: why wait to work there? Why not blog about it now, and evangelize to the world just how great DOW2 was why the world needs more games like it? Why not nerd out about what makes my favorite RTS so special? Thus, we have this blog.

This will be equal parts open letter to Day[9] and the Artillery Team, a response to their design blog, and a springboard for some fun discussion of RTS game design. Seasoned, of course, with some healthy fanboying over Dawn of War II. I truly believe DOW2 had the potential to be one of the best RTSes ever made, but was tragically held back by not having sufficient tutorials to help its players understand its multiplayer. I intend to correct this error here.

Full disclosure, I did apply for a job with Artillery recently, hence the craziness of doing something like this. But I figure, screw it, if this is the closest I ever come to setting foot in their studio, then I’m going to go all out on this one.

To fully understand what DOW2 did and why it worked so well, lets start with some general ideas around RTS game design.

Golf, Goals, and Obstacles

In Day[9]'s first design article, he describes the game design process as one of “making your life hard”. He shares one of his professor's stories about golf. In it he describes how putting a ball into a cup isn't that exciting, but that adding additional “inconveniences” made the experience more enjoyable. Hitting a tiny ball with an awkward club at a target many yards away all made the player's goal more elusive. Placing challenges between the player and their goal made that goal more satisfying to achieve. While I think this is a good way to describe the design process, I would argue that perhaps we aught to consider a different perspective.

The development of golf started when people realized hitting rocks with sticks was fun. To make this more fun, they experimented with different ways to hit the ball: different sticks, different rocks, different targets, etc. Each feature became a layer that enhanced the the players enjoyment. Each made it take require more finesse, have a cool sound, or give the player more control. The goal of getting the ball in the hole was merely one such layer. Despite what some players and designers might say, the true heart of the game is not getting the ball in the hole, but what that feat represents - the culmination and validation of one’s time and effort in honing their skills. Each layer is a lens to magnify the enjoyment of what began as a simple activity.

Writing a story works much the same way. Starting with a focal character, dilemma, or setting, the author adds layers to flesh out the world and enhance the emotions they are trying to conjure up. A writer might ask "I want to make a stone-cold badass, so what kind of scenarios can I put this guy in to make him more badass?" So perhaps they could add some rivalry with an old partner who killed his daughter. Perhaps his daughter was blind, and reminded him of his ex-wife. His goal of revenge sets the stage for resolving this escalating emotional buildup. Each new layer adds to the emotional weight of a character, action, or scenario.

Constraints as Focusing Devices

Another way to look at constraints is not hurdles to overcome, but the natural characteristics of a fictional world. For non-abstract games, the designer is taking all of human experience and cutting it down to just a handful of familiar activities. The choice of what gets included and what does not dictates the focus of that game, and thus the actions available to the player. Like in the real world, if done correctly, these "constraints" do not feel forced or limiting, and instead fade into the background and become invisible. In our daily lives, we do not think about the "inconvenience" of breathing or the strain on our legs to keep us standing upright - we intuitively overcome these challenges unconsciously and without anxiety.

The challenge is to leverage player expectations around the familiar elements you have chosen . The very act of reducing the world in this manner is itself a form of escapism, as players are transported into a simpler, more predictable world. A world players can easily wrap their head around, and where they have a strong sense of agency in their ability to control it.

Consider the times when you played a game and came across a knee-high wall you couldn’t cross. It was frustrating, wasn't it? Then think of something like the original Super Mario. Did it bother you that you couldn’t punch koopas? Did you feel shackled by the inability to talk and discover the backstory of a Pirahna plant? No, because the world was presented in such a way where your mind could intuitively define a range of possibilities of what you could and could not do.

Abstract games, or abstract game elements, obey the same dynamic. Often they too are inspired by a small subset of the human experience. The further from reality they become, the more you must take a bottom-up approach in presenting them. Like a newborn child experiencing the world for the first time, you must start small, and build upon the players knowledge until they can model the entirety of this alien world within their own mind.

Story Arcs

The topic of Day[9]’s second and third posts is story arcs in the context of a multiplayer game. He describes the desire to have a strong beginning, middle, and end, and how to make each of these key points feel significant, while steadily growing in intensity. The challenge, as he rightly points out, is how do you make these key peaks in the story arc happen, and how do you keep people engaged in between them?

I think this dynamic can be approached from two schools of thought, which I will call the mono-arc and multi-arc approach.

Mono-Arc

A mono-arc story follows your traditional story slope - steadily rising tension, punctuated by moments of great intensity, building to an ultimate climax. The theory is to spread out your peak moments so the audience does not get desensitized, as that would diminish the impact of major plot reveals. At each peak you give your audience a taste of the good stuff, then withhold it from them and have them follow your narrative path until they reach next one. This is a perfectly reasonable model for thinking about narratives, but I feel it is vulnerable to misuse, particularly in games.

Sometimes you end up with a game where the player is stimulated during certain peak moments, then is handed little more than breadcrumbs during the downtime. At its worst, the game teases the player, padding out the length of a game with hurdles and filler, demanding the player "earn" the right to be entertained (as if shelling out their hard-earned cash wasn't enough). Sure, the tension may rise over time, but player retention ends up resting perilously on their patience. Particularly in a modern world of Youtube, Netflix, and a million other distractions, why should I slog through 5 hours of random encounters just to see what Sephiroth is up to next? I suspect this is was the fear one commenter had in rejecting Day[9]’s “Making your Life Hard” theory.

Multi-Arc

An alternative is the multi-arc approach. I came up with it from reading “How to Win Friends and Influence People." In its first few pages, the reader is instructed to take their time reading the book, and set it aside when it inspires you with thoughts of your own. In following this suggestion, I discovered why people celebrate books so much. In these moments of mental reflection, the reader experiences a very different kind of stimulation from other media, a kind that relaxes some parts of the mind while exciting others.

This got me thinking that perhaps the best way to approach engagement is to not think about a single arc, but many parallel arcs, each representing a different kind of activity or emotion. As one arc would dip, another would rise, keeping the overall experience consistently stimulating. This bypasses the problem of desensitization, as you are not dialing back stimulation altogether, merely tuning in to a different frequency.

RPGs offer the clearest way to visualize this, as they often broken down into distinct and separate "modes" of play. In Mass Effect, for instance, you would have intense firefight one minute, then be talking galactic politics with an NPC the next. As the combat arc declined, the narrative arc ramped up, picking up the slack. The overall engagement level never dropped, it merely changed its flavor.

The beauty of RTSes is that they already follow a multi-arc structure naturally. Within the span of a few seconds, the player is transitioning from scouting, building, upgrading, moving units, formulating plans, microing abilities, and planning expansions. The player is never locked on a single thought for long, transitioning between many different strategies, tactics, and sub-tactics in the blink of an eye. It turns the seemingly chaotic and manic task of juggling memory, critical thinking, visual perception, and physical coordination all at once into something exciting and engaging. There is never a dull moment, as each decision and action tickles and prods at every corner of the brain, without any single part burning out. For this reason, I regard the RTS genre as having the best gameplay of any other genre out there.

The key to using the multi-arc to its full potential is to provide a variety of different stimuli to your player, as well as a broad range of strategic and tactical choices at any given moment. This is something Dawn of War II did phenomenally well.

Equilibrium and Cognitive Dissonance

Day[9] cites Settlers of Catan as an example of "equilibrium" dynamics. He describes its transparent scoring system as one that naturally nudges players toward withholding trade and ganging up against players in the lead. This produces a natural balance to the game without any explicit rule to enforce it. This dynamic helped most games remain close down to the very end.

If we look a little deeper, we understand that this dynamic is based on eliciting a sense of fear, while giving player's readily available tools to handle the threat. Every good strategy game uses this in one way or another, but when pushed to its extremes, things get really interesting. Cosmic Encounter offers one such example.

In Cosmic Encounter, there is a much wider range of power levels in its races than you would see in most games. Some races have relatively modest powers, like the Pacifist, who can win a battle by playing a negotiate card (which ordinarily causes a loss). Others, like the Virus, seem completely overpowered. Ordinarily, combat is resolved by both players playing attack cards face down, revealing them, and adding the number of ships on their side to the total number on their card, with the highest total deciding the victor. For the Virus, their power allows them to multiply the card’s value by the number of ships they had in the fight. So a measly 10 attack card backed by 4 ships could turn into an attack of 40 - matching the highest attack card in the game.

When I first (terrestrially) encountered the Virus, I was convinced the game wasn’t playtested. How could you possibly win a fight against those guys, let alone against races with arguably even stronger powers? I was turned off from the game for a good 10 years. I later learned that Cosmic Encounter has been around since 1977, and has been reprinted at least three times without  any fundamental changes to its core mechanics. It is still going strong today, and even has its own convention! So I decided to give it another look.

What I discovered was that there were in fact many mechanics that constrained any potential inbalances, including:

• Random assignment of each player's race (from 50 different choices)
• Attack card values range from 0 to 40, weighted more on the lower end of that scale, so your chances of getting a bad card at some point is high
• Players only draw new cards under very specific circumstances, so getting stuck with bad cards can be tricky to deal with
• Players can lose an encounter by playing a negotiate card when the Virus plays an attack card. They lose ships, but as compensation they can steal cards from the Virus player, potentially weakening to weaken them
• Players cannot control which player they encounter with, so a powerful player can't bully any one particular player

And most importantly:

• All alien powers are public, which means everyone knows just how powerful someone is. They can use this to either get the other players to turn against an “overpowered” player, or tactically ally with them to ride their coat tails to earn a few easy colonies.

In other words, with the right combination of mechanics, coupled with an equilibrium dynamic, you can allow players to wield far stronger abilities, well beyond what conventional wisdom would call “balanced.”

It was this very dynamic that was at the heart of the Dawn of War II experience. Unlike Starcraft II, DOW2 was balanced a bit more heavily in favor of hard counters. Yet it did it in a clever way- even though any individual unit had a clear counter, the player often had the choice of what they wanted their unit to counter through their choice of upgrades.

The best example of this is the Tactical Marine. As a heavy infantry unit, it was pretty beefy against most things, but had a hard time against melee weapons and plasma guns. It could be upgraded with a flamethrower (to counter light, swarmy infantry), a plasma gun (to counter other heavy infantry), or a rocket launcher (to counter vehicles). Likewise, even more specialized units, like the Assault Marines, could be upgraded to have a secondary support role. While they ordinarily countered artillery and weapons platforms (like machine guns), they could be upgraded with a bomb that allowed them to soften up and slow enemy vehicles for your main anti-tank weapons to take them out easier. It is the simple but weighty choice of what units and upgrades you got that was at the crux of the game’s tactical experience.

DOW2 did not go out of its way to keep itself balanced, and instead merely handed the tools to the player to balance it themselves. The first brush with this dynamic was often when a player first encountered a tier 2 mech unit, like the badass Wraithlord. They found out quickly that their puny tier 1 infantry could barely leave a scratch on it, and that it could inflict heavy damage if allowed to come into melee range. As scary as the Wraithlord was, they were not that hard to deal with - they walked slowly, and could be taken down fairly quickly with a few well-placed anti-vehicle weapons. Of course, this fear could come right back if your opponent destroys those anti-vehicle units on you. Thus was the rollercoaster ride of your typical DOW2 match.

Getting the most out of equilibrium dynamics involves leveraging cognative dissonance. Consider the thrill of riding a rollercoaster - your irrational side is terrified of falling off and dying, but your rational side knows you are safe. This is the mindset you want to put your players in. Make them afraid, yet confident they can overcome what's in front of them. That way when a player is facing down a massive superunit that requires an entire army to kill they know they can outrun it, dodge its special attacks, and defeat them with minimal casualties by dancing their injured units around. This makes players feel empowered as they take down what feels like a massive threat.

The most important thing is explaining this dynamic to your players clearly so they always know what they can do and how to handle such threats. Otherwise you risk your players ragequitting, uninstalling, and declaring you game imbalanced. This was a lesson Relic learned the hard way.

Speaking of dodging, lets talk about Skillshots.

Skillshots

Skillshots are one of several interesting mechanics Artillery is considering for Atlas. They seem to be going for a hybrid of a tactical RTS feel with a few MOBA mechanics thrown in. I think this is a great idea, and in fact DOW2 experimented with this as well, though had its own twist on it. Before I get to that, lets step back a moment and trace the evolution of this mechanic.

Warcraft III featured an early version of skillshots in the form of line-based area effect abilities. There were only three of them in the game (the Tauren Chieftain’s Shockwave, the Crypt Lord’s Impale, and the Pandaren Brewmaster’s Breath of Fire), and they weren’t quite what we would call skillshots today. You had to target a unit to aim them, which mean that the “shot” part of it really wasn’t there yet. They were essentially little more than an area-based attack,  resolving too quickly to be dodged, limiting the tactical depth of executing or defending against them.

Dawn of War II took it a step further. Unlike WC3, you could aim at the ground to line up and time your shots in anticipation of enemy movement. Most of its skillshot attacks had a windup period, with a visual telegraph showing when it was coming. Thus the onus was on the defender to dodge them. Since units moved fairly slowly, players needed to react quickly to avoid getting hit. Since these were mostly area of effect attacks like in WC3, their system focused more on punishing the mistakes of the defender than rewarding execution of the attacker.

MOBAs like League of Legends perfected the skillshots to what we know today. Not only can you aim them at the ground, but units (Champions) can move pretty quickly, and there are plenty of minions and terrain around you can duck and weave around to evade them. Thus both the attacker and defender feel empowered by the mechanic. Both are rewarded for success and only modestly punished for failure. It is worth noting, though, that unlike DOW2 or WC3’s “skillshots,” LoL’s were mostly designed around hitting a single target, rather than being used against large groups.

While we're on the subject of player empowerment...

Randomness and Player Agency

Going back to Catan, lets talk about randomness. I’m not a huge fan of Catan myself, as I feel too much of the game hinges on a random dice roll. Sure, people can mitigate it a bit in how they conduct trade with other players, but for me I just can’t get over that RNG element taking agency away from players.

That’s why I love Dominion. As a deckbuilding game, it balances randomness and player agency in a really clever way. Over the course of the game, players choose which cards to buy to add to their deck. Each choice a player makes directly impacts the probability of what kind of hand they will get. Want more reliability? Buy a Village. Want to buy better stuff sooner? Buy a Silver. Want to trash your weaker cards so you have an awesome hand every turn? Buy a Chapel. Every card allows the player to directly impact probability.

Dominion also has interesting scoring mechanic, which synergizes well with this idea of probability control. You win by collecting victory cards, which each have a point value on them. The player with the most victory points at the end wins. The problem is that most victory cards (with rare exceptions) are worthless during the game, and don’t do anything but clog up your hand. Thus you have to choose carefully when and which victory cards you take. Get overzealous early game, and you’ll have a hand full of victory cards and not be able to buy anything. It also acts as a balancing mechanism as players that are ahead “pollute” their deck with useless victory cards, allowing other players to catch up.

Player Decisions in Randomized Scenarios

In addition from the RNG you get from drawing cards, when the game is set up you randomly select 10 kingdom card types. These will be the 10 type of action cards players can choose from during the game. While completely random, this dynamic is carefully balanced in a way that still retains strong player agency.

The RNG in the game's setup creates a situation for players to react to - it does not decide the outcome. Players choose how they react to what is available, and tailor their strategy around it. Its not unlike the randomized map rotation of something like Starcraft II, where a player would study the map to plan out their expansions, identify choke points, and determine which map features best suit their race's unique traits.

Compare this to something like D&D, where the player chooses what they want to do, then rolls a die to see if it happens. Execution is handled by RNG, robbing the player of a sense of ownership of the outcome. In Dominion, the consequence of a player's actions within its randomized scenarios are entirely their own.

The Power of Variables

Variables are the key to a great strategy game. Both variables and RNG contribute to emergence - the generation of novel scenarios and interactions within a game that are not pre-scripted. Both add uncertainty to a game, but the key difference is that variables are things a player can control, and RNG is not.

Part of what made Starcraft II the massive success that it is was its maserful use of variables. Controlling where every unit goes, what units get produced, where your units are positioned - everything is controlled and decided by players. These choices leave artifacts behind, usually buildings, to give clues to scouting opponents of what to expect. Rather than being a game where anything could happen, observant players could control or predict the probability of certain outcomes.

It is this strong sense of agency that players really crave out of their competitive games, knowing that their actions are the deciding factor in their success or failure. It is the reason why Starcraft II and League of Legends achieved the level of success they enjoy now, and games that used too much RNG (like Dawn of War II and Company of Heroes) did not.

Player Expression

Beyond just giving players more control over the outcome of a match, variables enable player expression. With the right number of variables, players can develop their own unique playstyle, and allow for metagame trends to emerge. I hear Day[9] mentioning all the time “Bomber’s style” or “Scarlet’s style” when he describes pro matches of Starcraft II. Being able to put a signature to your play makes a big difference in how the eSports community responds to your game.

To pull this off requires the right amount of variables tailored to allow for a wide range of viable strategies. This is where Warcraft III fell short for a lot of people, as if you watch enough pro matches you will see that players of a given race will almost always have the exact same build, with the only difference coming down to how they micro their units.

SC2 allowed for much more player expression compared to WC3 thanks to a few subtle but important design choices. Units in SC2 move much faster than in WC3. This means that there is a lot more “physical” micro of units moving around. It is far easier for a casual spectator to see and delight in physical movement of units, as it reminds them of players running down a field in a sports game. Warcraft III was about targeting abilities and microing of injured units out of harm's way, but lacked this dynamic “dance” of well-microed units you see regularly in Starcraft II matches. Enhancing this one variable in SC2 made all the difference in its popularity, as it allowed micro to be much more visible to eSports spectators.

Player Interaction and Resources

This brings us to Dawn of War II, and how it put the above theories and dynamics into practice. A big part of the pacing and strategy of the game revolved around its resource system, which worked very differently from just about any other RTS out there.

Overview of Resources

DOW2 featured four main resources: requisition, power, victory points, and “red.” The first two dealt specifically with building units and upgrades, while the second two dealt with controlling the pacing and escalation of the game. All four of these were governerned by dramatically different dynamics, aligning nicely to the multi-arc principle I talked about earlier.

Each resource (aside from Red) was obtained by using infantry units to capture and hold nodes on the map, passively increasing a player's income of that resource. Its worth noting that while held, these nodes gave a small area of vision to their owner, working as a good substitute for traditional scouting. It allowed players to see what upgrades and units their opponent was using whenever they attempted to steal a node.

These nodes could only be captured using infantry units or commanders, often your most vulnerable units. Combined with DOW2's focus on small armies  (typically maxing out at just 4 to 6 squads), your units could not be everywhere you needed at any given time.

Requisition

Requisition is the most basic of these resources, like the minerals of Starcraft II. You gained some automatically from your main HQ building, but could obtain more by capturing Requisition nodes. The income each provided would  “mature” if held long enough, yielding up to triple its original value.

In a typical match, Requisition was often overlooked in favor of the seemingly more valuable Power and Victory nodes. Their nodes were both the easiest to capture, yet also the easiest to steal. They were a resource that could be easily taken for granted, and thus players exploiting the maturation mechanic could do some serious damage to their opponent’s economy if they just “decap” their opponent’s nodes, if not capture them for themselves.

Power

Power was the premium resource of DOW2, like gas in Starcraft II. It was used to fund high-value units, upgrades, and to unlock higher-tier units. They could be reinforced by "activating" them, causing a structure to built over the top of the node. Opponents would have to destroy this structure before attempting to capture the node for themselves. Additionally, once a Power node was activated, you could build up to three Power Generators around it, providing up to 8x its original value.

Power Nodes were often one of the biggest areas of contention on the map, and the most frequently harassed. Attackers had a few choices for dealing with them:

• They could destroy the generators and the node structure, forcing their opponent to pay requisition to rebuild them.
• They could destroy just the power node structure, capture the point, then steal the power generators for their own use.

Your choice ultimately came down to how much economic damage you wanted to inflict on your opponent, or how power starved you were at that particular moment.

Victory Points

Each player started the game with 500 victory points, and each map contained three Victory nodes, one near each player/team, and one in the center. When a victory node was captured and held, it started eating away at an opponents victory point total. The amount of “damage” to their victory points was equal to the difference in the number of victory nodes each player held (so if both players held one, neither player was “damaged”).

In a sense, victory points were a combination of a player's life total in Magic: The Gathering and Victory Cards in Dominion. Like in MTG, getting your opponents victory points to zero is the only way you win. It does not matter how many units you had on the field - this was the only victory condition. Additionally, like victory cards, holding them not give the player any tangible in-game benefit. They didn't give you any more resources, nor unlock any new units or powers. In fact, you could argue capturing them hurt you, as you could be spending your time capturing requisition or power nodes instead.

What it did, though, was add a sense of passive pressure to the other player. Before any shots were fired, merely holding a victory node “damaged” the opponent’s victory point total, forcing them to respond. Ignoring victory nodes in favor of rushing for other resources was not viable, as even a superior army would lose to a player that holds the victory nodes the longest.

But what if both players decided to ignore victory nodes and just amass their army? You couldn't do that either, as Relic also built in a gradual sudden death mechanic. After around 20 minutes, the victory point total of both players began to steadily drop, down to as low as 1. This not only made it impossible for games to drag on too long, but it meant holding and contesting victory nodes became increasingly more important the longer the game went on.

Red

Finally, we have what players called “Red.” It went by different names depending on which faction you played, such as “Eldritch Might” for the Eldar to “Waaaaagh!” for the Orks, but functioned the same way regardless. Each time you killed a unit or destroyed a structure, you earned Red. The defending player also earned 75% of the Red value of what they lost, so both players were constantly earning red throughout a match.

Red could be spent toward using Global Abilities, of which there were four for each Commander (DOW2’s equivalent of heroes from WC3, with three per faction). These abilities varied wildly, from providing a temporary damage buff, to calling in airstrikes, to summoning powerful special units. The first two abilities could be used right from the beginning. Abilities three and four become unlocked as you upgraded your HQ to advance to a higher tier. These abilities became increasingly more powerful, culminating in devastating 4th ability.

The 4th ability for every faction was always some equivalent of an SC2 Ghost’s nuke, capable of wiping out an entire army in a single click. As mentioned before, this is one of those things that seemed massively overpowered, but in fact wasn't. It could help a losing player even the odds, or a winning player finish out a game, but only if the defender ignored the very obvious glowing mark on the ground (which was even more conspicuous than the Ghost’s tiny red dot, mind you). They could also only be used once per game, as matches never went long enough for you to earn enough Red for a second shot. As such, skilled players often skipped these in favor of their lower-cost alternatives.

DOW2’s resource system was more than just a serviceable hurdle to getting fancy units and abilities. It opened up a multitude of paths to victory and gave players many different options to juggle during a match. It is the multi-arc principle at its finest.

Game Pacing - Punctuated or Gradual

Dawn of War II also had an interesting approach to game pacing. Like the Zerg in Starcraft, the HQ building of all factions could be upgraded up to three tiers. Unlike in Starcraft, advancing to a new tier was a major turning point in the match.

As I mentioned earlier, this was a game of hard counters. In tier 1, there are no vehicles nor any way to counter vehicles. Thus if you encounter an opponent that has hit tier 2 first, they will hold an advantage over you until you can catch up. This is likewise the case with tier 3, when aforementioned super units and army destroying abilities become available.

Arguably you could say this mechanical enforcement of a very clear early, mid, and late game transition may feel a little too forced. Perhaps, but I think having these punctuated moments added a lot of tension to the game, and complemented the game’s commitment to putting very large amounts of power in their player's hands. Compare this to the more gradual build-up of armies in Starcraft II, where the main pivotal moments come from players pulling off some awesome play and overcoming a superior force with great micro.

I think there is value to both player-authored turning points and mechanical turning points, but for me advancing in tiers felt far more compelling in DOW2 than in any SC2 match I played.

What Held Back Dawn of War II

So with all this awesome wrapped up in a single game, why didn’t Dawn of War II take off in popularity like its older sister Company of Heroes? I touched on a few of these already, but in brief:

• No multiplayer tutorials: This is obviously a huge one, compounded by the fact that the single player felt like playing almost a completely different game.
• Unintuative Balance Model: As awesome as leveraging fear and equilibrium can be, you absolutely have to tell your players how this works. To a casual observer, your game will seem massively broken and unfair if you don't explicitly show them the tools at their disposal. Of course, I say "unintuative" only when coming from other RTSes, as in real life it is perfectly reasonable to assume your pistols will not harm a massive, horrifying abomination.
• Too much RNG: There wasn't enough RNG to tip the scales most of the time, but there was just enough to turn off hardcore eSports folks. For example, the awesome but random sync kills, first introduced in DOW1, caused a unit to become essentially invulnerable for a few seconds as they performed a dramatic takedown of an enemy unit. This kind of stuff made the game a blast to play and watch, but not eSports material.
• Backlash from Company of Heroes Fans: There were many fans of CoH and the original Dawn of War that were dissappointed by the game. They were drawn in by the similarity to those two beloved games, but turned off for the above reasons, and thus gave it undo negative criticism beyond what it deserved.

This all could have been avoided if they just took the time to explain just one of their brilliant but punishing mechanics - Retreat and Reinforcement (DOW2's answer to Atlas's "respawn" mechanic, if you will). With the single click of a button, you could tell a squad to automatically run home to your base, moving much faster and taking significantly reduced damage as they did so. When they arive at your HQ, you could then "reinforce" to replace the squad members you lost. It costs roughly twice as much to buy a new squad as it does to reinforce each member one by one. Your decision of when to use just one single button was one of the weightiest decisions in the game.

This mechanic was carried over from COH, though Relic did not fully account for how its dynamics would change in DOW2. As mentioned earlier, this was a game where the biggest army would have maybe 6 squads total, whereas COH routinely had around twice that, and used infantry a lot more and made them cheaper (iirc, I'm not as experienced with COH). This meant that unlike in most RTSes, none of your units were expendable. If you failed to retreat even a single squad early game, you were done. You would fall way too far behind to recover. In fact, for experienced players, it was not uncommon in most games to go an entire match without losing a single squad, or at least evenly trading with an opponent after a big battle.

They didn't neccesarily have to change anything about the mechanic, all they had to do was tell people how it frikken worked!

Opaque Mechanics

Another thing that both Dawn of War II and Company of Heroes had in common was very opaque mechanics. While you could see a particular squad’s health, you could not see how much damage they did, how fast they were, how fast they attacked. The majority of their stats were just completely hidden from you. In fact, for many of the mechanics I even described earlier, like requisition point maturity, no player would ever know about them without some digging.

Instead, they did things like have more qualitative descriptions of things. Like “this unit is good against vehicles” or “deals fire damage in an area.” The player is left to experiment and feel out what these meant. I suspect this was done to make the game more intuative, or possibly to hide the insane damage modifiers they used under the hood. While never vague, they lacked the concrete numeric values something like Starcraft II thrives on, allowing for quick comparison between units across factions.

This often colluded with RNG, such as with the “melee skill” mechanic, where each squad had a rating of 1 to 100. When facing a unit with a lower melee skill, they have a chance to do a special attack, usually a knockback. This makes for some interesting visuals, but robs the player of a sense of agency, as they neither know when it will happen nor what the melee skill of any given unit was. Agency and control over outcomes is key when making a game with eSport potential.

TL;DR

Dawn of War II is a tremendously underrated game marred by a few bad decisions, and Project Atlas is shaping up to be its spiritual successor. Words cannot describe how pumped I am to see that happen (though I suppose the previous 6000+ words give you a good idea). I really want to see more games carry the torch and experiment with high power ceiling design the way it did.

So go out there and play some Cosmic Encounter and Dawn of War II. It is a crime that this style of game isn't tried more often, and I intend to correct that injustice.

Thanks for reading, and happy gaming to all.

EDIT: After some thought, I think I may have overstated how underrated Dawn of War II was by a little bit. At launch, it got great reviews, but even at its peak it never had a high server population on its multiplayer ladder. My perspective here comes more from my own personal experience of the game fading out of the public eye strangely quickly, while its predeccessor Company of Heroes remained strong for far longer. Thus consider this speculation and conjecture, not based on hard and fast research.