MTQ Update

Hi folks,

This project’s still kicking! We should have some new content on its way in the coming weeks. In the meantime…

Material, Time, Quality was one of the most interesting articles on the site for me to write. It was an experimental topic, and I wasn’t fully satisfied with how I was interpreting ‘Quality’ in lane-pushing games.

I’ve now updated that article with my more recent interpretation that sees ‘Quality’ relative to the ‘board state’. It reads more smoothly now, and with the third piece finally in place, the framework is complete as a tool for assessing decisions and trade-offs which might otherwise go unexplored. Check it out!


Game Design: Armours and Resistances


The moba genre has largely settled on an equilibrium with two armour types: Physical Armour and Magical Armour. Why is this the case, and could expanding the use of armours or resistances lead to greater complexity and depth?

In this discussion we’re focusing on armour types and resistances that players interact with. The rules that decide how much damage a ranged troop deals to a catapult are “environmental tuning”, and aren’t the focus of this article.

Fixed Factors

To begin our search for answers, let’s first review what armour types and resistances contribute to the genre, and establish why the current trend exists.

At any point in time, players face a constant fluctuation of variables which provide advantages and disadvantages over each of their opponents. Maybe one enemy is close to a wall and is advantageous to engage, but they’ve also just levelled up or got a new item that puts them ahead of the curve. We can visualise this abstractly as some kind of 2D surface, which I call an advantage map:



Here, perhaps the enemy is vulnerable to our magic damage, but we are vulnerable to their physical damage. Adding more armour types and resistances would increase the dimensions of the chart, allowing more room for different contours of matchup:


Armour types have a role to play in providing long-term strategic texture. As the damage type of an ability or attack is fixed and does not change during a match, the same is true of the relevance that damage types have in the game. Their permanence makes them an important tool in the designers kit for adding long-term dynamics to hero matchups. Many factors in decision-making are constantly changing (such as proximity to a wall, or current health); fixed factors are more stable, and create opportunity for strategy.

The presence or absence of fixed factors affects the weight of long-term choices, such as drafting or a player’s overall build plans. Generally, we want a certain ratio where both short and long-term choices matter, and fixed factors are key to managing that ratio. Too little emphasis on fixed factors, and smart short-term choices won’t tie into anything and become unrewarding; too much emphasis on fixed factors and we get long-term disadvantages that are very hard to turn around, such as complete outdrafts.

With armour types, we can manage the availability of items to control the timing curves for certain heroes/damage types. Physical damage might take a lot of items to become effective, while magical damage can taper off in the late-game due to a lack of steroids. Thus, a team which picks lots of magic damage has a fixed factor in their plan (they don’t have good options late-game), and both teams can manoeuvre around that.

Rock, Paper, Resistance

If you want fixed factors, resistances are about as fixed as it gets. Unlike armours where the strength of the defensive stat can be modified in-game, a resistance system uses predetermined values.

In such a system, each hero gets an “affinity” which alters any damage they take, depending on the affinity of the incoming damage. A familiar example is the pokémon resistances chart:


Image from

Here we would say that Fire-affinity attacks deal increased damage to units with a Grass or Ice affinity, and reduced damage to units with a Rock affinity. Players can’t modify the resistances table at all; it is part of the landscape on which they battle.

Flexibility is necessary for a resistance table to be fun! Players need opportunities to leverage the resistances table better than their opponents, proving their skill and providing satisfaction. Pokémon understands this: battles involve at least twelve combatants, and swapping or using off-affinity moves is core to its gameplay. Each player gets lots of opportunities to use different affinities, and there are even dual-affinity pokémon to increase the complexity.

In a moba context, we would aim match those properties: by making lane rotations easy/frequent, multiple-hero engagements common, or letting players alter what affinity they or their damage has, so they can actively adapt to bad matchups.

On the macro level affinities might work, but on a micro level they tend towards producing solved situations where one hero’s affinity advantage leaves little room for outplays. Getting into situations like this isn’t fun for either party, and there is constantly pressure on the weaker party to seek a more favourable engagement.

Let’s consider some possible implementation details. Damage is a fundamental mechanic, and making it more complex is like making movement more complex: it potentially affects almost everything about the game. Players need to understand affinities to play properly.

Hence, we would like the system to be as simple as possible.

The smallest affinity table has three affinities: enough to make a rock-paper-scissors pattern. Adding more is an option, but we’d need to support that additional complexity with very clear aesthetics: like pokémon does by making all its Fire-affinity moves look like fire. That places restrictions on the game’s aesthetic: entangling visuals with balance, and is not an ideal choice when the genre’s business model relies on letting players use a variety of cosmetics.

For existing examples of how affinities can work:

  • In Age of Myths, hero attacks are subject to a resistances table, but abilities are not. This is done with the purpose of balancing attack range and the items available to each hero class, and it works great!
  • Master X Master implements a rock-paper-scissors affinity system, which applies to all damage to/from heroes (and some neutral units too). The system works because, like perSonas, players in MXM pick two heroes and can swap between them. When faced with a bad matchup, swapping should ensure at least a neutral matchup (unless the player picked two heroes with the same affinity, which this system distinctly discourages).
    • The text showing damage dealt to units becomes more emphasised when an affinity advantage is being leveraged, which helps build familiarity with the system.
  • Elemental Wars made affinities and resistances its central premise, offered nice ways to interact with them, and was quite savvy to the potential issues.

My verdict is that resistances are potentially a good thing, but they come with a lot of responsibilities: for balancing, making them intuitive for players, delivering on flexible gameplay, and ensuring they contribute meaningful depth.

Weaknesses and Rules

Another important role for armour types and resistances is giving weaknesses to damage.

As the primary means of clearing creeps, taking towers, and killing enemy players: damage is universally useful. In order to give other mechanics a chance to shine, we want there to be some situations in which raw damage isn’t as effective.

One way to create those situations is write rules that say “X takes less damage from Y”. A simple example would be “buildings take less damage from heroes”, which forms the beginning of a resistance table as described above.

However, there are many more rules we could write to describe when damage will be less effective. Some rules will be general game rules, while others will be attached to a specific passive or active ability on a hero or item. Some will be categorical in nature, while others will depend on some numerical values like positioning or current health.

The chart below plots four classifications of rule for damage modification:


  1. General & Categorical: A general game rule which reduces damage based on fixed types like “is the damage from an ability” or “is the damage tagged with affinity fire“. This is largely equivalent to having a resistance system!
  2. General & Numerical: A general game rule which modifies damage based on some current numbers. A simple example would be “damaging an enemy who is facing away from you always deals +50%”.
  3. Specific & Categorical: A rule specific to an ability or item, that modifies damage from a certain category of damage source. For example, an item which blocks an incoming single-target ability every 20 seconds.
  4. Specific & Numerical: A rule specific to an ability or item, which modifies damage based on some current numbers. Example: reduce damage from enemies who are more than a certain distance away from the hero.

As we know, categorical properties don’t change and they form part of the long-term strategic texture of the game. Numerical properties do, and hence they are open to tactical manipulation.

All players must abide by general game rules, making them formative of the game experience. Ninjas on has a rule where heroes are slowed upon taking damage, and this alters the gameplay considerably.

Conversely, any rules which are specific to certain heroes or items can receive varying emphasis depending on how much players interact with them, hence their being marked on the chart as interactive rules.


Of course, there is always the possibility of overlap between the rule classifications. We could have a rule that says “buildings under 50% life take less damage from heroes”; this is both numerical and categorical.

In fact, there is one specific damage modification rule that overlaps with all the categories:


Armours are an adjustable stat that modifies damage flagged with a certain ‘type’. This makes them both categorical and numerical. Armours are formative because they apply to everything in the game, but also interactive because their influence can vary on a per-unit basis. This makes them a very versatile mechanic for introducing weaknesses in damage.

To take it slowly, lets examine a game with no armours whatsoever, and introduce an armour for the first time. Having a single armour type is possible: we end up with damage that ignores armour (“pure” damage), and damage which doesn’t.

This isn’t a good arrangement, because either:

  • Pure damage is uncommon and armour protects against most things (armour becomes too valuable).
  • Pure damage is common and there’s no instrument to interact with or counter it.

This may have been the case for some old AoS maps which didn’t use magic armour. In those games, abilities would deal a fixed amount of damage, which was countered by building max health to reduce the proportional impact. Not ideal for the designer: max health has so many other uses that it can’t be tuned for countering pure damage in isolation, and players have to deal with fuzzy/overlapping choices.

Having two armour types like physical and magical is enough flexibility to allow high-damage heroes like nukers to exist without breaking the game, because while their damage is high, it has an inherent weakness: either it’s all of the same type and can be countered efficiently, or it’s of split types and becomes difficult to build for.

Tropes and Variations

The most common forms of armour resolve to a percentage modifier on damage. However, we defined armour as “adjustable stat that modifies damage flagged with a certain type”, and we could apply that stat in other ways. For example,

  • ‘Physical armour’ could offer a flat reduction on incoming physical damage.
  • ‘Magical armour’ could determine the health/duration/cooldown of an anti-magic shield that can be temporarily activated (as an active ability that all heroes have).

These have some quirks (how does physical damage over time overcome a flat reduction?), but there’s nothing wrong with doing it this way.

We’re accustomed to strong correlation between damage types and damage sources: “physical attacks” and “magical spells”. This arrangement means that heroes usually have access to both damage types, even if one is weak. These correlations aren’t a bad thing, as they establish trends, and exceptions which break a trend make things interesting.


Should stats like Evasion and Block count as armour types? Well, they can if you shift your perspective! One way to look at it is that evasion (as a counter) “defines” attacks, in the same way that physical armour (as a counter) “defines” physical damage. The equivalent for evading spells is spell immunity, which is an armour type that operates on a temporary activation basis, like we described above. (In DotA, the duration of spell immunity is a variable value!)

So if you have evasion and spell immunity as two varying stats, that’s effectively an armour system and it covers most incoming forms of damage. Anything that neither evasion nor spell immunity can block is the equivalent of pure damage. There’s no reason why we couldn’t have a moba where this was the primary perspective!

The Third Shield

armour-types-2In most lane-pushing games, the physical/magical split has a huge influence on how players draft, choose items, and use their abilities. Imagine the complexity if a third armour type and corresponding damage was introduced…

Let’s call it Runic Armour.

I’m going to pick some correlations that I think would integrate relatively seamlessly into the existing moba format. Namely, we’ll say buildings and boss objectives have minimal Runic armour, so heroes with Runic damage have an easy time taking objectives, and said objectives get back some physical armour to compensate. This fixes an unhealthy side-effect of having only two armours (which is appropriate for PvP), where heroes with physical dps or which reduce physical armour get to be effective against buildings as a freebie.

We’ll also declare that Runic damage is usually delivered in melee or close range, so heroes need to directly engage objectives to get the most of it (but maybe some hero gets to be the exception to that), and that it’s uncommon, with maybe 15-20% of heroes having native access to it. Catapults probably deal and are weak to Runic as well.

For hero combat, Runic damage would be relatively low (it deals almost full to buildings), but it doesn’t have to be worthless. We can supply items/abilities that apply negative Runic armour to heroes only, or give a high percentage of mana leech on Runic damage: giving Runic abilities the opportunity to fill niches that others don’t.

These are undeveloped thoughts, but I think they demonstrate some of the room that armours have to potentially increase depth, if they’re assigned a purpose and properly integrated.

On a smaller scale, we could define narrow-interest armours: such as a support armour which protects against supportive nukes and reduces the duration of crowd control, or a creature armour which protects against damage from summoned units. Consider a nuke which reduces a hero’s creature armour, or choosing to give towers an on-hit passive that reduces an enemy’s support armour, making them more vulnerable to usually-harmless supports.


At this point we should have a clear understanding of why the two-armour system has been the standard for so long: it adds a fixed factor that the designer can easily tune, and gives damage some much-needed weaknesses. Resistances and affinities are a high-maintenance addition, which most games have understandably chosen to avoid (for hero vs hero interactions).

Introducing new armour stats, or more generally, variable hero stats (which might not correspond directly to damage) helps to to formalise interactions, and improve players’ understanding of modifiers that are currently fixed percentages, and don’t stack intuitively. These stats typically aren’t formalised yet because they’re used infrequently, but as a game expands, having a system in place becomes more valuable. For example, Heroes of the Storm recently reached a point where using percentage-modifier buffs to apply damage modifiers was proving cumbersome, and the game has now moved to a more standard armour system.

We are in a genre where deep mechanical interaction is so important, that exposing or introducing more stats for manipulation feels like a natural progression as games expand. Whether it will be new armour types or otherwise remains to be seen!

MOBA Geography

During the summer I had the pleasure of being interviewed by esports journalist and writer Josh Calixto as part of a feature piece about the ‘geography’ of commercial mobas: addressing the question of how the ubiquitous “three lanes” layout evolved, and why it’s been so enduring. We also covered the evolution of various other map features, and mused about possibilities and constraints on where map designs might be headed next.

If you like the sound of that, check it the full article out over on Killscreen.

Brief Note

Hi folks,

I am acutely aware that I haven’t been able to update lpg in a while (due to the demands of another project). Rest assured that my work here won’t be left unfinished; I will get back to writing articles as soon as time permits.

Review: Ultimate War, Part 2

Continuing from Ultimate War, Part 1


Scattered around the map are a number of objectives which may be captured by either team to enjoy certain benefits. They are all initially controlled by the neutral faction Chaos, and guarded by a reasonably strong defender unit. Upon dealing the finishing blow to the objective (all of which are structures), it will be restored to full life under its new owner’s control, along with a fresh defender.

Among the objectives are five Spawn Buildings: the team which controls them will periodically spawn a certain unit type from that location, to march along the nearest lane. Ownership of a Spawn Building will periodically grant the human players with Magical Runes, and importantly: increases the number of friendly units that will spawn during a Great War. This can be the difference between taking or losing towers on the mid-lane each time a Great War takes place.

The remaining map objectives are the three Obelisks, each of which grants a different global aura to the owner’s team. This is a familiar mechanic which appears in many an AoS, but Ultimate War adds a twist: each Obelisk controlled by the human team also enables an extra Team Spell for their main base. The Obelisks are:

  • Obelisk of Power: Passively increases attack damage, its team spell grants a further +100% attack damage to all units for 20 seconds.
  • Obelisk of Magic: Passively increases life regeneration, its team spell is a global-range nuke that deals 30% of maximum life to enemies in an area (this is crazy-strong late-game).
  • Obelisk of Protection: Passively increases armour, its team spell grants a ton of bonuses to friendly towers for 35 seconds (including a huge amount of regeneration).

The Obelisk of Magic, recently captured.

Even if it’s not possible to hold an Obelisk long-term, grabbing one for just a moment is enough to activate a spell to swing the battle (the spells have high cooldowns anyway). While in control of all three Obelisks, an “ultimate” team spell becomes enabled, which summons extra-powerful reinforcements on all three lanes.

I’m slightly disappointed the Obelisk team spells are locked behind one of the higher-tier Glyphs of Knowledge, essentially meaning that the humans need sink a lot of money into “tech” first, and can’t use the spells early-game in a weaker form. They’re among the more interesting perks for controlling a map objective that I’ve encountered, and properly-implemented would create a more meaningful distinction between the utility offered from controlling each lane. After all, it’s possible to get very creative with spells compared to auras, and there’s more opportunity for utility spells which don’t directly snowball in the way that aura implementations tend to.


Obelisks have another important role: they generate Divine Power for their team. At the start of a match, both teams have 0 Divine Power, which will gradually increase over time. Once it reaches 7500, that team’s Deity will be summoned to march down the mid lane, where it will either eventually destroy the enemy main base, or die trying (losing the game for that team; you can’t let your Deity die). This process takes approximately 90 minutes, depending on Obelisk control.

In this genre it’s important to put a clock on the match, so that victory isn’t secured the moment heroes have grown strong enough to “not lose”: they need to be kept under pressure to actually win. (Aeon of Strife does this too, using a 2-hour countdown timer.) Here we see the value of having a large quantity of towers in each base: distributing the defences over many units gives players better feedback about whether they’re taking the enemy base fast enough to win.

The enemy Deity, surrounded by the usual late-game chaos.

If the humans can’t stop the enemy Deity from appearing, they can still race to summon their own, or try for a direct kill (that global nuke from the Obelisk of Magic is one way to do it). Maintaining control of Obelisks is usually the deciding factor (since they affect Divine Power), but the enemy often sends squads of Elites specifically to contest them.

Finally, it’s worth noting that all the map objectives are on side lanes: there’s none on or directly accessible from the mid lane (where most of the enemy units will be pushing). A pair of Waygates allow units to bypass the mid lane entirely if they want to focus on objectives without getting caught in the crossfire, which some squads of enemy Elites will take advantage of.


The cast of Ultimate War features familiar pop-culture characters, ranging from superheroes and plumbers to vampires and sith lords. There are 33 in total, and many of the enemy Elites are also based on well-known villains.

There are three things to note about heroes. The first is that quite a few of them are classified as Divine Beings, which means they gain experience at a much slower rate, have substantially longer revive times, and are forbidden from carrying Unique items (the strongest type). It’s an option for players who want to spend their resources on team assets rather than personal equipment, or just want a fancier hero model.

The second is that while hero abilities are mostly standard issue, they typically gain increased range/cast range as they’re levelled up. This turns out to be crucially important on a large battlefield where you’re trying to cover a lot of ground or need to hit critical targets on the back lines.

A Divine Firelord throwing a long-range burning disc at enemy Elites.

The third is that each hero starts with a Spellbook: a sub-menu of additional abilities they can use. At the start of a match, this has only the essentials (usually True Sight, Dispel Magic, and a few others), but as more Glyphs of Knowledge are obtained it can reach up to 21 passives and abilities, depending on the hero. This is a lot of abilities, considering that heroes also have items and hero skills to manage.

The enemy Elites are designed around the fact that all heroes have True Sight and Dispel. I don’t think that’s a bad baseline: it makes it easy to create an Elite which requires “the presence of any hero” to defeat, or “the presence of 2 heroes” if it can cast a strong buff twice (Dispel Magic has a cooldown). It also allows some heroes to be special because they possess a second dispel, or can provide True Sight remotely. This implementation relies a bit too much on specific use of Warcraft mechanics, but I think it shows confidence in understanding of the genre to tweak the baseline of what a hero is, to better suit a particular game.


Ultimate War packs plenty of ideas, and does a reasonably good job of delivering a co-op “humans versus Elites” experience. Unfortunately, it has its issues: namely poor compartmentalisation of its features, and too much addition when implementing content. We’ve described already some of the issues with Elites and the lack of simple clarity when dealing with them; the game’s strategic options (the ways in which resources/gold can be allocated) suffer in a similar way.

It’s not problem that every building in the base offers some answer to “how do I improve the base’s defences”, but it is a problem that their solutions overlap so heavily, particularly for repairing towers or summoning more units. The minutes-long cooldowns on many of these means that often it’s not even a matter of which is better, but rather, which option is still available. The solution to this is getting rid of overlap: if the Palace can summon units globally, then I shouldn’t be able to hire a unit which has a global blink. If I can hire units at all, there shouldn’t be a consumable item that creates summons.

Another option for making things tidier would be having buildings correspond to the problems players are addressing (repairing and towers, capturing objectives, pushing lanes) rather than game mechanics like items or spells. However: we can see that the Palace having solutions for a lot of things is “untidy”, but that being able to execute only one order at a time adds depth. Unless we’re in an ideal world where all mechanics correspond to solutions (all items repair towers, all units capture objectives), there’s always going to be compromises when defining categories for players to navigate.

I think it’s healthy to examine the “humans versus Elites” format of lane-pushing games. PvP has plenty to learn from the what makes a good co-op experience, and having good co-op is a great way to ease players into a game if both modes are designed with congruency in mind. As an exercise for the reader: what would Ultimate War‘s PvP mode look like?

Download: Here

Review: Ultimate War, Part 1

In a return to the roots of the genre, Ultimate War (Plasma Boy, 2013-2014) presents a co-op AoS scenario, in which six players must demonstrate how their team of heroes can not only survive against a computer opponent with superior numbers and forces, but eventually grow strong enough to conquer it.

While many games pit human players against an AI, Ultimate War follows in the steps of Aeon of Strife: giving both teams the same basic composition of forces and buildings, and using “Elite” troops to create varied tactical complications. To avoid confusion with playing against bots in a regular AoS, I’m going to refer to this sub-genre as “humans versus Elites”.

50+ towers apiece. Even that won’t be enough to save you.

To better mimic a battlefield, troop waves in Ultimate War spawn less frequently than in other maps, but in larger numbers. That gives time for front lines to form, and once one team punches through, time to attack the enemy towers before a fresh wave of troops can draw fire. There’s never so few units on the map that it feels empty; the armies’ ebb and flow has a good rhythm.

The mid lane is comfortably twice as wide as the other two. It consistently has the most troop traffic, but every 20 minutes that’s taken to an extreme in the form of a “Great War“: an event in which both factions will spawn extra-large waves on the mid lane. The ensuing fights are chaotic but a nice change of pace, and whichever team wins will enjoy a stronger push than usual.

A “Great War” in action.


Much like Aeon of Strife, the enemy team’s lack of heroes is made up for by its Elite units. These specialist soldiers are a tougher breed, guaranteed to cause problems for the human players unless they respond quickly with the appropriate counter at the ready.

Some Elites are an issue because they have defences that the regular army can’t deal with, and simply won’t die until a hero addresses them. This can include physical immunity, invisibility, hyper-regeneration, or long range that towers can’t respond to. Other Elites have dangerous offensive tools, such as channelled AoE damage spells that will slowly annihilate a base. And some rare specimens are a mixture of both: simply too powerful to allow their devastation to run its natural course.

I like that Hydralisks got to be Elites in this game too.

There is some remuneration for heroes that spend their time chasing down Elites, in the form of occasional consumable item drops, and awarding a bounty of Magical Runes: the game’s secondary player resource which is used alongside gold for most purchases. This is awarded for the finishing blow, so securing last hits on Elites is quite important.

Ultimate War makes an effort to help players assess and respond to threats, using an alerts system, giving Elites a unique colour on the minimap, and even providing a reference book with some of the different types. However, this doesn’t resolve the underlying problem that are over 150 different types of Elite, each of which can have up to 16 different abilities and passives.

Many of these abilities are unlisted, have no accompanying visuals, and completely overrule the benchmarks that players normally use to assess a unit’s strengths and weaknesses. Maximum life, armour, and damage type are meaningless when so many units have hidden passives like ‘Demolish’ granting up to ×18 damage against buildings, or Hardened Skin reducing all incoming attack damage by massive amounts. Even if we ignore issues with hidden information, there are far too many different ‘quirks’ in the game, and furthermore, too many on each unit.

I believe the correct way to implement Elites is to give each unit type minimal special properties, and send varying combinations/quantities to form different challenges. All the better if the Elites have auras, buffs, or other abilities which let them synergise with each-other to create memorable pairings. Aeon of Strife was a successful game with exactly two types of Elite, and while that definitely leaves room for expansion, Ultimate War went too far with its additions.

Scripted Events:

Throughout a match, a number of random events will occur. The most common is the appearance of enemy Elites, which might march down a lane, focus on contesting objectives, or have some special behaviours. Some types of Elite will show up and remain at a fixed location; applying global pressure (through auras or spawning more units) until the humans manually eliminate them.

There is a third faction in the game called Chaos: hostile to both armies, and with unit types and scripted events of its own. Its typical behaviour is to randomly spawn units in the middle of the lanes, which either lash out at the nearest thing they find, or try to capture an objective and return it to neutral status.

A swarm of aliens showed up! Thankfully, disrupting an enemy push.

To add to the commotion, there are also a bunch of tornadoes which roam around the map, randomly tossing units into the air, and occasional meteor showers which can wipe out even some of the stronger Elites. I’d rather that Chaos and the weather effects weren’t in the game, but statistically they favour the underdog (who has less units), so they do serve a purpose.


Each team’s base in Ultimate War is protected by over 50 towers, in a variety of shapes and sizes. The enemy definitely enjoys most of the variety, with fire, frost, poison, and even invisible towers peppered along the cliffs. The human towers have aesthetic variety, but they all do the same thing; varying primarily in attack damage.

The sheer number of towers gives a granular measure of progress, first to measure how much ground is lost each time the enemy forces come knocking (and it’s hard to avoid them knocking), and later to gauge how quickly progress is being made when assaulting the enemy base.

Ultimate War has a special mechanic paired with towers, called Generators. These small glittering buildings, presumably inspired by Tower Defence games, grant an aura to nearby buildings and wards in a small radius. They can be seen in the image below, where they’re amping select towers in the enemy base with double damage, regeneration, or huge amounts of armour. They don’t have much life, and cannot be repaired, but their aura bonuses are highly valuable.

The red circle shows the approximate radius of a Generator.

For human players, the interaction with Generators is pretty simple: kill them first! Simple, but it does require that heroes be present when pushing, since the friendly army won’t focus the Generators on its own: it will prioritise towers which are outputting damage.

The subtler side of Generators is that humans start with almost none, but can purchase and place them near towers to shore up their defences. The catch is that they’re very limited in supply, and certain enemy Elites have a tendency to destroy them (with bouncing attacks or other spells) if they’re not carefully guarded. Furthermore, all the Generators in the world won’t be enough if an Elite with the right immunities shows up…

A Siege Behemoth cuts through towers like butter, leaving some now-useless generators behind.

I really like Generators as a feature for this kind of map. All three (damage, regen, armour) are practical against different threats, their small radius puts them at risk, and they force players to be picky about which towers they augment. Towers on their own are just the right amount of underwhelming, so there’s pressure to use Generators wisely. This mechanic probably wouldn’t work so well in a PvP scenario, where as we saw above, human players simply know to attack the generators first, no matter where they’re placed.

The Swiss Army Base:

Other than defensive instalments, the human base has a large number of buildings with tools to assist the players in their endeavours. The most familiar will be a line of item shops, which sell RPG-categorised items (heroes are limited to: 1 Weapon, 1 Armour, 2 Accessories, 1 Unique item each). Also available are consumables, including Serpent Ward (permanent once placed), a repair kit to instantly heal towers, a global blink, a single-target debuff for enemy towers, a rocket launcher needed to kill certain enemy Elites, and of course the Generators themselves.

Next up, there is a Mercenary Camp where players can spend gold to hire extra units (up to a relatively low food limit for each player). They are varied in purpose: some are cannon fodder, some are specialised to destroy buildings/generators, another is a flying unit that effectively is a mobile generator! However, the restock time on each unit type is typically over 10 minutes, so they are not a disposable investment as they tend to be in other maps.

This ‘Magic Sentry’ owl applies Generator auras in the same small radius, but at half-power.

The Reinforcements building allows players to spend gold to buy additional waves of troops for the team. Options include sending a squad of either regular or siege troops to a chosen lane, sending Elite troops to all lanes, or sending an Elite squad to each of the map objectives. Furthermore, for a massive financial investment, an Emerald Dragon can be summoned to provide the backbone of a push along the mid lane. Here, players may also buy up to five permanent Reinforcements Upgrades, which improve any future reinforcements that are purchased.

At the Statue of Divinity, players can buy some special items that are either recommended or explicitly required for dealing with certain Elites. That is to say, there is at least one Elite in the game who is impervious unless a named item from this shop is in the hero’s inventory. Also on sale are Glyphs of Knowledge, of which there are five, each more expensive than the last, and incrementally unlocking new abilities for all the human heroes, and for the human main base, which is called the Palace.

The Palace has a number of spells, which may be cast by any player. The (quite long) cooldowns are tied to the base itself, and hence these spells are a shared asset for use by the entire team. The stronger ones are locked behind Glyphs of Knowledge, but some are available innately as well. These “Team Spells” include:

  • Repair: a channelled ability which will gradually spend the friendly AI’s gold to repair the target building over time.
    • The friendly AI gets gold from any last hits that it takes, meaning it will accumulate quite a substantial sum over the course of a match. (Mechanically, this occurs in a lot of AoS maps, but the gold is never used for anything.) Here, that gold can be used exclusively for Repair.
    • Despite it seeming like an entire army’s war chest would fund a lot of repairs, it proves to be quite expensive, and with so many exposed towers, repair gold will routinely run dry.
    • While Repair has enough range to work on any building inside the base, I was disappointed that it can’t reach the buildings held as objectives. I think it would have been great to let players spend their precious repair money to buy time for heroes to arrive and contest capture attempts.
  • Cloud and Earthquake: These two are channelled abilities which last for up to 40 seconds, and may be cast at global range. The first disables towers in an area (helping units push), while the second damages towers in an area while also slowing enemy units. Both are useful, but it is worth noting that the main base is only able to perform one action at a time: it can’t repair while using these skills, nor can it use them both at once.

    Using Cloud on one of the stronger, Generator-buffed towers at the back.

  • Gun Bots and Serpent Ward: Another pair of global abilities, these can place permanent units anywhere on the map. Serpent Wards can’t move, but are strong and benefit from Generators, which makes them efficient to place beside existing towers.
  • Magic Meteor and Rain of Fire: These are a single-target and AoE nuke, respectively. Sometimes you just gotta call in the artillery! Note that some Elites have Spell Shield, which can render the 13,000 damage Magic Meteor nuke completely harmless.

Continued in Part 2.

Review: Overdrive

Overdrive_Preview Overdrive (Pzygho, 2008) is one of those AoS maps which delivers a surprising amount of emergent gameplay by introducing a single twist. The twist in question is using the Overdrive mechanic from Final Fantasy X to give heroes their ultimates, rather than the usual method of granting them at level 6 and having lengthy cooldowns.

In this map’s implementation, a hero’s ultimate becomes available once they have taken a certain threshold of damage. Once cast, the ultimate vanishes and the hero must again take damage up to the threshold to re-enable it. A hero with their ultimate available displays a clear graphic to all players.

Ultimates level up exclusively through usage, with the damage threshold increasing at higher ranks. Hence, any hero aiming to have a maxed-out ultimate is going to need to tank damage and use their ultimate many times to get there. One of the best ways to achieve this is using neutral camps, from which any hero can absorb damage while staying out of enemy sight. This interaction elevates neutral camps to being an important game element, for every hero.

The two lanes are very close, and are easy to move between.

Creep Camps:

It’s not strange to see a two-lane map, given that Overdrive is a 4v4 game with a considerable emphasis on jungling. The lanes are close and it’s easy to hop from one to the other. The jungle entrances are mostly covered by towers, which makes ganking a bit more difficult early in the game.

Inside the jungle itself, creep camps are haphazardly scattered around. There are 8 camps on the south side, 6 on the north, and given their loose placement it seems fair to assume that there wasn’t any particular intent behind this. Committing to this idea by having a “hotly contested” jungle in the south with a “quieter jungle” in the north could have worked well: it would add some much needed asymmetry to the map’s layout, and make jungling (and tanking for ultimates) a more nuanced decision.

The “corner camps” located just outside each team’s base are perhaps the most useful ones in the game. They’re furthest from the possibility of enemy ganks, have the shortest escape route, and are also closest to friendly healing: making them ideal for working towards ultimates. They are so good that I’d rather see them removed, since they trivialize the overdrive mechanic.

The short walk from corner camp to the north base entrance.

The logic behind how neutral creeps spawn is a bit unconventional. Each camp houses a fixed number of creeps, with unit types randomly chosen from the camp’s encounter table. When a creep dies, its respawn timer begins immediately (instead of waiting for the rest of the camp to die), and it is guaranteed to be a different unit type upon revival.

Also unusual is that in Overdrive, it’s possible to find a lowly Sea Giant (level 3) and a burly Revenant with chaos damage (level 8) in the same encounter table. So the corner camp a player might be relying on could randomly be too weak to provide good dps for tanking, or too strong to kill for experience. The lack of consistency isn’t to my taste.

The game explains that there’s a chance for neutral creeps to drop a random tome on death, permanently granting +1 to a stat for whoever picks it up. However, the chance is abysmally low, and since the jungle already gives heroes a reason to visit, this feature doesn’t make much difference.


There are two bosses on opposing corners of the map. They are about equal in terms of power, but each has their own name, and set of abilities (both single-target and area-of-effect) which they will use whenever the opportunity presents itself. Unusually for boss creeps, these abilities cost mana, of which the bosses have a large (but exhaustible) supply. In practice, the regeneration is too high for “bleeding” a boss’s mana before attacking it to be a viable tactic, but that’s a pretty neat direction that some other game could experiment with.

Each boss inhabits an enclosure, which initially looks decorative, but actually serves a purpose: heroes can’t attack a boss unless they’re inside. This seems to be a solution to ranged heroes kiting a boss to death, and to an extent it works (though the boss can still be affected by spells from any distance).

Murkius in his lair.

Bosses are a great way to take lots of damage quickly at low risk due to their location (there is still some risk, as the bosses have single-target disables). However, bosses are a one-time objective: they don’t respawn, and hence killing a boss means it can’t be tanked in the future. For this reason, it’s slightly disadvantageous to lose the boss closest to your own base.

The only reward for defeating a boss is experience for heroes who are nearby, and a special permanent item which grants flat bonuses (and hence will be more valuable earlier in the game). However, with Overdrive being a fast paced game on a small battlefield, it’s easy to see when players are missing from the map, so early boss attempts are hard to slip under the radar.


Players acquire a secondary resource in Overdrive called Favour. It’s awarded for participation in hero kills, tower kills, and map objectives, and is required (along with gold) to buy just about anything that can be bought.

One of the nice things about Favour is that players start the match with a modest amount of it: enough to get started on some items and not be under immediate pressure to get kills, or at risk of being starved after losing the first two flags. Most AoS maps start players with 0 of the secondary resource, and players are immediately racing with the enemy team to keep ahead. Having seen Overdrive’s approach, I have to say I prefer it.

The other point of interest is that Favour makes hero kills important for long-term progression, as pushing lanes and taking towers awards relatively little Favour. Since ultimates make it easier to secure kills, players have an additional long-term incentive to work on their overdrive.


The item system is divided in two: those items which do not cost Favour, and those which do. The former category is a tiny minority: it consists of cheap consumable items like teleport scrolls, wards, and potions. Healing efficiently in Overdrive is important, so weighing cheap non-combat potions against more expensive in-combat potions, against a refillable bottle, against saving for a permanent healing item, could be the difference between getting an ultimate off before dying, or not.

The Favour-costing items are divided into classes: Weapon, Helmet, Armour, Accessory, and Miscellaneous. At most 2 Weapons, 2 Accessories, 1 Helmet, and 1 Armour are allowed per hero. Some of these items are also primitive, which denotes that the item can be upgraded into one of 3-4 slightly better items, often having a special passive. For example, Speed Boots is a primitive accessory which has three upgrades, while Wooden Staff is a primitive weapon with four upgrades.

One of the disadvantages of a recipe system (like the one seen in DotA) is that a player must understand their late-game plan before they can efficiently choose early-game items. A system with primitive items that can serve as starting points should resolve this, by presenting players with meaningful choices one step at a time. (Dawngate showcases a solid implementation of this.) However, Overdrive completely misses the point, and instead has six primitive weapons all in the same price bracket, all giving roughly the same bonus attack damage, and providing no tangible decision-making data to players whatsoever.

Which will you choose?

It seems these primitive weapons were intended to divide the items based on the icons they used, rather than their effects. Not the most helpful categorisation…

Given the need to absorb damage, healing items are an appropriately diverse commodity: including the classics like life leech, a regeneration aura, and instant group healing, but also some alternatives like a mass-rejuvenation, sacrificing an allied creep to heal, or healing nearby allies on death.

One of the most expensive items in the game passively adds an extra level to the owner’s ultimate, which is a nice way to let players get more out of their investment in overdrive-related activities.

There are three available faction upgrades, for melee troops, ranged troops, and towers respectively. These cost favour and gold, and despite their exponential cost, they become a mandatory investment once one team has started using them. (As we’ve discussed in previous reviews, this is not a healthy system.) Upgrades complete instantly, so there is at least no pressure to start early due to long research times. I did enjoy that the towers’ upgrade increases their attack range; a small twist I haven’t seen before.

Flags and Runes:

The main map objective in Overdrive is a periodic quest. A flag will spawn every 6 minutes in the middle of the map, and upon being returned to a team’s base, it will award Favour to all players on that team. Each flag returned will grant more favour than the last, and the only restriction while carrying it is that dying or teleporting will cause the flag to drop.

Time to earn some favour…

Being situated on high ground encourages teams to show up early to secure the spawn zone and a vision advantage, which can lead to some pre-flag brawls. However, the open-plan terrain, and lack of any tangible disadvantage for the flag-bearer mean that the first team to pick up the flag can (and should) immediately disengage and bring it home, minimising the risk of losing it. If the purpose of flags was to create engagements, this isn’t a complete or effective implementation.

The existence of the flag makes pushing towers early more rewarding, as neither chasing a flag past enemy towers, nor conceding it, are very appealing.

There are four runes which spawn on the map, in familiar DotA style. Unfortunately, their duration is too short for them to make any practical contribution to gameplay: they cannot be bottled, they barely even help with jungling, and the one rune that would have proved useful (regeneration) isn’t available.


There are 12 heroes in total, which is comfortable for a relatively small 4v4 AoS. They each have three skills + one overdrive ultimate, done to a decently high standard of creativity. Sometimes it’s subtle variations which work best: Voodoo Shaman has a skill which hexes one unit, causing it to apply a poison to any other nearby enemy units. The overdrive ultimates of course are the stars of the show, and they deliver by being suitably exciting and high-impact spells.

Powerful huge area damage: ✓

When a hero dies, an invisible ghost spawns at its fountain, and may be ordered to travel to the hero’s corpse where it may perform an instant full-health revival. This isn’t possible at early levels where revive timers are too short for the ghost to even leave the base, but it eliminates a lot of downtime at later levels.

There were unimplemented plans to mark some areas on the map as “dead-zones” where in-place revival would be disabled: such as inside each team’s base, and around the boss areas. Aside from being necessary to stop abuse, this would have encouraged defending teams to make aggressive plays outside their base, before they get boxed in with no recourse.

Hero deaths incur no gold penalty, which means that dying to neutral creeps early-game is by far the most efficient way to get home and back to full life. Losing one’s ultimate or overdrive progress might have been a fair penalty here.


Some of Overdrive‘s features are less-than-ideal in their implementations, and a lot of that can be attributed to the map being incomplete. What’s important are the overdrive ultimates: a core mechanic which impacts player decisions throughout the game: when it comes to healing, jungling, killing enemies to keep up in favour, or using the ultimates to level them up. There are also subtler consequences, such as how much one should harass an enemy in-lane, or how to conduct a teamfight.

I really like that buying regen and heading to the jungle for the first minute to ready an ultimate is a valid decision: one that creates action early, has its own trade-offs, and wouldn’t otherwise be available. I like that maxing out a hero’s ultimate as soon as possible is an alternative means of progression. Those are great dynamics!

I was lucky enough to chat with one of the players who was tangentially involved with the map’s development, and enjoyed hearing a story about a player who arrived a bit late to lane, but without an overdrive: as though they’d been afk. The enemy jumped, only to discover that the player had been jungling to just under the damage limit, and immediately had an overdrive ready as rebuttal!

Of course, big ultimates like this cost a lot of mana for a level 1 hero.

Reportedly there were plans to expand upon the overdrive system. The changes, still modelled on of Final Fantasy X, would allow players to pick one from five different activities that would be counted towards their overdrive; a choice made at the start of the match. The proposed options were:

  • Taking damage
  • Healing self or allies
  • Crowd control applied to enemies (in seconds)
  • Last hits
  • Heroes killed

Personally, I would hate to lose the universal option of taking damage, since it gives everyone a reason to jungle, but there might be cases where some heroes would appreciate something different? I’m not sure. There’s definitely a risk of taking damage being paired up with healing and proving far too effective. Heroes killed might be either too weak (difficult to have it ready at the start of a fight), or too snowbally, as getting 2-3 ultimates in a single fight could be ridiculous.

If this hypothetical system were to become reality, I would propose that part-way through a match (say around level 10), players should be allowed to activate a second means of building toward overdrive, to freshen things up and let them adapt. Committing to a single overdrive activity all game (particularly a niche one) would be constricting and tedious.

I hadn’t invested much time into thinking about ultimates before encountering this map, but I think the ideas speak for themselves and make a pretty good case for exploring alternatives to the familiar “level 6 + cooldowns” approach. Hopefully we’ll see more experiments like Overdrive‘s in future lane-pushing games!

Download: Here