D4 Proposal 1 - Skills Selector - Tree VS Virtual Matrix

4K Proposal

https://i.ibb.co/jRXsY6x/diablo-4-proposal-1-skills-matrix.jpg

Summary

Don’t: Trees
Do: Virtual Matrices
Why: Billions of combinations, Better UI (compact, scalable, responsive)

Skills Matrix

Rules

  • All skills and runes available at Level 1
  • Maximum points per skill can be exceeded using items
  • Point cancellation is free and immediate (adaptative build)
  • For gold sinks: Search for « nwn2 crossroad keep » https://www.google.com/search?q=nwn2+crossroad+keep

« AND » runes examples

On hit

  • Pummelers (5) : +5% Hit: Stun (1s)
  • Freezers (5) : +5% Hit: Freeze (1s)
  • Brawlers (5) : +5% Hit: Immobilize (1s)
  • Mockers (5) : +5% Hit: Berserk (1s)
  • Plaguers (5) : +5% Hit: Poison (1s)
  • Injurers (5) : +5% Hit: Bleed (1s)
  • Carvers (5) : +5% Hit: Generate Corpse

When hit

  • Quickeners (5) : +5% When Hit: Accelerate (1s)
  • Ghosters (5) : +5% When Hit: Bone Spirit
  • Breakers (5) : +5% When Hit: Break CC
  • Brothers (5) : +5% When Hit: +1 Skeleton (3s)
  • Crushers (5) : +5% When Hit: Next hit is crit
  • Armorers (5) : +5% When Hit: Harden (3s)
  • Sharders (5) : +5% When Hit: Bone Shards

No hit condition

  • Skirmishers (5) : x5% Damage to CC
  • Slaughterers (5) : x5% Attack Speed
  • Smashers (5) : x5% Critical damage
  • Zweihänders (5) : x5% Area damage
  • Spikers (5) : x5% Skeleton thorns
  • Followers (1) : Share your damage
  • Grounders (1) : Risen from ground

Explainer

Skills selection

  • Matrices rather than Trees
  • Freedom rather than Linearity

OR VS AND

  • Don’t force OR choices (developers choice, OR inheritance)
  • Do: Allow to check all possible boxes (players choice, AND traits)
  • Available points, max points per skill are already a big restriction

Tree VS Virtual Matrix

  • Trees (oriented graphs) have
    • Physical links (UI don’t scale, invasive)
    • Graph dependencies (slow cancel, few combos)
  • Virtual Matrices have
    • Virtual links (scalability, compactness, density)
    • No dependency (granularity, composability, decoupling, billions of combinations)
  • Mathematics
    • Let N be the number of nodes, L the number of links, C the number of combinations then L = N x (N - 1) / 2 and C = N! (also called Factorial)
      • https://www.google.com/search?q=pmp+formula+number+of+links
      • https://www.google.com/search?q=wikipedia+factorial
    • The number of links and combinations is exponential: UI cannot display all of them
    • To avoid perf bottleneck, devs reduce links, nodes, combinations: They create a tree
    • Better solution: Don’t display the links. Use Virtual links from Virtual Matrices
    • Bonus: Virtual Matrices simplify UI (PC and console)

Mathemarics

2 options for ARPGs:

  • Matrix → Amputated and Ordered → Tree
  • Matrix → Compacted → Virtual Matrix

For Matrices:

Let N be the number of Nodes
Let L be the number of Links L
Let C be the number of Combinations

L = N x (N - 1) / 2

C = N! (also called Factorial)

// Factorial
const f = (n) => {  
  if (n < 0) { return undefined; }
  let c = n || 1; 
  for (i = 1; i < n; ++i) { c *= i; }
  return c;
}
console.log(f(-1)); // undefined
console.log(f(0)); // 1
console.log(f(1)); // 1
console.log(f(2)); // 2
console.log(f(3)); // 6
console.log(f(4)); // 24
console.log(f(5)); // 120
console.log(f(6)); // 720

Bonus - Accelerate

Spell « Accelerate » in Dungeon Keeper
https://dungeonkeeper.fandom.com/wiki/Speed_Monster

  • Increase all speeds:
    • attack speed
    • movement speed
  • Could be a curse for minions (10s)
  • Base effect would accelerate minions
  • A rune would also accelerate enemies (for Thorns)

Bonus - Runes as Gems

Crazy idea:
Runes are collectibles gems.
Every game is different (runes will drop in a random order).
All skills stay accessible at level 1.

3 Likes

I am not a fan of skill trees (too linear, too restrictive). The goal should be a skill-web.
With the difference being more branching, more connections; more paths to get from A to B overall.

However, also not zero paths (or “all paths” if you will) as with the virtual matrix scenario.
Having to take a path to get from A to B is more restrictive for sure, but that restriction also comes with one major benefit; it can be used as a balancing measure. Adding more opportunity cost to getting specific nodes.
Adding more pro’s and con’s to taking different paths across the tree or web. Which generally can lead to more interesting choices.
Instead of just picking freely across everything, allowing for full optimization with no downsides.

That would be really bad.
An adaptative build is an oxymoron tbh. Builds should be sticky. Choices should matter.

All skills definitely should be available at lvl 1! Absurd idea that we have to lvl up to get access to some skills. That also leads to a linear lvling experience.

In general, the skills themselves should not really be part of the skill tree or web. They should be the start of it.
Preferably by giving each skill their own separate tree/web!
But if the current D4 model, with a large shared tree/web for all skills are used, then the skills all need to be at the center/root, with no skills having any prerequisites. Only the skill upgrade nodes should have that.

What do you mean virtual matrix? A board with many holes like POE? No thanks. Tropophobia, dude.

Not necessarily.

Temporary association

I see the build as a temporary snapshot of nodes, just like a project is a temporary association of people. It has a beginning, and a end.

Stickyness sources

The build has already stickyness sources.
It is sticky to:

  • the activity performed (N monsters, 1 boss)
  • the current level (spendable points)
  • the gear looted (available items combinations)

Armory

The need for an armory that stores builds is a good example of this. Players want to try different styles and save them.

Commitment

The character has a name, a personal link to the player who invested time in it, and matters more than the build the character temporarily embodies.

Emotion

This why some players prefer to change the build rather than recreate another character, as they are emotionally bounded with their current character. Gamification is all about emotion.

Leveling

Error and trial always happens during the leveling phase, as randomness is imposed on player. Player needs options to answer this unexpected changes (tactics known as contingence in project management) so it can reverse a bad fate (RNG) and keep a good player experience.

Subtrees are trees

Subtrees are trees, and trees (oriented graphs) are a mathematical structure that drastically reduces the number of possible combinations.

Slicing a doomed mathematical structure into smaller doomed parts is a commendable effort to digest its inherent defaults, but is probably the wrong approach. Choosing a better structure would be the rational choice.

Global view VS Subscreens

A global view on runes, while super fast and very practical for points spending / points cancellation, can be intimidating for new players.

Valid argument.
Tropophobia has to be taken into consideration.
Either Richness or Emptyness can be intimitading at first glance.

I’ve seen in the past two interesting approaches:

  • Last Epoch use a subscreen per skill to display the runes
  • D3 / Wolcen use a tab per skill to display the runes

They are both good compromises to mentally separate the runes from the skills and contain the visual complexity for beginners.

That being said, I think if done properly (consistent color, spacing and separation for color blindness), the global view should be the target, as the endgame, thus the number of allocable combinations, and how fast actions can be performed (the less clicks the better), should prime over all other considerations.

Maybe a configurable approach in the game options would serve both types of players:

  • beginner (subscreens by default)
  • experimented (all in one screen).

OR choices for runes

If devs really want to introduce OR choices for runes, I see several options:

  • Sets with many pieces: Strong identity, only one set can be equipped at a time, which effectively realizes the choice: Acceptable from players perspective
  • Book of the Dead: Allow max N minions and type the minions
  • Skills having same tag:
    • Skeletal Warrior: [Bone][Summoning]
    • Skeletal Mage: [Bone][Summoning]
    • Skeletal Wraith: [Bone][Summoning]
    • Skeletal Reaper: [Bone][Summoning]
      Each skill have different runes.
      The number of minions can be limited either:
      • globally, priority given to the first equipped skill
      • By allowing only Z summoning skills to be equipped
  • Pluggable runes: Typed Gems like Final Fantasy 7 Materias having their dedicated socket type (N Gems but only one can be socketed in the given socket). Warning: Requires a lot of discipline to never fall into subtrees.

Items linked sockets (Bad)

If devs take the FF7 option, which is the way Path of Exile took, an additional (Bad) strong OR option is to let RNG gear restricts sockets (runes):

  • How many
  • Which types
  • How they are connected (links)

I would never recommend this FF7 / POE approach as it carries all the flaws of the subtrees, thus considerably restricts the number of combinations.

In conclusion, choices often feel imposed to beginners, especially when they cannot go back, or if the price to go back is too high.

Players should have the right to make mistakes (error and trial), and error should be as free as possible to reduce the friction.

This is why I’m advocating players choice first (AND choices, virtual matrix, free refunds).

Give the player building blocks (legos) and its various preferences will naturally be expressed over the time.

i do understand the problem with prerequisite skills
but you somehow gotta make it feel…feely aswell
not just make it look like an exel sheet you know
in theory it’s “more choices”
sometimes finding a way through a tree is a fun game in itself tho (if they are funnier than D2 trees)
the everything always mentality often leads to quick burnout in games
i don’t have a clear stance in this
both can work, both can fail

1 Like

That is the opposite of stickiness. One enemy is the smallest indivisible unit a build can encounter.

Decisions in combat should be the primary thing a player can alter, to adjust for unexpected changes (from enemies etc.), as well as RNG fate in gear, and not the least, the long-term consequences of their build choices.

Enemy does X? Well, I might have to adjust my combat gameplay with Y to counter it.
Gear drops push me in X direction? Well, I might have to adjust my combat gameplay with Y to counter it.
Build choices push me in X direction? Well, I might have to adjust my combat gameplay with Y to counter it.

The best approach for sure.
Well, with a web instead of a tree.
For one, it means the different skills dont have to compete for the same resource (skill points). Instead of having to choose between boosting one skill or the other, you can boost both.
In a genre that too often end up with people being “pushed” toward focusing on a single skill or two (usually attack skills), to maximize power, allowing players to enhance each of their skills to the same degree (from skill points at least, not from other sources such as gear) is just positive imo.

The different approaches are about much more than just about beginner friendly UI.

A tree or a web does not mean everything becomes OR choices, you can pick both A and B, it will just cost more points. So you might need to pick less A and less B to get both, or skip out on C (sure, an OR choice, but not between A and B, and since you could pick A and C instead, C also isnt bound to be an OR choice), for example.

In general, I would say there should be no, or at least extremely few, outright OR (as in “you can only pick A or B”) choices in skill upgrades. That is indeed way too restricting.
But it is quite fine if one AND choice costs more than another AND choice - due to the path needed to take to reach them.
Like; it is cheaper to pick Meteor and Fireball skill upgrades than Meteor and Frozen Orb skill upgrades (in a scenario with one big shared skill web/matrix)
Or if each skill has its own web: cheaper to pick Improved Single Target and Increased Burn Effect on Fireball together, than to pick Larger AoE and Increased Burn Effect.
Making both of these choices possible, but at different costs. For example because the latter choice would be stronger, and in a system with no restrictions, would be the best choice always - hence as said earlier, prerequisites and paths gives devs another tool for balancing.

basically D3 you mean :clown_face:

Can you illustrate a use case of what you mean by “web” with a picture ?

A web is where you can say pick a skill and that skill then has a radius of its own modifiers that you can pick as well.

They branch out. And that can spread around, like for cold, fire, electrical, etc.

1 Like

Yeah, no D3 runes :smiley: Well, even D3 breaks its own OR limitation in many cases, by giving items that activates all runes.

Well, I dont even mind the D3 skill runes. Could be a fine system in some other system. But not the skill tree/web. Imo.

PoE and Wolcen are the best examples I know from any A-RPGs


But basically just a system with lots of connections between the different nodes, leading to multiple paths to the same nodes. Unlike in a tree design, where you go down one branch, to reach a node.

Note, a web does not need to be as absurdly expansive and complicated as PoEs and Wolcens.
Like, if they had one web per skill, it would of course be a lot smaller and less complex.

Like, Last Epoch skill nodes is mostly a tree layout, branches going in different directions.
But here is an example, in the upper right corner, where there are two connections, allowing two paths to the same end nodes. More of that imo.

literal clown game lol xD

D3 developers often said that activating all runes was not always possible because of the original OR design of the runes.

On the other hand, players often want to activate ALL the runes, because of their natural greedyness, which argues in favor of a AND design.

Got it, by web you meant cyclic graph, which is still a tree, but with intermediate connections acting as shortcuts.

Node reachability

It is very clear than in Wolcen, POE, Last Epoch and any tree-based skills selector, once you start in one direction (ex: far right) it is nearly impossible to get a node from the other side (ex: far left) without spending lots of points, as the positionning of nodes obeys a UX law rather than a gameplay law.

Combinations are reduced for an irrational motivation.

Rotate layers does not help

Wolcen tried to circumvent this problem by adding rotating layers, but in practice it doesn’t work, because when you rotate to connect an interesting mini-tree at the left, all the layer is rotating, breaking your mini-tree connection at the right.

The rotating Gate of Fates is a beautiful example of footgun.

Just like for Path of Exile, the existence of mini-trees, plus the fact that the positionning of nodes obeys a UX law rather than a gameplay law, breaks in practice the connectivity thus artificially reduces the number of combinations.

Justify precedence

Another problem with the trees is to justify the precedence of one node to another (dependency).

In 4X games like Civilization, where research and discoveries naturally follows the order of history, precedence is meaningful.

In ARPG, it seems very hard to justify a precedence of a node VS another, other by developers’ will.

There is often no rationale at all.
This become rapidly frustrating.

Players ask: “Why should I waste X points to get this node ?”
Answer is: “For no reason. Even developers have no idea. UX department said it looked good.”

Given points reflect time investment in the game, and the fact players investment should be respected, the game should avoid forcing to waste points just to traverse useless nodes to reach the node it really seeks.

Respectful rule

  • All nodes (skills, runes) should be accessible at level 1
  • Which implies no link
  • Which implies no tree
  • Which implies a Skills matrix

active skills should never have active skill prerequisites.
but passive upgrades can be tier based and require finding a path to reach them
which is more interesting than just hording points in order to activate them with a universal accessibility

2 Likes

Somes runes are indeed more powerful than others.

True if the node is limited to 1 spendable point.

But in the proposal, you can spend multiple points. :100:

As the maximum number of points vary depending on the rune, you can reflect the tiers by the number of spendable points.

For example, a rune of Army of the Dead has a maximum of 50 points, which clearly indicates that it is a powerful rune.

Zoom on Army of the Dead 2/50
https://i.ibb.co/LzD88Wm/diablo-4-proposal-1-skills-matrix-aotd.jpg

The number of points is relative to the power of the rune, in order to bring balance.

The more the rune is powerful, the more spent points it requires to fully express its power.

By using this system, any rune, even the strongest, is accessible at level 1.

Not at full power, but still accessible.

This design eliminates the annoyance relative to the waste of points.

To illustrate, for Army of the Dead it could be:

  • A)
    • +1 Skeleton, 3/15
    • +1 Second, 2/50
      3 Skeletons, lasts 2 seconds
  • B)
    • +1 Second, 3/15
    • +1 Skeleton, 2/50
      2 Skeletons, lasts 3 seconds

Or in favor of devs not giving players what they think they want.

Hence, not a tree. A tree usually dont have branches growing back together.

But yes, intermediate connections between more of the nodes. However, not between all nodes.

Making it harder to reach the opposite side can very much be a gameplay law.
The UI is not a cause, but an effect, of the gameplay design.

It was an interesting idea for sure, but I’d be careful with using it generally. It risks offering too much freedom, defeating the purpose of the web design in the first place.

One node can quite naturally lead to another.
Like, take Fireball as a skill. Maybe one node adds a DoT. And another adds an additional effect to the DoT. Wouldnt make much sense if you could pick the second one, without the first. At least it would only serve as a “noob trap”, since the second effect would not work on its own.

But even when there is no direct connection, where effect B only can work if effect A is active, the two can just be connected from a gameplay design perspective, where the devs want you to go through A, B or C, to get to D, because D is considered to be a higher tier of power, which they want to make more costly to get.
Or to enforce themes around choices. Like, maybe effect D is some very iconic fire effect, and only a sorceress who has dedicated themselves to the art of fire spell casting, should be able to realistically reach it. And so on.

I dont want a lot of those kinds of linear prerequires (hence arguing for a web, with multiple paths), but they can certainly make perfectly fine sense from a game design and gameplay perspective.

It is not a waste of points. It is paying a fair cost. It is making choices, that have different pro’s and con’s.
You still get those things you spent points on along the way. It might not be what you truly wanted, but it still shapes your character.

Being able to choose everything freely, inconsequential, with no direction or coherence, might offer more freedom, but also make everything feel pointless. Making it lack a sense belonging/fitting to a larger whole; the development of a character.

Much more than time spend, it reflects a journey. The journey of the character that is being played. The time spent is more of an afterthought, that comes from the journey taking time. Both due to gameplay pacing, devs wanting each part of the journey to have some breathing time before you move to the next step, and because, well, the actual stuff you do to lvl up, such as killing monsters in an A-RPG, is going to naturally take some amount of time, unless the combat depth is paper thin.

All skills should be available at lvl 1 for sure.
All upgrade nodes (what you call runes) should not. Or rather, they technically should be available at lvl 1, there should be no lvl requirement on them (unlike in D3), but you just wont have the points to reach them at lvl 1.

Indeed.

1 Like

I think it would be a fun visual if the skills were rotatable and interlocked into and with each other, like gears, with the points of convergence being slightly better than the others so that one can customize their options yet still retain synergy.

LE has the best skill model in my opinion. You can access most skills, provided you unlock them, then you pick a path within that skill.

It isn’t over engineered. It offers enough configurations for different builds. Enough choice to even turn dps skills into support skills.

They did really well with skills, however, their item effects could use more synergy. I think most players don’t appreciate’s D3’s legendary item interactions enough.

2 Likes

I understand the intent.

That being said, you can equip an item that gives an effect on a skill / rune you never equipped, and basically while it does nothing (brings no bonus), you can still equip the useless item.

I’m not saying this is bad or good, just pointing that it is possible to equip useless powers.

It is part of the combination theory, and the job of the player to understand the underlying mechanic to find the best possible combination.

Developers choice VS players choice (Meme 1)
https://i.pinimg.com/736x/fc/cc/a7/fccca79daa4ef22b91b4ad80cab3695b--user-experience-design-ux-design.jpg

It works with the current proposal, as powerful runes for Army of the Dead require more points to spend to acquire their full power.

This is a well-known programing problem. :robot:

Rather than precedence (inheritance), it is best to use synergies (traits).

It is a small push in one direction, but never a forced option.

One problem with a web (possibly cyclic graph) is the player don’t control the shortcuts.

Players will eventually trace their own shortcuts.

Developers choice VS players choice (Meme 2)
https://miro.medium.com/max/960/1*pMk3h0dIYMb_I1iJCjriPQ.jpeg

Like an optional campaign is needed to give a sense of linearity.

This is why I would advocate for small synergies / traits.

Optional Linearity.

Wolcen’s Gate of Fates proved it was a bad idea (not working).

Skills synergy is okay if not overwhelming (small and balanced, like 5%).
It is rewarding, but never coercitive.

For now.
But we can do so much better.

thats what i said basically

It is unlikely however that the item is completely useless. Even if the legendary effect is. It still has other affixes, and at minimum some Defense/Armor value, dmg or whatever. You might want to use the item, even if specific affixes are not useful.
Albeit imo the goal should overall be, that most affixes always do something, for all characters. Even if they might not do much.
Items can also easily be equipped and unequipped.
Skill nodes should not.

A good designer will make sure to close off the alternative path when it is found.

It is ALWAYS developer choices. The game was designed by them. Even if you can pick freely, that was the developer choice.

I am strongly against predesigned builds. That is a big part of what is wrong with set items in D3.
But overly deterministic paths being bad, also does not mean that no limits is the right approach.

But not more points to acquire some of its power.

Synergies might mean you will be stronger with fire by focusing on fire, a real problem that D2 suffers from, greatly limiting build options. That should not be the goal imo. A hybrid should be just as strong, in its own way, as a build dedicated to fire.
But the build dedicated to fire might have fire effects that the hybrid cant get. Which is what prerequisites can offer. Making the dedicated fire build feel unique and different from a hybrid that could just pick whatever fire skill for their toolset, due to no restrictions existing.

Now, of course, you can still have restrictions in a virtual matrix. Like, fire nodes might require X points spent in other fire nodes, to unlock access. That can work fine to create tiers of nodes. But it still lacks the benefits of interesting character build decisions, that having to come up with a path from A to B can offer.

That isn’t a problem, but specifically the intended benefit of the system.

Which again is something you can relatively easily stop in a game (while somewhat harder in a real world scenario).

Imo skills should never have “artificial” synergies D2 style (like points spent in fireball randomly increasing your meteor dmg, for reasons). Synergy should come more naturally, from skills interacting in combat. Like, a frost nova skill might synergize well with meteor, by making it easier to catch enemies within the meteor blast radius.

Yeah. Haven’t played LE. Really looking forward to it. But from afar, it seems like the best skill-upgrade system in an A-RPG so far.

1 Like