45 replies to this topic

[Alistair Winter]

A little while back, there was a little bit of talk in GD about community councils and town halls and showing Russ that he should sit down and listen to the players. As expected, it was mostly just talk and nothing came of it, as far as I know.

Now, there's a lot of of discontent about PGI's new skill trees. So here's my question:

To everyone who feels that PGI should actually use some sort of skill tree to promote role warfare and give players more freedom to customize... can you do better than PGI did?

In other words, if you think that skill trees are a huge mistake and PGI should just remove skills all together, then this is not for you. If you prefered modules or have a completely different solution, then this is not for you. But if you want to see some sort of skill tree that resembles what PGI is working on today, can you actually do better than them?

Note: The goal should be to avoid giving every mech in the game all the best skills (e.g. speed tweak, radar deprivation, seismic sensors, bonus armor, etc) and end up with the majority of skills being completely pointless and overlooked. If you give players total freedom, there's really no "tree" at all. Then we're just shopping skills, just like we used to shop modules. So if that's what you want, again, this is not for you.



I'd love to see people post some suggestions in this thread, and then I will edit them into this opening post.

Here's one suggestion, which is very tentative, just to get the ball rolling. Obviously, the idea is that you want the most valuable skills (armor bonus) to require some investment, so they're not low hanging fruit. But you also want certain skills that are critical to certain roles to be available without requiring full investment into that tree. So I can have a triple AMS Kit Fox without maxing my 40 different armor skills, and I can have reduced fall damage if I have a jump jet mech, but I don't want to invest fully in the armor skill branch.


So here's one suggestion from me. I'd be happy to see other suggestions, to check if the players are able to come up with some sort of consensus, or if the players are completely divided and don't know what they want as a group.

Edited by Alistair Winter, 03 March 2017 - 03:41 AM.

[Doctor Dinosaur]

I'm so free to post the following topic here with a suggestion for the firepower tree:
https://mwomercs.com...80#entry5637580

[Zeleglok]

I don't like both, they're zero variety. The only difference is that you spend more points in PGI version, which is probably better, since other branches require points as well and it might result in having to actually choose what branch to unlock.

[WolvesX]

This is soo much better!!! LIKE

---

On a side note... why have links between the nodes at all?

[AngrySpartan]

Alistair Winter, on 02 March 2017 - 01:13 PM, said:

If you prefered modules or have a completely different solution, then this is not for you. But if you want to see some sort of skill tree that resembles what PGI is working on today, can you actually do better than them?

That's not completely in line with PGI "tree" aand it's so called "structure". Yet, I'll leave it here, since my suggestion does exactly this:

Alistair Winter, on 02 March 2017 - 01:13 PM, said:

Note: The goal should be to avoid giving every mech in the game all the best skills (e.g. speed tweak, radar deprivation, seismic sensors, bonus armor, etc) and end up with the majority of skills being completely pointless and overlooked.

...idea is that you want the most valuable skills (armor bonus) to require some investment, so they're not low hanging fruit. But you also want certain skills that are critical to certain roles to be available without requiring full investment into that tree. So I can have a triple AMS Kit Fox without maxing my 40 different armor skills, and I can have reduced fall damage if I have a jump jet mech, but I don't want to invest fully in the armor skill branch.

https://mwomercs.com...le-alternative/
Witcher 3 style structure of a skill tree. Main Idea - to get to the most desirable nodes, you have to unlock more nodes in a respective tree.
Image:

Spoiler

Mobility tree example:
Tier 1:Acceration, Decceleration and "junk" nodes opened all the time.
To open Tier 2 (Twist speed, Turn speed, Torso Yaw+Pitch, etc.) you have to spend unlock 8 nodes in that tree
To open Tier 3 (Speed tweak) you have to spend 24 nodes in that tree.
Nodes are not connected, number of skill points is reduced, all "junk" nodes have 1-3 levels max.

Edited by AngrySpartan, 03 March 2017 - 12:24 AM.

[Alistair Winter]

Doctor Dinosaur, on 02 March 2017 - 11:47 PM, said:

I'm so free to post the following topic here with a suggestion for the firepower tree:
https://mwomercs.com...80#entry5637580

Definitely.

The reason I don't like that tree system, and all the other similar ones that have been posted, is that their tree structure is irrelevant. They have removed so many restrictions that it becomes a pointless aesthetical representation, which effectively gives you total freedom, with a few minor exceptions. Check out these different paths on Andi's skill tree:

Max range (10) = 10 skill points, of which 0 are mandatory points
Max heat gen (8) = 10 skill points, of which 2 are mandatory points
Max cool down (11) = 12 skill points, of which 1 are mandatory point
Max velocity (5) = 8 skill points, of which 3 are mandatory points
Max laser duration (10) = 13 skill points, of which 3 are mandatory points
Max missile skills (12) = 16 skill points, of which 4 are mandatory points.

Also, the mandatory points are usually completely random. For example, if I want max missile skills, then I have to take range and heat gen skills to get the shortest path. If I want ballistic skills, then I have to take 1 range and cooldown.

Why is cooldown the path to ballistics, while range is the path to missile weapons? What if my missile boat needs cooldown more than range, what if my ballistic carrier needs range more than cooldown? Then I have to take more mandatory skills.

I want a system that doesn't force me to take all these tiny mandatory skills with their tiny bonuses, filling up my list of quirks. Like "30% heat reduction... and 1% extra range, 1% lower cool down, 1% extra velocity and 1% smaller missile spread". That's annoying to me, personally.

Zeleglok, on 02 March 2017 - 11:52 PM, said:

I don't like both, they're zero variety. The only difference is that you spend more points in PGI version, which is probably better, since other branches require points as well and it might result in having to actually choose what branch to unlock.

The idea would be that no branches should require you to take pointless skills, relative to whatever role you want to play. You may not get the very best skills immediately, but you shouldn't be forced to take pointless skills. For example, if I want to have the fastest mech on the battlefield, I shouldn't have to take arm pitch skills before I get speed tweak. Maybe I have to take acceleration skills, which are related to speed, in order to reach speed tweak skills, which are more valuable. But I shouldn't have to take 5 random skills in order to get the one I want.

That's my take on it, anyway.

In reality, by forcing people to spend more points on random skills, you're reducing variety, you're not creating variety.

WolvesX, on 03 March 2017 - 12:07 AM, said:

This is soo much better!!! LIKE

---

On a side note... why have links between the nodes at all?

You can go with the Witcher 3 structure that Angry Spartan has suggested, in order to eliminate the need for nodes. But I'm not sure PGI would go for something so radically different. After all, they already paid some graphical UI designer to make those hexes, so I'm sure they are determined to use them.

AngrySpartan, on 03 March 2017 - 12:23 AM, said:

That's not completely in line with PGI "tree" aand it's so called "structure". Yet, I'll leave it here, since my suggestion does exactly this:

https://mwomercs.com...le-alternative/
Witcher 3 style structure of a skill tree. Main Idea - to get to the most desirable nodes, you have to unlock more nodes in a respective tree.
Image:

Spoiler

Mobility tree example:
Tier 1:Acceration, Decceleration and "junk" nodes opened all the time.
To open Tier 2 (Twist speed, Turn speed, Torso Yaw+Pitch, etc.) you have to spend unlock 8 nodes in that tree
To open Tier 3 (Speed tweak) you have to spend 24 nodes in that tree.
Nodes are not connected, number of skill points is reduced, all "junk" nodes have 1-3 levels max.

I would be very happy with this, if PGI was up for it. However, I don't see why you need Tiers. Why not put skill barriers for each individual skill? E.g. you need 5 skill points in mobility to unlock torso twist, but you need 10 skill points in mobility to unlock speed tweak. This makes it easier to balance the system.

[AngrySpartan]

Alistair Winter, on 03 March 2017 - 12:56 AM, said:

I would be very happy with this, if PGI was up for it. However, I don't see why you need Tiers. Why not put skill barriers for each individual skill? E.g. you need 5 skill points in mobility to unlock torso twist, but you need 10 skill points in mobility to unlock speed tweak. This makes it easier to balance the system.

"Tier" is a strong word for me as well. Consider it just a term to define how the system works, there is no particular reson to put nodes in Tiers, except it makes the whole tree more structured with clearly defined horizontal levels.

Structure (or rather lack of structure) is my biggest complain with PGIs UI so far.

Alistair Winter, on 03 March 2017 - 12:56 AM, said:

You can go with the Witcher 3 structure that Angry Spartan has suggested, in order to eliminate the need for nodes. But I'm not sure PGI would go for something so radically different. After all, they already paid some graphical UI designer to make those hexes, so I'm sure they are determined to use them.

In fact it's not that different in terms of UI:
1. Nodes are not connected
2. Different node levels are "stacked" into single hexagon.
Not perfect, but suitable for reuse if PGI is not willing to develop different graphics for skill tree.

Edited by AngrySpartan, 03 March 2017 - 01:09 AM.

[Old-dirty B]

My suggestion / idea for the skill tree is rather simple but perhaps a bit too much to draw it al out in such a nice drawing as the op.

My idea is perhaps not really a tree, maybe a title such as "Mech tune-up system" or "Mech performance control unit" or ...?

Anyway, the idea is as following:

1. Each and every individual improvement that can be made gets its own "upgrade" bar. A full bar means 100% or the full improvement. The total improvement is a global percentage value applied to base values (that are set by tonnage and are derived from engine choice) and apply to all mechs across the board, the same amount for 20t to 100t. There's one area of exception, weapon specific improvements are grouped: velocity, duration and spread (max effect should be determined per weapon, for example unique velocity increase for srm's, lrm's, lbx's, ac's, ultra-ac's, ppc's etc).

2. Each bar has 10 steps, with each step the effect of the improvement is gained by 10% until the bar is full or 100% of the improvement is in effect.

3. Each 100% improvement or a full bar should be "weighted" against all other 100% improvements in order to establish a global economy system for valuing all improvement steps. A max acceleration improvement of 10% might be valued more then a max weapon range increase of 10% (or whatever the max improvement should be), a full acceleration bar costs perhaps 10 skill points while a full range increase costs 6 points.

4. Now that all steps have been weighted and valued in "skill points" a total amount of skill points should be determined per chassis and variant. The weakest or worst variants (perhaps the Cicada CDA-3C ?) get more skill points then the best variants (perhaps the Kodiak KDK-3 ?).

5. The value of a skill point in terms of XP and C-Bills is determined by dividing the global costs to fully master a variant by the total amount of skill points for both XP and C-Bills.

Balancing gameplay can be done in two ways, via the costs of a step (node) in terms of skill points and via the total amount of available skill points. I would advise against balancing via the total amount of improvement for individual chassis and variants as that limits diversity and forces users into a specific direction (as we already have with the current quirks system)

Well, this would be my way to fine-tune mechs, to add build diversity and to balance gameplay.

EDIT:
To give an example of how this system plays out in regards of balancing mech variants. Lets assume the KDK-3 is the very best variant which gets the lowest amount of skill points while the CDA-3C is considered the worst and gets the most skill points. The kodiak, already being a high performance mech, is considerably more expensive to upgrade. Perhaps only upgrade one major (expensive) improvement (cooldown for example) and a few minor (cheap improvements.
While the Cicada could perhaps get a big upgrade with many important improvements and almost all minor improvements.

Edited by B3R3ND, 03 March 2017 - 03:24 AM.

[Kuaron]

Alistair Winter, on 02 March 2017 - 01:13 PM, said:

I’ll take 4x yellow, all green and all blue.
On all Mechs.
Thanks.
Decision making is tiring anyway.

[Oberost]

I posted on another topic during the PTS1...

This is a tier based skill tree:



In this tree you can choose only one skill from each Tier, and every skill is balanced with the other skills from that Tier. The result is this:



If you balance things to be almost equal in usefulness you will have to make REAL choices, because picking one skill will prevent you from picking the other two useful skills on that Tier.

Do the skill trees different AT LEAST for each weight class and you'll have some more balance and variety.

Hell, you can even use hexagons instead of rectangles if this fits you better...

Edited by Oberost, 03 March 2017 - 03:36 AM.

[Alistair Winter]

B3R3ND, on 03 March 2017 - 03:03 AM, said:

My suggestion / idea for the skill tree is rather simple but perhaps a bit too much to draw it al out in such a nice drawing as the op.

I'm not sure I understand your idea. Would it be fair to say that it's basically a module system, except each mech gets X number of modules? And instead of 1 module per ability (e.g. 1 radar deprivation module), you are essentially splitting each ability into different modules? So you might need 10 radar deprivation modules, or 6 speed tweak modules, etc?

I'm just trying to frame your idea using terms that are already familiar to us, in order to avoid making things too complex. It's kind of hard to visualize 148 different skill bars between 0% and 100%.

[Kuaron]

Ah well, since we were asked for our suggestions:

Measure (1) from this thread is one:
https://mwomercs.com...-trees-banaced/
(Measure (2) was rather a quick idea at the time posting.)

This is ofc. not the only one, the forum has several.
PJohann posted a nice one here, for example:
https://mwomercs.com...ost__p__5637557

(All links provided contain self explaining pics!)

[Alistair Winter]

Kuaron, on 03 March 2017 - 03:27 AM, said:

I’ll take 4x yellow, all green and all blue.
On all Mechs.
Thanks.
Decision making is tiring anyway.

Joke's on you. You'll never enjoy playing a double AMS Wolfhound that can fling itself from roof tops without taking damage. Armor and structure is overrated.

Oberost, on 03 March 2017 - 03:28 AM, said:

I posted on another topic during the PTS1...
This is a tier based skill tree:
In this tree you can choose only one skill from each Tier, and every skill is balanced with the other skills from that Tier. The result is this:
If you balance things to be almost equal in usefulness you will have to make REAL choices, because picking one skill will prevent you from picking the other two useful skills on that Tier.
Do the skill trees different AT LEAST for each weight class and you'll have some more balance and variety.

I like it. It's basically like the Witcher 3 skill system, from what I understand. It's similar to the Diablo skill tree, except you can't have your cake and eat it too. You're forced to make choices on each branch. In MWO terms, this would be like if picking Radar Deprivation meant you could not also have Seismic Sensors.

Personally, I prefer a straight linear skill tree, because mutually exclusive skills eliminates freedom. For example, what if I am forced to choose between max range and max projectile velocity? Suddenly all the best long range mechs will be laserboats, instead of gauss snipers. This is just a hypothetical example, but you can imagine similar situations. What if I want extreme jump jet power for my light mech, but now I have to choose between jump height and jump acceleration? That would be sad.

That being said, Tiers do not necessarily mean that you get up with mutually exclusive choices. Diablo effectively had Tiers by restricting certain skills and spells until you reached a certain level. So you couldn't have Fire Novas at level 5 or whatever.

I like Tiers, but I'm not sure I like mutually exclusive skills.

[Kuaron]

Alistair Winter, on 03 March 2017 - 03:40 AM, said:

Joke's on you. You'll never enjoy playing a double AMS Wolfhound that can fling itself from roof tops without taking damage. Armor and structure is overrated.

FIFY.

[Alistair Winter]

Kuaron, on 03 March 2017 - 03:45 AM, said:

FIFY.

But I preferred mine.

[Old-dirty B]

Alistair Winter, on 03 March 2017 - 03:30 AM, said:

I'm not sure I understand your idea. Would it be fair to say that it's basically a module system, except each mech gets X number of modules? And instead of 1 module per ability (e.g. 1 radar deprivation module), you are essentially splitting each ability into different modules? So you might need 10 radar deprivation modules, or 6 speed tweak modules, etc?

I'm just trying to frame your idea using terms that are already familiar to us, in order to avoid making things too complex. It's kind of hard to visualize 148 different skill bars between 0% and 100%.

Yes, my idea is to make a branch per ability / module, each divided in 10 steps or nodes. I would list all the abilities horizontally and each ability gets a string of 10 nodes vertically, using your example (which you have made very well ) you would get a green string, a red string, a blue string, a yellow strings and so on.

To fill all nodes for an important branch you need more skill points then to fill all the nodes for a lesser branch.

Does that make sense?

(perhaps i should make a drawing anyway lol)

Edited by B3R3ND, 03 March 2017 - 03:53 AM.

[Alistair Winter]

B3R3ND, on 03 March 2017 - 03:50 AM, said:

Yes, my idea is to make a branch per ability / module, each divided in 10 steps or nodes. I would list all the abilities horizontally and each ability gets a string of 10 nodes vertically, using your example (which you have made very well ) you would get a green string, a red string, a blue string, a yellow strings and so on.

Does that make sense?

(perhaps i should make a drawing anyway)

Yes, I think I understand what you mean now. But in my mind, it's better to talk about this like an update to the module system rather than a skill tree. Because it's not really a tree in the sense that it has different nodes that branch out from each other. I'm not accusing you of this, but I think there's a problem where people are talking past each other, because they're using the wrong words. They're saying "I want my skill tree to look like this" when they actually mean "I don't want a skill tree, I want something completely different".

As for your idea, I am not strictly opposed to that kind of a module system. It's certainly better than what we currently have. Especially the idea of making a full seismic sensor require more skill points than, say, a full torso twist upgrade.

[Old-dirty B]

agreed, i was aware that my idea wasn't really a tree - i notified about that in my post anyway, i just noticed in your drawings that my idea would be rather similar to the system shown on the very left (full freedom - modular). Important in this idea is how to handle the improvements, the node costs and available skill points.

All in all, my idea is more about a system to balance and add another layer to create variety and not so much about advancing along a system to upgrade (like rpg's and such). If it would be totally on me i would take out the whole grind, xp and c-bills aspect and give all the skill points right away with no costs for speccing / respecting just to fine tune or customise a mech tailor made to the user's preferred way of play.

Edit:

Alistair Winter, on 03 March 2017 - 03:57 AM, said:

... Especially the idea of making a full seismic sensor require more skill points than, say, a full torso twist upgrade.

tnx, personally i like it even more when the same upgrade on a different, better or worse, mech has a different value. That means you have to determine per mech what the best value would be that you can get out of the available skill points as that differs per chassis and variant.

Edited by B3R3ND, 03 March 2017 - 04:19 AM.

[Radkres]

https://drive.google...MEhYSTQzdWxEVHc

I was thinking about if You converted what we have now to a node system what would be the cost break down.

in General lets say that all range modules are now 5 nodes with the normal progression of 2,4,6,8,10 percent once you have them all unlocked at a cost of 500, 600, 700, 800, 900 GXP or 3,500 GXP
So You would have to unlock all 5 nodes to get your 10% Range node for that Weapon.

Similar for the Cool Down Module if it also became nodes Same Cost but greater Percent per node. 2.4%,4.8%,7.2%, 9.6%, 12% . cool down for that Weapon. Same Cost in GXP

But Here is Where the cost change for those Nodes.
Examples:
6 million Cbill Module would cost 1,200,000 Cbills per node.
4 million Cbill Module Would cost 800,000 Cbills per node.
3 million Cblill Module would cost 600,000 Cbills per node.
2 million Cbill Module would cost 400,000 Cbills per node.

I Also broke down the current Skill tree into nodes also.

Each skill now has 5 nodes
Basic Skills Mech
Cool Run node 150 GXP per node 3% node
Kinetic Burst node 200 GXP per node 3% node
Twist X node 500 GXP per node 1% node
Heat Containment node 200 GXP per node
Hard Brake node 300 GXP per node 3% node
Twist Speed node 500 GXP per node 1% node
Arm Reflex node 300 GXP per node 1% node
Anchor Turn node 700 GXP per node 1% node

Elite Skills Mech
Quick Ignition node 800 GXP per node 6.6% node
Fast Fire node 1200 GXP per node 1% node
Pinpoint node 600 GXP per node 3% node
Speed Tweak node 1700 GXP per node 1.5% node

Master Skills Mech
Module Slot node 4300 GXP per node at 20% node

I drew out a spider node for this but have not found a way to post it.

Anyway was just thinking a way to convert the current system to nodes.

[Ruar]

A big piece of missing information is what value does PGI assign to different skills. For example we can assume speed tweak has a high value because it's at the bottom of the tree. Which means they want people spending 15-20 points to get speed tweak.

The problem is they buried speed underneath a lot of crap people may not want. Which causes several problems. The first is people feel like they are being forced to take skills they don't want. Second is we get inadvertent boosts which homogenize mech attributes instead of making each mech unique. Third is it becomes difficult to balance new content because skills associated with that content are entangled. Fourth is the effort necessary to click down through the tree and figure out all of the connections.

What they should do is go linear with all of the skills. So speed tweak should be something I can immediately select without having to unlock anything else. At the same time they adjust the cost so we have to spend 15-20 points if we want to improve our top speed. Perhaps they give every skill 10 nodes with varying amounts of return and varying costs. Perhaps they just do five nodes with higher costs.

It honestly doesn't really matter the exact proportion of nodes to cost as long as the end result is the player can click directly on the skill they want without having to dig through unwanted skills. PGI can then adjust the cost and returns based on player usage, current meta, new equipment, etc.

This makes skill selection much easier for new players, less tedious for older players, more intuitive for build purposes, and avoids the feeling of being forced to get something we don't want.

I honestly have no clue why PGI insists on burying skills under other skills considering all of the drawbacks it presents. My best guess is it was the brainchild of a higher ranking dev who refuses to admit there is a better way to make the skill trees happen.