25 replies to this topic

[MischiefSC]

In the dev stream we see some conversation from the devs on why they think a linear tree is bad. It concerned me because it ignores the point and nature of a skill tree - making choices to customize your mech.

All gating does is make any tradeoff that's not the best possible optimization a bad one. For example that you'd need to give up 5 points worth of useful quirks in ops or mobility for 4 bad quirks in sensor and 1 pt of Seismic or derp.

A linear tree works when you make the points of relative value. Like 0.5% reduced heat gen is 1 SP and 7% hill climb is one SP. This lets you trade relative values and make differing builds that are of approximately equal value. Giving up 2 or 3 engine ratings worth of speed plis other mobility quirks to get paper dolls to fill a fraction faster and 100m Seismic is a bad trade. Giving up 7% laser burn duration on an energy boat plus several % in reduction to heat costs and improved weapon cooldown to get miniscule buffs to JJs or 6 pts of total health on locations average is also terrible.

Gating makes tradeoffs possible, just a poor choice. It defeats the purpose of it. It makes the skill tree irrelevant to customization save in the same way single heatsinks and choosing ff over end is. It's the opt-in to make bad choices. That's it. It gives no good exchanges to make.

All customization is min/maxing. It's why we all pick DHS and endo and why all clans run CXL. For a good skill tree the point is to make a good value of exchange to make those tradeoffs worrh it. That means granular and of approximately equal value. Arm quirks are not worth as much as cooldown, so make 2 arm pitch quirks worth 25% each and 8 cooldown quirks worth 1% each. This means no matyer what I do I wont get more than 8% cooldown and with arm mounted weapons i may go 6% (a relatively small loss) to get better performance on most maps. If I have to give up 4%, plus some cooldown and range quirks to get that arm movement and some things I don't need then that's not a real tradeoff. It's just some bad decisions I can make.

In the existing system everyone gets cooldown, speed, etc. If they all get those to varying degrees in the new system that's not bad - there's things that are always a good idea. The ability to vary them by degree to make equally viable choices is what would make a good skill tree.

Gating doesn't. It just draws a clear line between a narrow set of good choices and a lot of bad ones. I get that in mech design because of aesthetics and hardpoints and lore there are going to be good and bad mechs and builds. There is no logic to forcing that limitation on the skill tree.

[B0oN]

How to stop a flood if there are no gates ?
In this instance gating has its uses, at least in my opinion .

[Bishop Steiner]

I am actually waiting to see someone demonstrate where the skill tree actually makes choices irrelevant. I keep building multiple different role focuses using it, so maybe I'm just too dense to see the issue?

Is it perfect? No. Is complaining about having to get 3 skill nodes you don't like to minmax your robot the end of the world? Well, I don't see it.

Note that even among Tier 1 and competitive players there is no consensus on good or bad, no matter how strongly you seem to be convinced there is.

Edited by Bishop Steiner, 30 April 2017 - 10:17 AM.

[MischiefSC]

Bishop Steiner, on 30 April 2017 - 10:16 AM, said:

I am actually waiting to see someone demonstrate where the skill tree actually makes choices irrelevant. I keep building multiple different role focuses using it, so maybe I'm just too dense to see the issue?

Is it perfect? No. Is complaining about having to get 3 skill nodes you don't like to minmax your robot the end of the world? Well, I don't see it.

Pick a mech/role.

[Bishop Steiner]

MischiefSC, on 30 April 2017 - 10:17 AM, said:

Pick a mech/role.

I have. Several times. I've built my PPC focused sniper K2, and then went totally differnet skill tree choices with my C1, becuase as a missile support mech, it didn't need the mobility that a sniper skirmisher build like the K2 did.

Most mechs I ended up only unlocking 3/4 of operations. A few I went the full monty, like the Jester, because it is a hot running build. My c1 could fire it's twin aLRM15s until it ran out of ammo, conversely, with 3/4 Ops nlocked, and never come close to overheating. etc, etc, etc.

But of course, onyl thing that matters is what you want to minmax for the Meta Box....of course, when everyone else has the same limitations....guess what you get? Generally a retardation of the power level.

You ignored or dismissed them last time, why should I go through all the effort for you to do the same this time?

Broken frikking record.

Edited by Bishop Steiner, 30 April 2017 - 10:28 AM.

[MischiefSC]

Bishop Steiner, on 30 April 2017 - 10:26 AM, said:

I have. Several times. I've built my PPC focused sniper K2, and then went totally differnet skill tree choices with my C1, becuase as a missile support mech, it didn't need the mobility that a sniper skirmisher build like the K2 did.

Most mechs I ended up only unlocking 3/4 of operations. A few I went the full monty, like the Jester, because it is a hot running build. etc, etc, etc.

You ignored or dismissed them last time, why should I go through all the effort for you to do the same this time?

Broken frikking record.

No I didn't. Havent discussed any specifics with you or anyone. Is the new pts up? You build one, I'll max all but 2 pts in ops and mobility each and then 19 pts in weapons and we can test it. Because more mobility, better cooling is better and gank > tank. Maybe some scout builds I'd go consumables over weapon quirks. Some sniper builds worth 2 or 3 pts in sensor for zoom. Of the 91 pts to spend there's 20 pts or less really "in play" and still making good choices. 90% of the time it's more like 5 pts.

Happy to test that.

[SuomiWarder]

In the current system there are skill boxes I didn't want to waste XP on like quicker restart that gated my advance to mastery for the extra module slot. No one cried back then.

[MischiefSC]

SuomiWarder, on 30 April 2017 - 10:34 AM, said:

In the current system there are skill boxes I didn't want to waste XP on like quicker restart that gated my advance to mastery for the extra module slot. No one cried back then.

Because getting quickstart doesn't keep you from getting rapid fire or cool running. You got all of it eventually.

[SuomiWarder]

Quick start did gate me from mastery though. Now that I think of it, getting full basic on two similar mechs I wasn't interested in playing 'gated' me from getting into elite on the one I planned to use.

[nehebkau]

MischiefSC, on 30 April 2017 - 10:13 AM, said:

Gating doesn't. It just draws a clear line between a narrow set of good choices and a lot of bad ones. I get that in mech design because of aesthetics and hardpoints and lore there are going to be good and bad mechs and builds. There is no logic to forcing that limitation on the skill tree.

Yep, the end result of the skill tree:

Potatos get even more starchy under the weight of bad skill-tree choices while the 'serious' teams simply develop a new meta to more quickly turn them into mash.

Edited by nehebkau, 30 April 2017 - 12:42 PM.

[MischiefSC]

SuomiWarder, on 30 April 2017 - 12:28 PM, said:

Quick start did gate me from mastery though. Now that I think of it, getting full basic on two similar mechs I wasn't interested in playing 'gated' me from getting into elite on the one I planned to use.

If you had unlimited points in the new system than your analogy would be correct. You don't, so it's not. Getting Cool Running never meant you would never be able to get fast fire. You did not have to choose between fast fire and speed tweak because if you got one you couldn't get the other and you couldn't get speed tweak without getting quickstart.

An unrelated analogy is not a good justification for a bad mechanic.

[Deathlike]

Bishop Steiner, on 30 April 2017 - 10:26 AM, said:

I have. Several times. I've built my PPC focused sniper K2, and then went totally differnet skill tree choices with my C1, becuase as a missile support mech, it didn't need the mobility that a sniper skirmisher build like the K2 did.

Most mechs I ended up only unlocking 3/4 of operations. A few I went the full monty, like the Jester, because it is a hot running build. My c1 could fire it's twin aLRM15s until it ran out of ammo, conversely, with 3/4 Ops nlocked, and never come close to overheating. etc, etc, etc.

But of course, onyl thing that matters is what you want to minmax for the Meta Box....of course, when everyone else has the same limitations....guess what you get? Generally a retardation of the power level.

You ignored or dismissed them last time, why should I go through all the effort for you to do the same this time?

Broken frikking record.

What Mischief is saying that the skill tree allows you to "get all of the relevant roles you want, but not really give up anything for the crappier ones". If each node isn't "balanced" against each other (in other words, 1 node of cooldown of some % should be equal to 1 node of something equally relevant, but not hill climb, +1 consumables, AMS overload, seismic, etc.), there are problems with the valuation between the nodes. The tree is constructed in such a way that even a min-maxer will realize which are the worthless choices (like JJs not really doing much boosting, specially on class I [Assault] JJs).

Choices should be more like "you can have duration or projectile velocity, but NOT have accel/decel or boosted JJs for better poptarting". Those are choices to be made... this tree gives you all the choices and all one (min-maxer) has to do is filter out "the worst options".

Edited by Deathlike, 30 April 2017 - 01:08 PM.

[Bishop Steiner]

Deathlike, on 30 April 2017 - 01:05 PM, said:

What Mischief is saying that the skill tree allows you to "get all of the relevant roles you want, but not really give up anything for the crappier ones". If each node isn't "balanced" against each other (in other words, 1 node of cooldown of some % should be equal to 1 node of something equally relevant, but not hill climb, +1 consumables, AMS overload, seismic, etc.), there are problems with the valuation between the nodes. The tree is constructed in such a way that even a min-maxer will realize which are the worthless choices (like JJs not really doing much boosting, specially on class I [Assault] JJs).

Choices should be more like "you can have duration or projectile velocity, but NOT have accel/decel or boosted JJs for better poptarting". Those are choices to be made... this tree gives you all the choices and all one (min-maxer) has to do is filter out "the worst options".

I am actually rather aware of what he is saying. My time in the PTS does not agree with it. It certainly is less "minmax" than the linear trees you guys are pushing for, or the one Solamha built.

[Deathlike]

Bishop Steiner, on 30 April 2017 - 01:23 PM, said:

I am actually rather aware of what he is saying. My time in the PTS does not agree with it. It certainly is less "minmax" than the linear trees you guys are pushing for, or the one Solamha built.

I think you are doing inadvertent min-maxing and not realizing it.

There are choices and there are "choices". That's the best way to say it.

[Khobai]

Quote

so maybe I'm just too dense to see the issue?

obviously. occam's razor.

if a bunch of other people say theres an issue, and you dont see it, there most likely is an issue and youre just too dense to see it.


but the problem isnt so much walling. the problem is lack of equality in skill nodes. If all skill nodes cost 1 skill point then all skill nodes need to be equally worth 1 skill point. lack of equality in skill nodes is the underlying cause of walling.

the problem is more the fact we have weakass skill nodes like -1% crit chance (receiving), +5% hill climbing or +2% sensor range that just arnt equal in value to other skill nodes like +2.5% structure or +1.5% max speed, etc... if every skill node was equally worth considering spending a skill point on, it would eliminate a lot of issues. So instead of +5% hillclimbing, buff it to +10%-15% hillclimbing so having to spend a skill point on it is not nearly as infuriating.

The other problem was having to pick weapon nodes for weapons you didnt even have on your mech in order to get the generic weapon skills, but PGI already said theyre fixing the weapon tree so you dont have to get laser skill nodes even if you dont have lasers.

Edited by Khobai, 30 April 2017 - 01:51 PM.

[Chrome Magnus]

Khobai, on 30 April 2017 - 01:39 PM, said:

if a bunch of other people say theres an issue, and you dont see it, there most likely is an issue and youre just too dense to see it.

Just because a group of people say there's an issue doesn't mean there is one. We only need look at the history of mobs to see that. Plus your statement doesn't account for the other mob that doesn't see the issue or at least don't see it as being nearly as big as some folks make it out to be.

[FupDup]

Khobai, on 30 April 2017 - 01:39 PM, said:

obviously. occam's razor.

if a bunch of other people say theres an issue, and you dont see it, there most likely is an issue and youre just too dense to see it.


but the problem isnt so much walling. the problem is lack of equality in skill nodes. If all skill nodes cost 1 skill point then all skill nodes need to be equally worth 1 skill point. lack of equality in skill nodes is the underlying cause of walling.

the problem is more the fact we have weakass skill nodes like -1% crit chance (receiving), +5% hill climbing or +2% sensor range that just arnt equal in value to other skill nodes like +2.5% structure or +1.5% max speed, etc... if every skill node was equally worth considering spending a skill point on, it would eliminate a lot of issues. So instead of +5% hillclimbing, buff it to +10%-15% hillclimbing so having to spend a skill point on it is not nearly as infuriating.

The other problem was having to pick weapon nodes for weapons you didnt even have on your mech in order to get the generic weapon skills, but PGI already said theyre fixing the weapon tree so you dont have to get laser skill nodes even if you dont have lasers.

Making all nodes equal in value would help the situation a lot, but there are still some skills that you aren't going to need on your particular mech/build.

For example, AMS Overload will never be useful on a mech that does not equip AMS no matter how much you buff that skill.

The only time "gate" skills are forgivable is when they benefit all builds and all mechs, like +armor or +accel.

Edited by FupDup, 30 April 2017 - 01:58 PM.

[Khobai]

Chrome Magnus, on 30 April 2017 - 01:57 PM, said:

Just because a group of people say there's an issue doesn't mean there is one. We only need look at the history of mobs to see that. Plus your statement doesn't account for the other mob that doesn't see the issue or at least don't see it as being nearly as big as some folks make it out to be.

uh what? people dont mob up unless theres a serious issue to get them riled up in the first place. like a frankenstein monster running around. #sharpenyourpitchforks

mobs dont just form randomly and persecute others for no reason. there is always a reason. even if that reason is manipulated or exaggerated.

and there have always been ignorant people in the world who are too dumb to recognize an issue, or refuse to believe something is an issue when it really is, or people who simply cannot accept that other viewpoints exist besides their own.

Quote

Making all nodes equal in value would help the situation a lot, but there are still some skills that you aren't going to need on your particular mech/build.

I agree. but the whole point of making all nodes equal is that every node is worth 1 skill point to some particular build. in an ideal skill tree there are no useless nodes.

Right now we have skill nodes that arnt worth 1 skill point to ANY build.

Quote

For example, AMS Overload will never be useful on a mech that does not equip AMS no matter how much you buff that skill.

AMS overloard isnt worth 1 point even if you equip AMS. Its probably only worth it for a triple AMS Kitfox.

Which is why we should have unique skill trees for each mech like PGI originally promised at mechcon.

Having unique skill trees for every mech made so much more sense. Because then you could actually get rid of quirks by just migrating that mechs quirks into its own unique skill tree. And you could fine tune every mech in the game based on its power level and intended role. Even IS and Clan mechs could have different skill trees to help balance IS vs Clan.

Instead theyve removed quirks that made mechs different and made all mechs more similar by having them share a generic skill tree. What PGI has done is the complete opposite of what they needed to do. Or rather what they promised to do.

So yeah in order to save this hexmaze skill system, I really think PGI needs to go back to having unique skill trees for each mech. Give us a generic skill tree in the meantime to start with, but ultimately work towards every mech having its own unique skill tree in the future.

Edited by Khobai, 30 April 2017 - 02:46 PM.

[MischiefSC]

Bishop Steiner, on 30 April 2017 - 01:23 PM, said:

I am actually rather aware of what he is saying. My time in the PTS does not agree with it. It certainly is less "minmax" than the linear trees you guys are pushing for, or the one Solamha built.

Except I don't want min/maxing.

I want real value in tradeoffs. Like 0 hill climb unless I buy hill climb. 7% hill climb being worth giving up 20% arm pitch or 1%speed or 0.5% heat cap or 1.5% weapon range or whatever the medium of exchange is.

I want to actually have to make a conscious decision on more than about 10sp and I want the spuds I play against who didn't make optimal choices in the skill tree to at least have an approximate value to those who did.

[Khobai]

Yep. min/maxing is a product of poorly balanced choices.

min/maxing is taking the most efficient path that yields the biggest benefits for the least expense in skill points

but if all paths give the same benefits in the form of balanced tradeoffs, min/maxing is impossible


but the more complex a skill system is, the harder it is to prevent min/maxing. Which is why PGI shouldve just gone with something simpler like most other game developers have learned to do. Simpler means you can control the choices the player can make more easily.

thats why the module system was actually pretty good. it was simple. and pgi couldve just chosen to expand the module system instead. And it probably wouldve been way better than whats on PTR now.

The hexmaze skill system is a total mess. The only way I can see it being saved and actually improving the game is if they get rid of the idea of one generic skill tree for every mech and give each mech its own unique skill tree. Taking away quirks and making all mechs share the same skill tree does not help diversify mechs and mech diversity is what we need.

Edited by Khobai, 30 April 2017 - 02:48 PM.