Power Source and Power Level Progression

[Jajo]

I'd like to point out a potential issue with power level with respect to power source progression.

 

One of the design goals of multiclassing is, to have multiclass characters at 75%-85% class efficacy of pure characters at the same level.

 

In today's presentation Sawyer has shown the current tables and values, which give pure characters 3 power source points per level and multiclass characters 4 points, that are split between the two classes (3 in currently leveled up class and 1 in the other class). This gives a multiclass character an average power source point gain per level of 2 (i.e.: if class levels are split evenly between the classes, each class would receive an average of 2 points per level), which results in a multiclass character having an average power source point progression at 67% for each class, compared to a pure character.

 

Assuming, that power level progression is linear (i.e.: character at power level 4 is at 80% efficacy of a character at power level 5), which seems to be an assumption also taken by designers (I'm basing this on the shown power level vs level table), this would mean, that given a linear relation between power level and power source would mean, that a multiclass character would be approximately at 66% efficacy.

 

Of course, relation between power level and power source is not linear, as shown in the following table:

Points	Power Level
0-2	0
3	1
8	2
14	3
20	4
26	5
32	6

BUT! As it looks like, nonlinearity exists only for the first 3 power levels (~ 3 character levels) and is then linear up to at least power level 6 (~ character level 11). If it remains linear the character efficacy will inevitably limit towards 67%, which is well below the intended 'sweet spot' range.

 

 

For curiosity's sake, I've made comparison of pure (PC) vs average multiclass character (MC) power levels (PL) and their ratio with respect to character level (CL):

CL	PC PL	MC PL	MC/PC
1	1	1	100%
2	1	1	100%
3	2	1	50%
4	2	2	100%
5	3	2	67%
6	3	2	67%
7	4	3	75%
8	4	3	75%
9	5	3	60%
10	5	4	80%
11	6	4	67%
-linear assumption from here on-
12	6	4	67%
13	7	5	71%
14	7	5	71%
15	8	5	63%
16	8	6	75%
17	9	6	67%
18	9	6	67%

Since I do not have the tables, I've assumed linear progression for pure character power level with respect to character level progression, as was indicated by Sawyer in the presentational video.

My interpretation is, that the trend is already showing. Multiclassed character would also spend only 6 out of 17 levels not being worse than the intended efficacy (level 1 being exempt, because all characters are equal at level 1).

 

TL;DR:

Power level with respect to character level or power level with respect to power source progressions should most probably be tweaked to be more non-linear.

[Ninjamestari]

I don't like this system. It's needlessly complicated and lacks eloquence. I feel like the boys at obsidian need a couple nights worth of good sleep; I only design stuff this complicated when I'm on psychosis induced by sleep deprivation and too much caffeine.

[Micamo]

Indeed: Change the system so that you get 2 points in that class's power source when you level a class, and 1 point in all other sources, then change the power level table to look like this:

0-1 | 0
2   | 1
6   | 2
10  | 3
14  | 4
18  | 5
22  | 6
26  | 7
30  | 8
34  | 9

Then you hit 75% efficacy for an even level split. (7th-level powers for both)

Edited February 3, 2017 by Micamo

[virgi26]

i don't think your math is correct. Even if the power level progression is linear, overall effectiveness is not, since not only power of spells increases but also amount of spells. So I would say (with doing all the maths in my mind ) that 5lvl wizard will be 40% effective as 10lvl wizard. And if you add another 5lvl class which is also 40% as effective you would get 80% effectiveness.

But... i'm not quite sure what does he mean by effectiveness=)

[evilcat]

I think you reach too far ahead. Unless you have some extra insight we cant really tell how things will look in game.

Wait for beta and then complain.

The general rule, that we have some points for class, but less for class we do not advance in sounds ok for me.

[Jajo]

i don't think your math is correct. Even if the power level progression is linear, overall effectiveness is not, since not only power of spells increases but also amount of spells. So I would say (with doing all the maths in my mind ) that 5lvl wizard will be 40% effective as 10lvl wizard. And if you add another 5lvl class which is also 40% as effective you would get 80% effectiveness.

But... i'm not quite sure what does he mean by effectiveness=)

My math is correct and gives the correct results if and only if my assumptions are correct. That is, that:

• power level vs power source progression remains linear from power level 6 onward (the trend from 3->4->5->6 indicates this) and
• ratio between different power levels is a good (or even precise) indication of efficacy.

The second assumption here being the greater uncertainty, which is what You are rightfully aiming at. However I will not believe any number that is just plucked from thin air, because it is just too convenient to come up with something, that fits the model producing a desired result.

There are also two other crucial things wrong with your numbers.

First one being, that the point was for each individual class of a multiclass character to perform at ~80% effectiveness, not 40%.

The second one is, that effectiveness is almost certainly not a simple sum. The inspiration for PoE have been D&D games and much of the mechanics is based on D&D. In D&D encounter levels were in a quadratic relation with the number of opponents, which means, that a level 5 NPC would have been at 1/32 of effectiveness of a level 10 NPC. This is, of course, another assumption, which I did not make in my initial calculations. The efficacy numbers would have been even worse otherwise.

 

 

I think you reach too far ahead. Unless you have some extra insight we cant really tell how things will look in game.

Wait for beta and then complain.

The general rule, that we have some points for class, but less for class we do not advance in sounds ok for me.

Oh, you mistake my purpose with this post. I'm not complaining at all. As I've said, I'm just pointing out a potential problem with the model. And of course I don't have extra insight, nor do I know how things will look in the game, but making two not really intrepid assumptions and the numbers we do know, I am just hoping to constructively comment on the presented model.

Waiting for beta will not help anyone and would not accomplish anything in this case. It is harder to change mechanics by then and I am almost certain, that we will not get an opportunity to play level 16-18 characters in the beta. As I've said, the potential problems described in my original post would be most prominent at higher levels.

[Eurhetemec]

 

i don't think your math is correct. Even if the power level progression is linear, overall effectiveness is not, since not only power of spells increases but also amount of spells. So I would say (with doing all the maths in my mind ) that 5lvl wizard will be 40% effective as 10lvl wizard. And if you add another 5lvl class which is also 40% as effective you would get 80% effectiveness.

But... i'm not quite sure what does he mean by effectiveness=)

My math is correct and gives the correct results if and only if my assumptions are correct. That is, that:

• power level vs power source progression remains linear from power level 6 onward (the trend from 3->4->5->6 indicates this) and
• ratio between different power levels is a good (or even precise) indication of efficacy.

The second assumption here being the greater uncertainty, which is what You are rightfully aiming at. However I will not believe any number that is just plucked from thin air, because it is just too convenient to come up with something, that fits the model producing a desired result.

There are also two other crucial things wrong with your numbers.

First one being, that the point was for each individual class of a multiclass character to perform at ~80% effectiveness, not 40%.

The second one is, that effectiveness is almost certainly not a simple sum. The inspiration for PoE have been D&D games and much of the mechanics is based on D&D. In D&D encounter levels were in a quadratic relation with the number of opponents, which means, that a level 5 NPC would have been at 1/32 of effectiveness of a level 10 NPC. This is, of course, another assumption, which I did not make in my initial calculations. The efficacy numbers would have been even worse otherwise.

 

 

I think you reach too far ahead. Unless you have some extra insight we cant really tell how things will look in game.

Wait for beta and then complain.

The general rule, that we have some points for class, but less for class we do not advance in sounds ok for me.

Oh, you mistake my purpose with this post. I'm not complaining at all. As I've said, I'm just pointing out a potential problem with the model. And of course I don't have extra insight, nor do I know how things will look in the game, but making two not really intrepid assumptions and the numbers we do know, I am just hoping to constructively comment on the presented model.

Waiting for beta will not help anyone and would not accomplish anything in this case. It is harder to change mechanics by then and I am almost certain, that we will not get an opportunity to play level 16-18 characters in the beta. As I've said, the potential problems described in my original post would be most prominent at higher levels.

 

 

If you're seriously concerned about this, you should ask Sawyer about it on his Twitter or Tumblr. He answers questions of a similar nature pretty regularly.

 

But I would suggest you are being silly in three ways.

 

Firstly, you are making too many assumptions about both the precise intent of MC'ing and the precise indication of efficacy. You kind of admit that latter but I'm not sure you're really thinking about how big a hole it is in your knowledge. They absolutely are "intrepid" assumptions, contrary to your claim that they are "not really intrepid".

 

Secondly, you're assuming that this is something the devs, particularly Sawyer, are completely unaware of this, which seems wildly, wildly unlikely, and could easily be found out by... asking Sawyer. It's not like he's in some sort of fortress. Man is answering questions every day.

 

Thirdly, your claim that "by beta it will be too late and not help anyone" is a totally unsupported assertion that has significant evidence against it, given the massive changes Pillars 1 went through, and the massive changes virtually every AAA CRPG goes through.

 

Still, why not actually go to Sawyer with this, if you're serious?

Edited February 4, 2017 by Eurhetemec

[hilfazer]

Talents will be at 67% but there's also accuracy, health and defences and they depend on character level.

75%-85% refers to overall performance.

[Jajo]

If you're seriously concerned about this, you should ask Sawyer about it on his Twitter or Tumblr. He answers questions of a similar nature pretty regularly.

 

But I would suggest you are being silly in three ways.

 

Firstly, you are making too many assumptions about both the precise intent of MC'ing and the precise indication of efficacy. You kind of admit that latter but I'm not sure you're really thinking about how big a hole it is in your knowledge. They absolutely are "intrepid" assumptions, contrary to your claim that they are "not really intrepid".

 

Secondly, you're assuming that this is something the devs, particularly Sawyer, are completely unaware of this, which seems wildly, wildly unlikely, and could easily be found out by... asking Sawyer. It's not like he's in some sort of fortress. Man is answering questions every day.

 

Thirdly, your claim that "by beta it will be too late and not help anyone" is a totally unsupported assertion that has significant evidence against it, given the massive changes Pillars 1 went through, and the massive changes virtually every AAA CRPG goes through.

 

Still, why not actually go to Sawyer with this, if you're serious?

 

 

I do not have a twitter account; I do not have a tumblr account, I do not have a reddit account. I do have a something awful account, but this is a topic about pillars of eternity 2. How exactly is the official pillars of eternity 2 forum the wrong venue for asking about this? It is quite clear, that obsidian employees do follow and even answer on these forums.

 

1. I am making exactly two assumptions - I have explicitly stated this in both my posts, so I guess I am kind of admitting it. Explicitly. The first assumption is exactly what Sawyer said in the video: power level goes up every other level, like it did in D&D. The second assumption is more questionable, but as I have explained, it is by far the more conservative model among the known alternatives (being linear instead of quadratic).

 

2. This thread is me asking about it. What is your point? Should people assume, that developers have thought about absolutely everything?

 

3. I have never claimed, that "by beta it will be too late and not help anyone". I have claimed, that by beta it will be harder to change things, not impossible. This is not an assumption, this is a fact.

My other claim was, that in beta we will not be given the end portion of the game to test. Too many potential spoilers. We sure didn't get to play it in PoE1 beta, TToN beta, [insert your game name here] beta...

Edited February 4, 2017 by Jajo

[Jajo]

Talents will be at 67% but there's also accuracy, health and defences and they depend on character level.

75%-85% refers to overall performance.

A very good point.

[MaxQuest]

TL.DR.:Power level with respect to character level or power level with respect to power source progressions should most probably be tweaked to be more non-linear.

So the power_source points to power_level conversion was changed....

Initially Josh posted this response.

And we could say that the formula was: power_level = Math.floor(power_source / 3)

 

With the new conversion table:

Points	Power Level
0-2	0
3	1
8	2
14	3
20	4
26	5
32	6

It looks like the formula now is: power_level = power_points < 3 ? 0 : Math.floor[(power_points + 4) / 6]

This changes the final power_level of a level 18 character in the following manner:

lvl (18 + 0) = power_source points: 54 + 18 | power_level = 9
lvl (17 + 1) = power_source points: 52 + 20 | power_level = 9 + 4
lvl (16 + 2) = power_source points: 50 + 22 | power_level = 9 + 4
lvl (15 + 3) = power_source points: 48 + 24 | power_level = 8 + 4 
lvl (14 + 4) = power_source points: 46 + 26 | power_level = 8 + 5
lvl (13 + 5) = power_source points: 44 + 28 | power_level = 8 + 5
lvl (12 + 6) = power_source points: 42 + 30 | power_level = 7 + 5
lvl (11 + 7) = power_source points: 40 + 32 | power_level = 7 + 6
lvl (10 +  = power_source points: 38 + 34 | power_level = 7 + 6
lvl ( 9 + 9) = power_source points: 36 + 36 | power_level = 6 + 6

Also, we can hardly estimate the relative power of a single vs multi-class character, until we know at least some basic coefficients for spells and abilities. For instance what is the formula for fireball's hit damage??

dmg = base_value * (1 + power_source_points/54 + might_coefficient)?

 

 

Edit: I mean that would make sense as a lvl 18 pure wizard would have by 20% more powerful spells than a 9/9 wizard/fighter of the same might, plus Josh mentioned in that video that damage of class specific abilities would scale with power source.

On the other hand, I remember that when explaining empower mechanic their idea was that empower would add 3 power levels to the next spell cast. Would that mean that there will be a conversion back to power source points? And how will they do it, if the function is not bijective?

Edited February 4, 2017 by MaxQuest