D&D (2024) Class spell lists and pact magic are back!

[mamba]

I saw above. It is missing the entire point. Again, it is QUITE clear that these surveys have been giving them good information. You are just insisting they aren't because you don't like it.

as I said before, you do not even know what the implications of the issues I am raising are, so you saying it clearly gives them good information is meaningless. I am not going into the part that you have never managed to demonstrate the 'clearly' part of your post, been there, done that, it's a dead end or an infinite loop, depending on how long I care to keep replying.

It gives them relatively good information, when it comes to avoiding things that are very unpopular and identifying things that are very popular. The issue is in the middle, where things that would be sufficiently popular to at least iterate if not squeeze into 'keep' outright are not always reliably identified.

If all you care about is avoiding the unpopular options and not so much about improving the game in general, then that is something you can live with. I think that WotC's main reason for the survey is exactly that, they care a lot more about identifying and avoiding very unpopular proposals and a lot less about some improvements slipping through their fingers.

That is my gripe with it, esp. since I believe this is why some of the changes I would have liked did not make it.

 

[Maxperson]

You are completely missing the point. By your logic every survey ever created is useless because you can't precisely get exactly what people mean. There is no direct line, there cannot be. You say it doesn't matter, but it is the entire point.

No. That's entirely incorrect. Most surveys don't make an attempt to figure out an ultimate percentage based on vague questions. They ask specific questions and report what percentage answered "important," which answered "very important," etc. They make no attempt to be as exact as WotC claims to be.

It was an example. They use it for non-number surveys too. It isn't a phenomena that only occurs when you have numbers. The entire theory rests on the idea that the extremes on both ends cancel out, and the closer to the middle you are, the more accurate the answer is.

80% isn't the middle. WotC isn't trying to get a middle percentage. Nor are they cancelling out the extremes.

No. You are completely making up this idea that 51% equals satisfied. If you were "satisfied" with your work would you redo it? Would you refer to it as something you believe you can still salvage? Heck, remember, a score of 51% doesn't even get that. It gets tossed in the trash.

If it gets tossed in the trash, their 70% or 80% is entirely inaccurate for multiple reasons, not just the one I used earlier.

This is like saying that getting a 51% on test, since you got more answers correct than incorrect, means you pass. It doesn't work that way. And it may sound weird to you, but that doesn't make it false. 51% on the One DnD survey means enough people said "dissatisfied" and "Very Dissatisfied" that it is getting thrown out. You can say "that's stupid, 51% should be salvageable and 49% should be thrown out" but that isn't what they are doing, and they are doing it the way they are for a plethora of very good reasons. Including

1) Getting a super majority
2) Better public perception
3) A goal of high quality.

Without knowing exact percentages, they cannot tell if something hits 70%, 80% or 5%. They can't tell if it got a super majority or not.

Then why is there a very dissatisfaction level Max? It would be like having a Living, Dead, and Super duper dead. There is no point in a category that is just the exact same as the first category, only with an exclamation point.

Very dissatisfied would be 0-25%. Dissatisfied would be 26-50%. Satisfied would be 51-75%. And very satisfied would be 76-100%. There are 4 categories to divide up 100%.

 

[mamba]

Very dissatisfied would be 0-25%. Dissatisfied would be 26-50%. Satisfied would be 51-75%. And very satisfied would be 76-100%. There are 4 categories to divide up 100%.

I’d go as far as saying very satisfied gets rated as 100%, it certainly cannot be 76%….

The rest is somewhat more fluid, and that too influences the average without any input by the person filling out the survey…

 

[Amrûnril]

Regarding the discussion of percentages, the developers are describing their results as approval ratings. In conventional polling terminology, that’s a measurement of how many people like something, not of how much the average person likes something (if a public figure has a 45% approval rating, that means 45% of respondents told the pollster they approve of the person, not that the average respondent approves of the person 45%). If the developers say something had a 60% approval rating, the default interpretation of that statement (absent specific evidence to the contrary) would be that 60% of respondents gave it either an “Approve” or “Strongly Approve” rating.

If the developers want a first impression of how popular something is, this seems like a good number to look at. If they want to know how strongly people feel, that would be obscured in the topline data, but hopefully still available in a more detailed breakdown of the results. If they want to know whether people would like to see further iteration on a design, then they need to collect data that actually corresponds to that question.

 

[Maxperson]

I’d go as far as saying very satisfied gets rated as 100%, it certainly cannot be 76%….

Sure it can. It's absolutely guaranteed not to be anywhere close to 100% as a group. An 80%(A grade of B) would definitely be very satisfied. Getting a B is a very good grade.

The rest is somewhat more fluid, and that too influences the average without any input by the person filling out the survey…

If the groupings aren't equal measures, it becomes even more lopsided. It's simply not possible for WotC to accurately gauge what 80% and 70% look like based on the questions that they give.

 

[Neonchameleon]

If the developers want a first impression of how popular something is, this seems like a good number to look at. If they want to know how strongly people feel, that would be obscured in the topline data, but hopefully still available in a more detailed breakdown of the results. If they want to know whether people would like to see further iteration on a design, then they need to collect data that actually corresponds to that question.

Which they get by reading the comments. I've long described 5e as the Happy Meal of D&D, designed so almost everyone will like it and almost no one will hate it. And Crawford's continuing that tradition more than competently while his work as an original designer/design lead in adding new elements over this playtest has been ... lacklustre, with the places where he's presented a new vision having been frequently and normally deservedly shot down either for bad planning or for bad execution when what was presented should have been a very early draft (even an outright first workable draft in the case of the brawler). I might prefer something spicier than the fatty salty mixture of the happy meal - but it's undeniably popular and a good compromise.

 

[mamba]

Regarding the discussion of percentages, the developers are describing their results as approval ratings. In conventional polling terminology, that’s a measurement of how many people like something, not of how much the average person likes something (if a public figure has a 45% approval rating, that means 45% of respondents told the pollster they approve of the person, not that the average respondent approves of the person 45%). If the developers say something had a 60% approval rating, the default interpretation of that statement (absent specific evidence to the contrary) would be that 60% of respondents gave it either an “Approve” or “Strongly Approve” rating.

so X% were satisfied or very satisfied here, could be. Not that it makes this any less wrong

If the developers want a first impression of how popular something is, this seems like a good number to look at.

possibly

If they want to know whether people would like to see further iteration on a design, then they need to collect data that actually corresponds to that question.

that is the decision they make based on how high that percentage is… I agree that it would be better to ask this outright

 

[mamba]

Sure it can. It's absolutely guaranteed not to be anywhere close to 100% as a group. An 80%(A grade of B) would definitely be very satisfied. Getting a B is a very good grade.

if it were 80%, nothing could exceed an 80% rating.

I am not saying everyone is 100% satisfied, but that if you assign a percentage to each of the four options, and then add those up to get the average %, then ‘very satisfied’ being assigned 100% is likely to me.

It will still represent a whole band of individual satisfaction %, but so do the other three, and they too would then be represented by a single number. WotC has no way of determining that your satisfied was 71% and mine was 67%, they have to go with one number that works with their math / thresholds

 

[Maxperson]

if it were 80%, nothing could exceed an 80% rating.

But it is 80% and 81% and 82% and 83% and...

That's the problem with their method. They are trying to achieve an 80% rating without having an inkling of what people actually rate the abilities. "Very satisfied" doesn't tell them anything about where in the real range a given answer is. If they are using them all at 100%, they are grossly overweighting the "very satisfied" answer. All of the other answers are similarly weighted by those of us voting, but unknown to WotC because they are asking the wrong question.

The question they should be asking is, "On a scale of 1 to 100, how much do you like X ability?" That would allow them to average the answers to accurately see how much those who are voting really like an ability.

I am not saying everyone is 100% satisfied, but that if you assign a percentage to each of the four options, and then add those up to get the average %, then ‘very satisfied’ being assigned 100% is likely to me.

I agree with that. I'm sure they do arbitrarily assign a percentage that is wrong. That's the problem. It's arbitrary and wrong. To be right they need to ask the better question I suggest above.

It will still represent a whole band of individual satisfaction %, but so do the other three, and they too would then be represented by a single number. WotC has no way of determining that your satisfied was 71% and mine was 67%, they have to go with one number that works with their math / thresholds

That's not true. They do have a way to determine if I am 71% and you are 67%. They just have to ask the right question instead of 4 wrong questions that are guaranteed to give them a wrong percentage score.

 

[mamba]

But it is 80% and 81% and 82% and 83% and...

yes, but if you want to create an average based on the survey, you kinda have to pick one number for each, which was my point

That's the problem with their method. They are trying to achieve an 80% rating without having an inkling of what people actually rate the abilities. "Very satisfied" doesn't tell them anything about where in the real range a given answer is. If they are using them all at 100%, they are grossly overweighting the "very satisfied" answer. All of the other answers are similarly weighted by those of us voting, but unknown to WotC because they are asking the wrong question.

that is one problem, the other is that what they really want to know is whether we like this idea better than the current one and want to see it improved (if it is not well enough received right away) or discarded, at least that is how they use the averages they derive

They are only overweighting the 100% if the other three percentages are picked 'wrong'. In any case, none of the four will be entirely accurate, no matter what value WotC assigns. With 40k votes this might not matter all that much unless they pick those percentages wrong however.

My concern is more that the percentages do not map very well to discard / improve / keep, yet this is what WotC decides based on them even though we cannot accurately communicate this.

That's not true. They do have a way to determine if I am 71% and you are 67%. They just have to ask the right question instead of 4 wrong questions that are guaranteed to give them a wrong percentage score.

I meant based on what they are asking today