5e Math, Take 2!

Enemies will stay usable the entire campaign, as saves and AC and attack bonuses no longer change.

All you have to worry about is HP and Damage.

You want ASIs for damage *and* accuracy. 5E's damage numbers are stupidly fucking low to the point where any +1 helps and AC isn't ignorable so the +1 to hit is also important.

>All you have to worry about is HP and Damage.
>All you have to worry about
>All

Oh, is that it?

As opposed to normally, where you also have to worry about AC and Attack roll scaling?

That makes things easier. I don't know how you could think otherwise.

Sure he would. But if your stat is already at 20, you've got your ASIs worked in. My point was that there's fewer parts scaling with level, and that even a level 1 enemy would have a decent chance to hit a level 20 enemy with such small scaling. They wouldn't hit very hard, but they would hit.

You'd likely also want to rework ASI progression if you're going to work several of them into up front attribute generation.

>
If you dropped the number scaling in your primary stat and proficiency, and started both of them at the max value, what would be the consequences?

I can think of two.

1) A lot of arguments over whether a paladin's primary stat is Strength or Charisma, and similar arguments for Bladelocks, Monks, Rangers, Eldritch Knights, Monks, Spellthiefs, MONKS, M O N K S, Barbarians, and FUCKING MONKS.

2) A really fucking boring character progression.

Oh, and

3) A really incongruous experience. So my snot-nosed barbarian kid fresh from the hill clans and just starting his path of adventure, is already as strong as Conan the Barbarian? My wizard who only yesterday graduated from the Unseen University already has intellect on-par with that of Merlin?

Yes, they're just lacking in experience, hence having less abilities, weaker spells, and lower proficiency bonuses

>using characters that are in no way related to dnd as a comparison to the mechanical side of dnd characters

Why don't you just change Proficiency to "add your level to skill checks you're proficient in"?

And adjust DCs as necessary.

But my 1st-level barbarian might still beat Conan around, say, The Hour of the Dragon (he probably wasn't 20th by then), in an arm-wrestling match (I presume the guy still wants the Barbarian to keep his capstone ability). And my snot-nosed wizard from the Unseen University is just as good at trivia night as Merlin.

Y'all need to stop looking at the raw numbers and start remembering what those numbers are supposed to represent.

(The guy I was responding to, you'll note, stated that he wanted to drop the number scaling from both primary stat AND proficiency, starting both at their max values. Meaning your primary stat starts at 20 and your proficiency starts at +6).

>Arguments over what starts at 20.
Why? You give them higher powered arrays to pick from, and let them put things where they want. What's the argument?

>Boring character progression
The interesting part is the new class features, not number scaling.

Did you read the OP, at all? How would that address "all kinds of group checks are shit"?

>The interesting part is the new class features, not number scaling.

How do you address current features and ASIs (Fighter and Rogue especially) where number scaling IS the new class feature.

Did you catch the reason why it was suggested?

A group check with +0/+1/+1/+2 is as effective as a roll with +11 for like all DCs up to DC17.

+5 attribute and +6 proficiency = +11

Feats?

For arguments sake, what if it was
Ability @20(+5)
Prof @+6
And then tack on an extra +(1/3.5)*lv to add in number scaling, without the group roll being way better than skill.

>Did you read the OP, at all? How would that address "all kinds of group checks are shit"?
Not sure what the fuck you're talking about. I just thought this would be an instant fix for bounded accuracy.

Personally I don't think it's even a problem, but OP looks like he's investing way too much time into this.

Oh, shut the fuck up, you know the point I'm trying to make and trying to distract from it won't change that it's a fundamentally good point that will have to be addressed if you want to implement the proposed change into your games.

>Why don't you just change Proficiency to "add your level to skill checks you're proficient in"?

Because that leads to 3.5 and Pathfinder. Well, slightly less since you cap out at 20 rather than 23 in a skill, but fundamentally the same thing.

It creates impossibly swingy problems. Like, say, a 10th-level seafairing adventure involving pirates attacking a ship.

- Oh no! A storm strikes. Make Acrobatics checks to stand on the rolling deck while fighting pirates.
- - A character with 20 Dexterity trained in Acrobatics has a modifier of +15. He can pass DC 26 checks 50% of the time.
- - A character with 20 Dexterity untrained in Acrobatics has a modifier of +5. He can pass DC 16 checks 50% of the time. He can't ever hit a DC 26.
- - A character 10 Dexterity untrained in this skill has a modifier of +0. He can pass DC 11 checks 50% of the time, DC 16 checks 25% of the time, and can't ever pass DC 26 checks.

So what do you set the DC at? A DC 16 check literally can't be failed by someone trained in Acrobatics at this level under your system, but it's a coin-flip for someone untrained but still dexterous, and a dangerous gamble for someone without any notable dexterity. It's too easy for one player and too dangerous for the others.

But if you set the DC at something high enough to actually challenge the +15 Acrobat player, you make things far too difficult for everyone else.

(cont'd)

The op says the "problem" with bounded accuracy is not the number scaling, so much as it's the shitty effect of group roll mechanics.

So shitty, that if you were to try to fix it by +#s instead of changing the mechanics for group rolls, it would mean starting with +11 in the stuff you're good at.

There are charts and numbers in the first 4 posts and everything.

What the fuck are you even on about? How are group checks in any way fucked?

THAT'S THE FUCKING POINT.

The main reason for bonded accuracy is that limiting how high skills checks can possibly get makes designing adventures much easier. This is particularly true for Wizard of the Coast themselves, who publish adventures like Curse of Strahd or Storm Kings' Thunder but don't have the benefit of actually knowing the players who will be playing it.

By capping the skill modifier for most characters at +11 and by having that modifier progress at a steady rate that's easy to anticipate, it becomes much easier for Wizards of the Coast to design an adventure for characters of a given level and have any given character be capable of actually completing the adventure, rather than being stymied by a lock that's too difficult to pick; or by making them roll for skill checks that are laughably easy. Instead, they can present DCs that are challenging for trained characters and difficult but not impossible for untrained ones.

Expertise, meanwhile, allows characters in classes with a focus on skills to notably shine by being much better at a given task than the rules normally assume. While everyone else is fighting pirates on a rolling deck and desperately to stay on their feet, the character with Expertise in Acrobatics is zipping along without a care. Expertise is a case of deliberately breaking bounded accuracy, basically.

> But if you set the DC at something high enough to actually challenge the +15 Acrobat player, you make things far too difficult for everyone else.

It seems like the primary complaint isn't finding a DC that challenges characters of different aptitudes, but that trained characters with maxed stats could ever fail "simple" categories of task.

In the example given DC 16 is appropriate. The graceful acrobat basically ignores the storm, because that's what he's specifically good at. The nimble guy is pretty good, but he's still got a chance at failure. The averagish guy is going to fail a lot.. because he sucks.

This is the point that 3.5/pathfinder missed. Once you reach a certain level of aptitude in your specialty, obstacles from that category SHOULD cease to be a challenge.

Are you blind, illiterate, or just retarded?

Look at the pretty pictures. And the relevant snippets.

>>For all DCs between DC5 and DC17 (except DC 10 and DC 11), the weaker (at skill bonuses) party rolling is like having a +11, and the stronger is like having a +12.

>>I. Unless it's for DC18+ checks, there's no reason to take Knowledge skills. Investigation and Social Skills are the same, unless there are escalating penalties for each failure.
>>II. You should never use "the whole group needs to individually succeed" checks, it's basically a guaranteed failure.

It's spelled out in math, in black and white.

It's a stupid fucking point from a game design perspective, as outlined here

No it's not, there is a very good reason nearly every single RPG does not do anything like that stupid fucking bullshit.

Most RPGs don't have a level progression. They just have a cap on attributes and skills, and let you buy up to it.

And often, attribute+skill = dice size, if it's a one die system.

>but that trained characters with maxed stats could ever fail "simple" categories of task.

Everyone has an off day. No NBA player has a perfect free-throw record, for example, as was pointed out upthread. This is particularly the case when that NBA player is trying to perform a free-throw - a simple task in and of itself - when a goblin is gnawing on his legs, and the thing he's throwing isn't a basketball, but a bomb, and his target isn't a hoop, it's a dragon's mouth.

>This is the point that 3.5/pathfinder missed.

It didn't miss it at all. 3.PF has a tendency to meticulously detail what the DC is for any given task and set it in stone for all time. The problem comes form the "swingyness" that results.

Like I said here, , it makes adventure design a nightmare, particularly if you don't actually know the specific people who will be in your player's party, because you don't have players, because you're Wizards of the Coast and you instead have millions of consumers, each of whom have between them millions of characters.

In 3.5, Wizards of the Coast would know that a 10th level Rogue could be reasonably expected to have a +18 to Open Lock. What Wizards didn't know was a) whether or not a given party has a rogue, and b) whether or not that rogue has taken Open Lock.

A coin-flip challenge for the rogue would be a DC 29. For anyone untrained in Open Lock, however, this would be impossible to hit. So what were they supposed to do? Design all their adventures so that only parties with rogues that had Open Lock maxed out could progress? Or design adventures where the DC to open a lock was never beyond the reach of a party without Open Lock? But if they were doing that, why bother having locks at all when a rogue who IS trained in Open Lock will just walk right through them?

>there is a very good reason nearly every single RPG does not do anything like that stupid fucking bullshit.

Regale me as to the reason why most RPGs would prefer, with their published adventures, to include parts where it is possible that a party will not physically be capable of progressing, due to the players not having foreknowledge of what skills would be requires to overcome a given challenge and so therefore are being stymied by a obstacle designed to challenge someone who has been dedicated to advancing that particular skill since character creation but which is as a consequence impossible for characters who have NOT been so dedicated to that skill to pass?

Because other RPGs, including other editions of D&D, are built with the assumption that you'll cover your bases or you'll contract out NPC help?

Thread is tl;dr

What's wrong with using saga edition math of:
half level + stat + 5 if proficient + 5 if expert?

Because certain people get triggered at numbers that are larger than what you can count on your hands and toes and tell you to go back to Pathfinder.

NBA players would all be pretty low level so that example is terrible.

Everyone who wondered how "group checks" should be handled can consider what I've been using the whole time as the DM:
>If the group gets into a situation where all of them need to make a check (such as sneaking past some enemies etc.), only the player who has the worst stats for the situation (in this case, the worst stealth) rolls. If they succeed, I as a DM believe that if the worst succeeded, everyone else did so too.

Group stealth checks (Type B checks) still shit that way, for one.

For two, those bonuses are too low on the non expert side. Change the +5 to a +6, or have attributes at +6, and proficiency at +5 and it works out alright.

No point putting ranks in stealth with those rules. Unless the whole party is built to be good at it, everyone is shit at it.

And is just based on the math from OP.

Half level may be too fast scaling for 5e though. If the bonuses are big enough, sticking with somewhere around 1/4 lv is probably fine.

Why not average the bonuses and roll a single check?

I suppose that's one way to do it.

Or have the guy with the best bonus roll, with some kind of modifier. Presumably everyone else is following his lead.

>Group stealth checks (Type B checks) still shit that way, for one.

But the way stealth checks work for 5e is that if 2 out of your group succeed, all of your group succeeds, no?

I guess "jumping a pit" would be a better example of Type B, but even an untrained, non-good stat person would have an about ~ 25% chance for those under Saga math (assuming level appropriate challenge).

>specialist tears through challenges meant for non-specialists with no effort
>this is somehow a problem
Why.

I may be misremembering stealth, don't have my book open.

Jumping a pit would probably be a better Type B example though.

Because sometimes that shit happens in the stories I read and I want my game to look like the stories I read.

Do you have any further questions?

Do you enjoy having the wrong sort of fun?

Gotta keep those pesky players down, where they belong.

>So what were they supposed to do?
Design multiple avenues for progression. You have a skill-monkey with Lock Picking? Great, you open the door. You don't? Well, you can either track down the key (maybe involving some knowledge checks or social checks) or you can take an alternate route that's more dangerous.

Or, since we're talking about D20, a Wizard just casts Knock and chortles disdainfully.

>it makes adventure design a nightmare
No, not really, you build around the assumption that the players covered the Thief/Fighter/Wizard/Cleric bases in some form and let them get fucked/let the DM modify shit on the fly if they refuse to do so AND won't use hired hands for whatever they're missing. Designing for 3E is a shitshow that rarely works out well because many parts of 3E don't work as advertised, NOT because gork the barbarian can't nail a Knowledge check outside of his realm of expertise or pick a lock.

Better solution is to tailor the end-result to the number of successes. For example, those characters that pass are able to reach advantageous positions and/or have a surprise round while the characters that fail are caught out in the open.

Make simple tasks autosucceed for sufficiently skilled characters

Don't let ameteurs roll for suffiently high level challenges.

Knowledge skills aren't one and pass. They're group checks and there's potential for misinformation on low rolls.

Social checks are usually roll once. Investigation can usually be solved by spending extra time, with the downside of whatever that time costs you.

You don't need to change the system when you aren't using the system right in the first place. Rather, there's multiple valid ways to run it, and it's only wrong if you don't like the outputa method gives. So use a different method.

Jump this pit or you die is one of those "pick this lock to continue" kind of barriers. If it's necessary, then those who can't jump have to go around a slower way or spend resources to get across.

I'll leave this here again.

Wow this devolved into autistic screeching and name-calling really goddamn fast.

The point of the skill system as is is partly to have all PCs in the group the ability to contribute, even however little they can, It's a safety net for a fucked up roll really.

Making something like this is essentially telling the others to just sit it out, they're not needed, don't you want all of the party to function as a team?

5e doesn't need all of this maths at all, that's not even the point of the system.

It's also really easy to design around.

Those with acrobatics could walk a rope connecting the rooftops (possibly after someone jumped over with a rope).

Those with neither DEX or STR will have spells to help them out.

Those with neither DEX, STR or spells better be lucky as shit to survive as adventurers.

Wow, turns out the buttblasted autists are from the proponents of this method, even when faced with well reasoned and well constructed discussion questioning its worth and and the fact that Wizard's method works for what it is. Very good to know.

>Wizard's method works
>fact
Wew.

Why not use different dice e.g 2d10?
It would make it harder to roll high and make point boni much more important to reach high numbers

>Most of the thread was a clusterfuck of namecalling,
Welcome to neo-Veeky Forums, enjoy your eternal summer.

Even the examples given in the other thread didn't display quite the parity between the skilled and untrained, coupled with the well formed and articulated arguments in this thread.
Calm your autism.

And that point is shit. There's a reason only a tiny handful of games try anything like it.

>...can aggravate the problem you're trying to solve
Wow it's like they knew that the bonuses were too small and went ahead and did it anyways. Fuck WotC.

Becuase it results in math that is too swingy, becuase in order to challenge someone trained in a skill you have to set the DC so high that a person untrained in the skill is unlikely or outright incapable of hitting.

This is fine if you want each class to have exclusively its own niche with no overlap. It's problematic if everyone on the team needs to, say, hop on a bantha and ride across the desert, and only one of them has any proficiency in ride.

Ironically the Saga system fails at its intended purpose, because the entire reason why they came up with that formula and ditched the skill rank system was becuase they wanted to make sure that if a GM wanted his players to hop on banthas and ride across the desert, they would all be capable of doing it without issue.

Or in other words SAGA was trying to do exactly what 5e has managed: ensure that only the most difficult skill tasks are beyond the reach of most players.

>this is somehow a problem

It's not a problem. The thing is that 5e's system ALREADY DOES THAT, without making the skill check DC so high that it is utterly impossible for non-specialists to pass - merely unlikely.

>Do you have any further questions?

No, but a comment: this is why you don't work at Wizards of the Coast or some other gaming company: because you design for YOUR players, not ALL players.

I'm actually not aware of another system besides 3.PF/4e D&D and derivatives that makes the local equivalent of skill checks utterly impossible for a character unless it requires specialized knowledge. Like, in White Wolf's World of Darkness, for example, any player could pass any talent or skill check becuase the Difficulty is never higher than 10, their dice pool is never smaller than 1, and only 1 success is ever needed to succeed.

I have a WIP 5e hack that addresses this problem. There's a lot of shit in there but typical DCs and scaling proficiency bonuses go a long way to fixing the math.

homebrewery.naturalcrit.com/share/HymIj_ZO6g

TL;DR: Bounded Accuracy designs its bonuses in part because of the way Armor Class generally falls among creature types. When you structure DC's the same way the game tends to structure AC, shit works.

There are 18 skills in 5e D&D. The Fighter/Rogue/Cleric/Wizard combination have between them, including skills from background, 18 skills. This means that the only way to have every single base covered is if the players deliberately build their characters to have zero overlap. More realistically there will be some overlap, however, because most players design the character they want to run, not the character the party needs.

>NOT because gork the barbarian can't nail a Knowledge check outside of his realm of expertise or pick a lock.

Usually it's more of an issue that Zot the Wizard is going to have to be left behind to die to the goblins chasing you because he can't make an Athletics check to jump a gap.

Or the group needs to travel overland from point A to point B really fast, but the only way to do so is on horseback. Once again Zot screws everything up for the party because he sucks at ride, can't make the DCs, and so if the party is attacked between A and B by outriders then he's going to be a giant load unless he gets of his horse, but doing that leaves him essentially immobile when compared to everyone else.

Or, more relevantly for my most recent play experience, the Thief is the only one with Stealth so half the game involved the Thief being 60+ feet away from everyone else and everyone having to rely on the Thief's lackluster Perception check results and/or the fact that if the Thief is spotted, she's gotta endure an entire round of combat by herself before help can arrive.

Or so on.

>Wow it's like they knew that the bonuses were too small

They anticipated that some autists would be stuck in the 3.5 mentality that it should be impossible for rank amateurs to ever even once succeed, and that specialists would never, ever fail.

There is nothing wrong with the math, the problem is the people whining that Gork the Barbarian got a lucky roll and beat Zot at trivia night once.

In most of the groups I play with we use guidance cantrip and the help action to make all out of combat ability checks with proficiency, advantage and +1d4.

Furthermore, many spells exist to help like Pass w/o Trace and Enhance Ability. Not to mention magic items.

If anything, these magical solutions call for ability checks to be more difficult, not easier.

Shadowrun. Try making a check a specialist failed as an untrained fuck, I dare you.

The party of amateurs dramatically outdoes the proficient guy from DCs 8-21 until level 17. Are you seriously suggesting that you only allow rolls above dc7 if trained?

Your suggestion is bad.

Only helps at all with type C scenarios after around level 10, and doesn't help type B scenarios at all.

It was brought up right in the OP.

What does reposting it accomplish?

Last thread people claimed the problem didn't exist because no math was provided.

This thread there is mathematical proof of when there is an issue right in the OP.

On group checks. Both when each person is allowed to take a crack at something and one success carries the group(Type C), and when the worst roll holds the group back(Type B).

The OP suggested changing the mechanic for group checks, and showed mathematically that under the way most people do group checks, for the professional to be *as* good in type C checks as the party of amateurs, it needs to be for between DC18 and DC21+ checks (depending on the size of his bonus), or he needs a +11 or better.

And the autosucces rule only helps with DCStart with ability at 20 or at level 10, use the autosucces rules, and don't allow nonprofit isn't characters to roll on DC 16+ checks.
Only addresses Type C.

If you've got nothing but name-calling to contribute, you're wasting your time.

Does someone wanna run anydice and modify OPs programs so we can see the odds for these and see how "use different dice" plays out?

I suspect it wouldn't *solve* either problem, but it might work as part of a solution.

I can see at a glance that your fix doesn't fix it. Have you run the numbers in anydice?

There's no reason for Zot the wizard to even take knowledge skills, unless it's for DC21+ checks. Anything easier, and his drinking buddies will succeed far more often than he will anyways, until he hits level 17 (or gets a total +11 or higher through expertise).

And that +11 only matches their output for rolls DC17-.

>there is an issue

There isn't, though. The intent of the system is that for all but the most difficult of tasks, anyone has a chance of accomplishing it, but some people are more likely to accomplish it than others if they've dedicated time and effort to training in the skill.

You are complaining about the necessary consequence of this system, viz., that Grok the Barbarian can occasionally at low levels beat Zot the Wizard at trivia night (and like problems) provided the trivia questions are all DC 20 and below. The thing is that any attempt to "fix" this relative non-issue will necessarily detract from the original point of the 5e skill system, which, to repeat myself, is to ensure that everyone has a chance at all but the most difficult of tasks.

And you're also forgetting that it's not a real issue because it doesn't break the game. From a purely mechanical standpoint Zot is still more likely than Grok to answer any given trivia question correctly (increasingly so as their levels advance), but the game does not collapse in on itself if Grok occasionally beats Zot. It's not a bug, it's not a problem, it's just a thing that you do not personally like.

Oh, also, a sub-intent of the system is to make Wizards of the Coast's jobs easier by allowing them to know, +/- a few points, what the likely skill modifier for someone of a given level is going to be, and thus be able to design adventures based around that assumption.

>Anything easier, and his drinking buddies will succeed far more often than he will anyways

You're gonna have to walk me through how someone with a +5 modifier to a skill is less likely to hit a DC 15 than someone with a +0 modifier to the same skill, because I am just not seeing how that is mathematically possible.

Zot is much less likely to succeed at trivia than Grok barbarian, Steve fighter, Billy rogue, and Sally cleric combined.

Trying to be any good at such tasks

Did nobody read the OP? There's visual charts and a table.

Click the images at the top of the thread.

Open up the anydice programs if you wanna see the math.

Tl:Dr;
>Because the amateurs get 4 rolls.
Though OPs hypothetical peanut gallery isn't all +0, it's +0/+1/+1/+2.

>than Grok barbarian, Steve fighter, Billy rogue, and Sally cleric combined.

So you're telling me that because one guy cannot reliably beat four guys in Trivial Pursuit who are all working together against him, the system is fundamentally flawed?

THEN PAINT ME BLACK AND CALL ME A MONKEY, because I thought that was a pretty accurate reflection of real life. Indeed more than anyone a Dungeon Master should know this, because 90% of DM woes are caused by his players, who outnumber him, being able to think up things that he, as one guy, could not.

>Trying to be any good at such tasks

Zot IS good at such tasks. It's just that his brain, fabulously learnéd as it is, is not as good as four brains combined.

Whoop dee fucking doo.

Not sure what would be considered a 'fix' in that case. The major assumption under this errata is that not only do the DC's go down a few points, any reasonably experienced character is going to get 2 ranks of proficiency in a skill they want to be considered experts in.

Everybody can take 10 under non-stressful conditions, and experts take 10 at all times. The proposed DC table just adds 10 to the average bonus of characters at creation, and upgrades in tier as their proficiency and ability scores rise.

Amateurs auto-succeed on easy to moderate as they level, while experts auto-succeed hard tasks from the get go and by the end of their careers can contest nearly impossible tasks at a little less than 50%, assuming they don't have advantage, which is made easier by the third rank of proficiency which they easily have by that point.

That said I didn't address the part 2 questions.

blogofholding.com/?p=181

So why should Zot bother rolling then?

Why should he even take the skill? Just throw another +2 into the peanut gallery and up it to 5 unskilled checks. the group clearly doesn't need knowledge skills, someone will always know the answer without a specialist, and with a specialist, it's still rarely going to be the specialist who solves it.

There's no take ten in 5e.

The closest to take 10 is the autosucces rule, which averages out to a take 5 rule.

Well, I didn't say that the problem didn't need fixing in 5e. I said that take 10 IS my fix. I buffed that auto-success rule.

I don't think it's perfect though which is why this thread is a good thing.

Gotcha.

I personally think the solution lies in somehow changing the mechanics for group checks.

Take 10 means you start to match the peanut gallery in effectiveness around +6 though, which I admit is better.

I'll have to run the numbers on that one.

>So why should Zot bother rolling then?

1) Because Zot is more likely to succeed on any one given roll, and often you only get the one roll. When the Sphinx asks a riddle of the party and allows them only one answer, what makes the most sense is for the party to use the Aid Another action on Zot's check, giving him advantage.

2) Because the system is designed to enable Zot's companions so that if they individually need to pass a given check, they can, even if it's difficult. Zot's reward for investing in Knowledge is the ability to hit DCs higher than 20, and in doing so enable his party as a whole to progress.

>the group clearly doesn't need knowledge skills

No one in the group, even 's group of +0/+1/+1/+2, is capable of hitting a DC 25, so in fact they clearly do.

How many DC25 knowledge checks does the party really need to pass?

And in the sphinx scenario, theyd all roll knowledge before anyone talks.

The party is better off if the wizard takes whatever he can that *isn't* knowledge or insight.

For knowledge and insight, you almost always get one roll, per character.

If DC25 comes up, the party fails. The GM isn't going to have the game stall when that happens. So the one time a year that actually comes up you take the badstuff and handle it.

>How many DC25 knowledge checks does the party really need to pass?

I dunno. But they're really not going to like it when it happens.

Besides which, it's not just about Zot's knowledge roll, it's about the system as a whole. A DC 25 Knowledge check might not be that common. But a DC 25 lock? A DC 25 Acrobatics check when on a rolling ship in a tornado? A DC 25 Persuasion check?

Have you never heard the maxim that it's better to have a thing and not need it, then need it and not have it?

Taking a somewhat different tack on this - the party is REALLLY not gonna like it when they decide that because the 4 on them can untrained on average hit any DC under 20 better than 1 person specialized in a thing, they don't need someone trained in Perception.

And that's when a series of thieves steals all their money purses when they walk through town, because the thieves are actually trained in Stealth and Sleight of Hand and can beat their shitty Perceptions with ease.

Exactly.

Zot (and his party) will be glad he took proficiency in a skill with Type A or Type B checks instead, and that Zots proficiencies somehow include stealth, perception, and athletics, instead of arcana or geography.

Perception is type A as often as it is type C. You often have to roll it without the team to back you up.

This seems rather arbitrary. Any skill can be A, B, or C, and you will never find a system that satisfies all three without necessarily sacrificing the ability of untrained users to ever pass any but the most trivial of checks in that skill.