Assorted Attempts At Augmenting Assurance

[Artificial 20]

As-Is
Take an absolute total of 10 at trained/15 at expert/20 at master/30 at legendary.

Alternatives
Double rank-based proficiency bonus (+2 at expert/+4 at master/+6 at legendary), treat critical successes as normal successes.
Improves with rank, but does nothing at trained.

Roll 2D20 and take the better result, treat critical successes as normal successes.
Scales, but does not improve with rank except for the existing bonus.

Roll 3D20 and take the middle result.
Scales, but does not improve with rank save the existing bonus.

Add +5 untyped bonus to roll, replace D20 with D10.
Scales, but does not improve with rank and is unconventional.

Add +5 untyped bonus to roll, replace D20 with D4 at trained/D6 at expert/D8 at master/D12 at legendary.
Improves with rank, very unconventional.

Remove rank-based proficiency bonus and treat any result on the dice lower than 4 at trained/6 at expert/8 at master/12 at legendary as that number.
Scales and improves with rank, numbers may need tuning (legendary is still worse on average than rolling, since the +3 is sacrificed).

Accepting analysis, additional adaptations and appreciation.

[Claxon]

I like a concept that would change assurance to let you "Take 4" at Expert, "Take 8" at Master, and "Take 12" at Legendary instead of rolling.

Which is a slight variation on what you're suggesting. It actually lets people be reliable good in skills they want to focus on. As it sits, a level 20 character that is legendary in a skill has a minimum roll of 24. 20 from level plus, +3 from legendary proficiency, and minimum of 1. But that's also assuming you have a 0 modifier in the tied ability, which is unlikely since by that level you're basically forced to put score increases into every score you'll probably have a +2 modifier at least. And for something you really want to focus in you could have a +6/7 from your modifier. Meaning that assurance...does nothing for you. Your minimum can be higher than what assurance gives you.

Now, that same character that can take 12 instead of rolling will have a 42, which is reasonable I think.

It's basically bringing back the take 10 rule, but just for a specific skill, and starts out worse and gets better over time. And, it's still not better than the best you could absolutely do, you're simply replacing the rolling part with an assured value thanks to your great competence with an ability.

[Azih]

Give it for free and have it scale with UTEML like Claxon suggests. Basically make it a core part of the skill system.

[WhiteMagus2000]

I like the idea of Assurance being a "take 10" type of thing. I would say take 8 at expert, 10 at master, and 12 to legendary. Just because I don't think anyone would take a feat that only let's you take 4. But that seems like a good and simple way to actually be useful.

[Mudfoot]

Take 10 and its ilk is much better than the current rule, but it suffers from the metagame problem that if you match the DC you automatically succeed, but if the DC is 1 higher you automatically fail. So it all comes down to knowing the DC.

So I would change it to: you roll the die as normal but the minimum die roll result is taken as:
Trained - 4
Expert - 6
Master - 8
Legendary - 10

So you can still attempt difficult things that would auto-fail the Take N variants, but you'll never screw up the easy stuff and it's extremely unlikely that you'll critfail.

[Artificial 20]

Thanks Claxon, and other posters who have replied. As I said in the OP, other ideas for Assurance are welcomed.

Mudfoot, your idea sounds like one of mine, but probably more elegant.

"Remove rank-based proficiency bonus and treat any result on the dice lower than 4 at trained/6 at expert/8 at master/12 at legendary as that number."

Yours avoids the "bonus removing" mess.

Paizo had Take 10 in P1E, and chose not to keep it.

I'm not sure we'll convince them to bring it back, for reasons such as 1 DC separating always-pass/always-fail.

Likewise, taking a fixed value being statistically superior to rolling, even at Legendary, may be a hard sell. Paizo seem to want the roll on the D20 to really matter.

A system where you treat dice rolls below X as X may have hope, if X is controlled. For a few examples, if X is 3/4/5/6, 2/4/6/8 or 5/6/7/8 at trained/expert/master/legendary, the dice roll still matters over half the time at legendary. I kind of like 3/4/5/6 myself.

[Talsharien]

Mudfoot wrote:

Take 10 and its ilk is much better than the current rule, but it suffers from the metagame problem that if you match the DC you automatically succeed, but if the DC is 1 higher you automatically fail. So it all comes down to knowing the DC.

So I would change it to: you roll the die as normal but the minimum die roll result is taken as:
Trained - 4
Expert - 6
Master - 8
Legendary - 10

So you can still attempt difficult things that would auto-fail the Take N variants, but you'll never screw up the easy stuff and it's extremely unlikely that you'll critfail.

I think that you are onto something here, but I would only allow the ability to be used when making Easy Difficulty checks. This fits with the original basis of being a master of simple tasks.

[Mudfoot]

I disagree. After all, if you're Legendary, you are something a bit special, and there's nothing legendary about reliably completing easy tasks. We all do that every day.

Besides, what about rolls vs monsters such as Athletics to trip, or those with no fixed DC like initiative? Are they "easy"?

[Themetricsystem]

Here is a new perspective.

What about instead of trying to emulate the "take a 10" or the like, why don't we instead try the opposite and have Assurance only apply in situations where the old "Take a 20" rules worked.

Think about it, what are you DOING when you are tying to 'assure' a thing.
as·sure
/əˈSHo͝or/
verb
make (something) certain to happen
to tell someone in a very strong and definite way that something will happen

Now, in its current form Assurance DOESN'T do this at all, in fact all it does it keep you from having to roll Trivial Checks (Which the Gamemaster section of the book already LITERALLY tells the GM they shouldn't be making you roll anyway...). That is heaping garbage, what we SHOULD be asking for is a way to ASSURE that we get the best result possible.

Allow a PC to take 10x as long attempting a given check as long as they are not in imminent danger to treat a given Check as an "unnatural" 20. There should probably be a specific carveout here to indicate that you can NEVER BY ANY MEANS Critically Succeed a task in this way.

Am I off my rocker here?

[Mudfoot]

Conceptually, being able to take 20 is a valid thing, like taking 10 is a valid thing. So a take 20 feat is equally as valid as a take 10 feat. How to name each one is a matter of preference.

[Starfox]

I've posted this elsewhere, but it seems very relevant here.

A rule of thumb I've been using is that anything that can be done in encounter mode is automatically successful in 10 minutes of exploration mode. Only when there is time pressure should we bother to roll. This is especially true in this edition, that requires such a lot of dice rolls for tasks like Open Locks or to resist a poison.