Why not allow stacked advantage/disadvantage?

[tchntm43]

I know what the rule is, I just don't understand why it is that way. For example, the rogue jumps out from a hidden place to strike an enemy who didn't know he was there, but the enemy is also currently fighting an ally of the rogue. Give the rogue an extra d20 on the attack role for being hidden, and an extra d20 for having an ally next to the target. So he'd roll 3 d20s and pick the highest one.

Or if the enemy also has some kind of magical effect that incurs disadvantage on attack rolls, then instead of just having a normal attack, the disadvantage cancels one of the advantages, leaving the rogue with 2 d20s on the attack roll.

It just seems like it would be way more fun and intuitive if it worked this way.

[titi]

I know what the rule is, I just don't understand why it is that way. For example, the rogue jumps out from a hidden place to strike an enemy who didn't know he was there, but the enemy is also currently fighting an ally of the rogue. Give the rogue an extra d20 on the attack role for being hidden, and an extra d20 for having an ally next to the target. So he'd roll 3 d20s and pick the highest one.

Or if the enemy also has some kind of magical effect that incurs disadvantage on attack rolls, then instead of just having a normal attack, the disadvantage cancels one of the advantages, leaving the rogue with 2 d20s on the attack roll.

It just seems like it would be way more fun and intuitive if it worked this way.

If I had to guess :
Because it's simpler this way
Because huge chains of advantage or disadvantages are easier to abuse

[Psyren]

1) It's not allowed because rewarding every source of advantage (or punishing every source of disadvantage) would mean counting up and keeping track of them all, which would take us back to the 3.5 days of tracking every single +/- modifier before we can resolve a turn and that would slow down play, e.g. he's prone and I'm attacking from higher ground because I'm on horseback, but my horse is balancing because the ground is icy and my hands holding my lance are numb from the cold, and also it's dark so I can't see him but he's also restrained because he's standing in the wizard's web spell and the monk is flanking... etc.

2) If your DM really does feel one side is overwhelming the other, they can just override the prevailing conditions and impose advantage or disadvantage to the test regardless of what the rule says. But the rule itself is that the two cancel out no matter how much one outnumbers the other (see #1.)

3) This would easily lead to a textbook "optimizing the fun out of the game" scenario. Even if the only thing the player needs to worry about in a given scene is advantage, they'd be encouraged to scrape for every source of it every time, because every d20 they add increases their chance of a natural 20. Consider how powerful a feat like Elven Accuracy is at just attacking with 3d20kh, now imagine a player being able to get 4d20kh or 5d20kh without any investment at all on a variety of checks.

[Ionathus]

Yep, it's to simplify the logistics for the DM: the flowchart gets a lot shorter if the only two questions that apply are:

"Does the roller have advantage from anything?"
"Does the roller have disadvantage from anything?"

This means that you don't have to think any farther once you've answered those two questions -- this also keeps players from trying to dogpile as many advantages as possible, and refocuses you on "can you do one thing to improve your chances? Great. That either gives you a clear bonus (advantage) or it negates a clear penalty (roll without disadvantage)."

As I understand it (mostly from other peoples' accounts; my experience with 3.5e was minimal), in previous editions players often spent a lot of time counting up different advantages and tallying the bonuses (https://www.giantitp.com/comics/oots0034.html). 5e's designers wanted to reduce the paperwork and math as much as possible, so they centered a lot of those minor bonuses in the advantage/disadvantage mechanic. Spells that used to give a +3 or something to your roll now just grant advantage. So if two spells would've given you a +6 total, now they just both give advantage, which doesn't stack (because 5e really wants "flattened math" where bonuses can't get too ridiculous).

Along with the Concentration mechanic, this has drastically reduced the pre-buffing before fights1, because there's reduced incentive to pile on a ton of bonuses. You're able to just pop one or two spells / potions, and rush into the fight, rather than spending half an hour idling in front of the boss door layering on buff spell after buff spell after buff spell. Which is something I really enjoy about 5e, my only other major D&D experience being Baldur's Gate II where every major enemy requires at least 10-15 minutes of juggling pre-buffs before you can even start the fight. I *do* enjoy the logistics there but it can definitely get old, and it would get a LOT more tedious if I was trying to share those logistics with 3-5 other humans.


1. A little too much in my opinion, but I think overall it's been effective at the goal

[Rerem115]

Yeah, stacking Disadvantage dice can pretty quickly lead to a place where failure is all but guaranteed for some situations, and vice-versa. Being knocked Prone in a Web spell while in Darkness would have a significant number of entities taking the lowest of 4(!) dice on most attacks, while any attacks against them would be taking the highest of 4 dice.

While not necessarily common, scenarios like that aren't uncommon, either - the one I listed can get set up in a single round by as few as 2 PCs - and players would be encouraged to stack the dice in their favor as often as possible. The end result would make stacking conditions far more debilitating, and swing combat towards whoever could inflict the most, the fastest.

[stoutstien]

You already have situational modifiers to work with so you don't need to adjust the base advantage/disadvantage.

[Silverblade1234]

And one other: it also increases the degree to which the DM has to negotiate with their players, who are now incentivized to argue for every source of advantage and against every source of disadvantage. Removing stacking simplifies the table management, in addition to the math management. To do otherwise would require a much more rigorously catalogued approach to adv/dis in the game mechanics, which is fine, but it's just not the game the designers wanted 5E to be (cf. 3.5E or Pathfinder but replace roll modifiers with adv/dis).

[Lvl 2 Expert]

I agree with the rest in that advantage was designed to be simple. In 3.5 some feats/actions/features/circumstances gave a +2, others a +3 or a +5 or a +10 or... This was because some advantages were considered bigger than others. In 5e this was scrapped for everything but a few particular bonuses, everything is straight up advantage now.

Now let's say your rogue attacks an enemy from hiding, while that enemy is already engaged with the fighter, but that enemy also has the indominable armor of "no rogue will fack with me". There's two sources of advantage here, and only one source of disadvantage, so the rogue, under a system where adding up is allowed, would still have regular advantage left. However, this armor is so magical that in 3.5 it would have been statted as a -25 to the rogue's attack, easily bigger than the bonus of attacking from hiding and from attacking while the enemy is engaged combined. Yet, because the system was simplified, this super expensive magic armor is now almost worse than useless as long as the rogue can find two very common sources of advantage and completely overpower it.

Okay, that isn't the greatest example, I admit. But the general principle is: the advantage system is based on not wanting to keep tabs on how much of an advantage every source advantage realistically gives. Ones you start adding them together you basically make that "the same, every source of advantage is actually equal, it's not just to do less math".

(Yes, sure, counterpoint: under the curent advantage system ome minor course of advantage can balance out two or three huge sources of disadvantage, which is unfortunate as well, but at least there's less bookkeeping and less thinking hard about where to find more advantages and disadvantages to apply there.)

Aside from that I'm not sure I would make hiding and flanking stack to begin with even in a stacking system. Both kind of boil down to "he didn't see it coming". So double advantage for those two would be... "he reaaaally didn't see it coming"? "He was looking even more the other way"? :smallbiggrin:

Edit: that last one might accidentally have been a better point than the rest of my post. Sources of advantage can overlap, do the same thing, or just not work together. That's another rabbit hole to go down when trying to stack them. Advantage from a crowbar and advantage from someone helping don't stack, because the crowbar isn't big enough to fit two people at it.

[Mindflayer_Inc]

I know what the rule is, I just don't understand why it is that way. For example, the rogue jumps out from a hidden place to strike an enemy who didn't know he was there, but the enemy is also currently fighting an ally of the rogue. Give the rogue an extra d20 on the attack role for being hidden, and an extra d20 for having an ally next to the target. So he'd roll 3 d20s and pick the highest one.

Or if the enemy also has some kind of magical effect that incurs disadvantage on attack rolls, then instead of just having a normal attack, the disadvantage cancels one of the advantages, leaving the rogue with 2 d20s on the attack roll.

It just seems like it would be way more fun and intuitive if it worked this way.

Simplicity, though just use poker chips and you can do stacking advantage/disadvantage really easily.

Personally I'm ok with either option but sometimes "gamey" ideologies just win out bc it's easier.

I do kinda love if a setting takes a gamey mechanic and incorporates it into the setting. Like, this is just how luck and anti-luck work, for one description of advantage, the fundamental forces are just lazy that way lol

Edit

Side note, one of the fundamental designs of 4e was to simplify the game, they took that and went further with 5e. In 4e you could get stacking +2/-2. Advantage (usually, kinda) equals out to +3/-3 to a roll.

[KorvinStarmast]

And one other: it also increases the degree to which the DM has to negotiate with their players, who are now incentivized to argue for every source of advantage and against every source of disadvantage.
Removing stacking simplifies the table management, in addition to the math management. And for this I thank them.

[Mindflayer_Inc]

And for this I thank them.

I don't know. If you're with "your group" (you know, your friends or whatever) then having something to engage players is fantastic. Sure a random group could get annoying but if you know each other you would know what to push for and what not to.

I've found this is also good with new players too... Just not when you get the type of player who is just trying to be disruptive for lolz.

[Rerem115]

Knowing my groups, the first thing we'd do with this hypothetical change is pull some Discworld-esque shenanigans with Lucky. Like, full on Guards, Guards! deliberately inflicting Disadvantage to get that million-to-one chance.

'The gnome Barbarian drops his swords to pull out a greataxe, go prone, close his eyes, and make a called shot on the bandit's 'voonerables'.'

'...So, that's 5? d20s, take the lowest? You sure about that?'

'Yes. I also burn a lucky die to make that 6d20, take the highest. And, I'm pretty sure that crits.'

[Psyren]

Knowing my groups, the first thing we'd do with this hypothetical change is pull some Discworld-esque shenanigans with Lucky. Like, full on Guards, Guards! deliberately inflicting Disadvantage to get that million-to-one chance.

'The gnome Barbarian drops his swords to pull out a greataxe, go prone, close his eyes, and make a called shot on the bandit's 'voonerables'.'

'...So, that's 5? d20s, take the lowest? You sure about that?'

'Yes. I also burn a lucky die to make that 6d20, take the highest. And, I'm pretty sure that crits.'

An effective illustration.


I don't know. If you're with "your group" (you know, your friends or whatever) then having something to engage players is fantastic. Sure a random group could get annoying but if you know each other you would know what to push for and what not to.

I've found this is also good with new players too... Just not when you get the type of player who is just trying to be disruptive for lolz.

For new players it's even more important to break them out of the X debits/Y credits mindset.

[Kane0]

Ive allowed them to cancel on a 1:1 basis for years, hasnt really affected much during normal play but does smooth out some edge cases.

Letting them stack for additional d20s i think would cut into Elven Accuracy, and in the case of savvy parties almost impossible to make rolls for NPCs (eg stacking multi-disadvantage on top of a mind sliver/bane or cutting words or what have you)

[InvisibleBison]

Advantage (usually, kinda) equals out to +3/-3 to a roll.

Maybe in theory it does, but in practice the odds of rolling two d20s and having the result differ by 3 is extremely low. Most of the time, advantage is going to be worth a lot more than +3, and disadvantage a lot less than -3.

[JackPhoenix]

Ive allowed them to cancel on a 1:1 basis for years, hasnt really affected much during normal play but does smooth out some edge cases.

Letting them stack for additional d20s i think would cut into Elven Accuracy, and in the case of savvy parties almost impossible to make rolls for NPCs (eg stacking multi-disadvantage on top of a mind sliver/bane or cutting words or what have you)

Pretty much this. Having them cancel each other 1:1, but only getting (dis)advantage in the end instead of stacking the dice is perfectly fine and better than what we have (Oh no, I'm blind, poisoned, restrained and lying prone 500' away, and the enemy is also lying on the ground, but he can't see me either, so I can shoot him pefrectly fine) I don't know why people always immediately jump to the 2nd interpretation.

[Mindflayer_Inc]

Maybe in theory it does, but in practice the odds of rolling two d20s and having the result differ by 3 is extremely low. Most of the time, advantage is going to be worth a lot more than +3, and disadvantage a lot less than -3.

I don't know, plenty of times I get numbers that are really close. There are also times I get a 1 and a 20 (or whatever number far apart).

So the actual worth being about +/- 3 feels about right in actual play.

[Psyren]

Pretty much this. Having them cancel each other 1:1, but only getting (dis)advantage in the end instead of stacking the dice is perfectly fine and better than what we have (Oh no, I'm blind, poisoned, restrained and lying prone 500' away, and the enemy is also lying on the ground, but he's can't see me either, so I can shoot him pefrectly fine) I don't know why people always immediately jump to the 2nd interpretation.

You don't need to cancel 1:1 though, just eyeball it. If I as the DM saw that you were blind, poisoned, and restrained, I'd probably stop there and override you to disadvantage without bothering with counting up the advantages on the other side unless they were overwhelming.

[RSP]

I’m a fan of them stacking though not currently playing that way.

One thing I like about it is it makes strategic decisions more worthwhile.

Certainly is simpler just having the rule as written though.

[Kane0]

Pretty much this. Having them cancel each other 1:1, but only getting (dis)advantage in the end instead of stacking the dice is perfectly fine and better than what we have (Oh no, I'm blind, poisoned, restrained and lying prone 500' away, and the enemy is also lying on the ground, but he can't see me either, so I can shoot him pefrectly fine) I don't know why people always immediately jump to the 2nd interpretation.


You don't need to cancel 1:1 though, just eyeball it. If I as the DM saw that you were blind, poisoned, and restrained, I'd probably stop there and override you to disadvantage without bothering with counting up the advantages on the other side unless they were overwhelming.

In my experience the most common instances are two-and-one and two-and-two. Three sources of either advantage or disadvantage are somewhat uncommon and usually take intentional synergy to get to, which is kind of the point (Disregard this statement if you use Flanking).

Oh, I should mention that under this 1:1 ratio having Elven Accuracy counts as two instances of advantage for the purposes of cancelling, so if said elf has the benefit of say Guiding Bolt but is Frightened they still have one regular advantage rather than what would otherwise be nothing, absent other factors.

[InvisibleBison]

I don't know, plenty of times I get numbers that are really close. There are also times I get a 1 and a 20 (or whatever number far apart).

So the actual worth being about +/- 3 feels about right in actual play.

Yes, that's my point. It's about as likely for the results of the two d20 rolls differ by 3 or less as it is for them to differ by 9 or more (32% vs 33%, respectively). That doesn't feel anything like rolling with a +3 bonus to me; rolling with advantage feels like you're getting a very big bonus, because a lot of the time you are.

[Slipjig]

I've never checked the math, but I've always read that Adv/Disadvantage works out to be +/-5.

[PhoenixPhyre]

I've never checked the math, but I've always read that Adv/Disadvantage works out to be +/-5.

Which is true only if the adjusted target number (ie DC - bonus, the number you need to roll on the die) is about 10 +- 1 or so.

[tKUUNK]

Knowing my groups, the first thing we'd do with this hypothetical change is pull some Discworld-esque shenanigans with Lucky. Like, full on Guards, Guards! deliberately inflicting Disadvantage to get that million-to-one chance.

'The gnome Barbarian drops his swords to pull out a greataxe, go prone, close his eyes, and make a called shot on the bandit's 'voonerables'.'

'...So, that's 5? d20s, take the lowest? You sure about that?'

'Yes. I also burn a lucky die to make that 6d20, take the highest. And, I'm pretty sure that crits.'

lol, this is great. ^

I for one love the simplicity of non-stacking advantage. And I enjoyed the days of algebraic D&D (3.5e). The simplicity we have now is golden. We can just enjoy the story, and new players aren't overwhelmed or left wondering why anyone even bothers trying to learn the game.

There's the strange exception of Elven Accuracy. But even that doesn't outweigh a single Disadvantage, so it isn't exactly what we're talking about here with stacking. Just a side note I guess.

[Dalinar]

I've never checked the math, but I've always read that Adv/Disadvantage works out to be +/-5.

If your base success chance is 50%, your advantage success chance is 75%, which is the same as +-5. The farther the chance goes from that, the less impactful advantage is. Going from a 5% success chance to a 9.75% chance in the extreme case, for the equivalent of a little under +1.

Things get a little wonky in "nat 1 autofails / nat 20 autosucceeds" situations because advantage helps you on that but +X does not. Advantage feels like it adds a ton of consistency in part because it reduces that nat 1 chance, regardless of how similar it is to +X or not. Situations where "one die would have failed, but the other succeeded" come up a ton with it. (Not that that stopped my 19AC Paladin from getting crit through a dodge in a recent boss fight. :smallmad: )

[Mindflayer_Inc]

If your base success chance is 50%, your advantage success chance is 75%, which is the same as +-5. The farther the chance goes from that, the less impactful advantage is. Going from a 5% success chance to a 9.75% chance in the extreme case, for the equivalent of a little under +1.

Things get a little wonky in "nat 1 autofails / nat 20 autosucceeds" situations because advantage helps you on that but +X does not. Advantage feels like it adds a ton of consistency in part because it reduces that nat 1 chance, regardless of how similar it is to +X or not. Situations where "one die would have failed, but the other succeeded" come up a ton with it. (Not that that stopped my 19AC Paladin from getting crit through a dodge in a recent boss fight. :smallmad: )

"1 is an automatic miss" needs to be thrown out so bad.

Works fine as an alternate rule but when some classes don't need to ever touch a d20 it feels like the game is flipping the bird to weapon users.

[stoutstien]

"1 is an automatic miss" needs to be thrown out so bad.

Works fine as an alternate rule but when some classes don't need to ever touch a d20 it feels like the game is flipping the bird to weapon users.

...why...if a one wouldn't miss them why are you rolling?

[Mindflayer_Inc]

...why...if a one wouldn't miss them why are you rolling?

You can still miss even if you don't have auto-misses.

Rolling a 2 may not hit that 20 AC (high CR creature), but it will hit that 12 AC (low CR creature).

[Amnestic]

Works fine as an alternate rule but when some classes don't need to ever touch a d20 it feels like the game is flipping the bird to weapon users.

I confess I'm not quite sure why saving throws don't have the same auto-succeed/auto-fail as attacks do.

[Chronos]

If you needed to roll an 8 to succeed, and your Advantage rolls were a 7 and a 15, advantage wasn't worth a +8 in that situation; it was worth a +1.

The designers might claim that Advantage/Disadvantage not stacking was for simplicity, but it actually has the opposite effect. Everyone trying for every edge they can get is the simple outcome, because it's what naturally and intuitively happens in any real competition. But really, the reason why they made it not stack was because they had a design philosophy of making sure that nobody ever got competent at anything, and stacked advantage could become too much like competence.

Quoth Rerem115:

Knowing my groups, the first thing we'd do with this hypothetical change is pull some Discworld-esque shenanigans with Lucky. Like, full on Guards, Guards! deliberately inflicting Disadvantage to get that million-to-one chance.

'The gnome Barbarian drops his swords to pull out a greataxe, go prone, close his eyes, and make a called shot on the bandit's 'voonerables'.'

'...So, that's 5? d20s, take the lowest? You sure about that?'

'Yes. I also burn a lucky die to make that 6d20, take the highest. And, I'm pretty sure that crits.'
That's not a reason to not stack advantage; that's a reason not to use that weird houserule for Lucky. RAW, Lucky doesn't work that way. You'd have the choice of either the lowest of those five dice, or whatever you got on the sixth one. Which, yes, would be slightly better odds than if you had just rolled normally, but it'd be much worse odds than you'd get if you used Lucky but didn't do all of those stupid things.

[Rerem115]

That's not a reason to not stack advantage; that's a reason not to use that weird houserule for Lucky.

Except it's not a house rule?


Whenever you make an attack roll, an ability check, or a saving throw, you can spend one luck point to roll an additional d20. You can choose to spend one of your luck points after you roll the die, but before the outcome is determined. You choose which of the d20s is used for the attack roll, ability check, or saving throw.

Lucky already turns disadvantage into super-advantage. Stacking more dice turns it into super-duper advantage just highlights it.

Getting back to the 'competence' thing though, at least to my understanding it was to minimize the capacity to brute force issues, to make it harder to stack as many modifiers or dice as possible in order to preclude the need for rolling.

And, by making explicit in the design, make players less likely - albeit, not completely - to chase every possible bonus, since most of them don't stack.

[RSP]

And, by making explicit in the design, make players less likely - albeit, not completely - to chase every possible bonus, since most of them don't stack.

How much is this really a problem though? Serious question.

Because I’ve always seen it as the opposite: once something is in play, it makes any other tactic decisions moot. Usually this just affects abilities though (like Reckless Attack - no one ever really thinks of “oh should I Help the Barb?”).

For RP, I’d imagine every adventuring group ever tries to stack as much as possible in their favor, but it essentially becomes meaningless.

[Doug Lampert]

Yes, that's my point. It's about as likely for the results of the two d20 rolls differ by 3 or less as it is for them to differ by 9 or more (32% vs 33%, respectively). That doesn't feel anything like rolling with a +3 bonus to me; rolling with advantage feels like you're getting a very big bonus, because a lot of the time you are.

You're assuming that without advantage you get the lower die, which is not always true. 52.5% of the time, advantage give +0, because the first die is as high or higher than the second. Another 13.5% differ by 3 or less, for 66% chance of an improvement of zero to three.

Meanwhile, only 16.5% improve by nine or more.

Most times, you're rolling against a DC, but if you're just rolling for highest value, then 1d20 averages 10.5, and advantage averages 13.825, or a bit more than 3 additional points.

[Chronos]

Lucky turning disadvantage into super-advantage is totally a houserule. The rules for advantage/disadvantage say how to handle situations like that: The extra roll only affects one of the two dice. If you're rolling at disadvantage, you roll dice A and B. A comes up 3, and B comes up 12. You decide to use Lucky, presumably on that 3. You roll die C, and it comes up as 14. You naturally choose to use that instead of the 3. Now your disadvantage roll is a 14 and a 12 (counts as 12), instead of the 3 and 12 (counts as 3) that you had before. It's better than rolling disadvantage without Lucky, but it's worse than rolling Lucky without disadvantage.

[RSP]

Lucky turning disadvantage into super-advantage is totally a houserule. The rules for advantage/disadvantage say how to handle situations like that: The extra roll only affects one of the two dice. If you're rolling at disadvantage, you roll dice A and B. A comes up 3, and B comes up 12. You decide to use Lucky, presumably on that 3. You roll die C, and it comes up as 14. You naturally choose to use that instead of the 3. Now your disadvantage roll is a 14 and a 12 (counts as 12), instead of the 3 and 12 (counts as 3) that you had before. It's better than rolling disadvantage without Lucky, but it's worse than rolling Lucky without disadvantage.

Here’s the actual wording for reference, if anyone wants it:

“When you have advantage or disadvantage and
something in the game, such as the halfling’s Lucky trait, lets you reroll the d20, you can reroll only one of the dice. You choose which one. For example, if a halfling has advantage or disadvantage on an ability check and rolls a 1 and a 13, the halfling could use the Lucky trait to
reroll the 1.”

[Rerem115]

Hm? I've been operating under the assumption that Lucky gets to break that specific rule for 7 years now, ever since that was confirmed in Sage Advice back in 2016. Relevant text in the spoiler.

How does the Lucky feat interact with advantage and disadvantage? The Lucky feat lets you spend a luck point; roll an extra d20 for an attack roll, ability check, or saving throw; and then choose which d20 to use. This is true no matter how many d20s are in the mix. For example, if you have disadvantage on your attack roll, you could spend a luck point, roll a third d20, and then decide which of the three dice to use. You still have disadvantage, since the feat doesn’t say it gets rid of it, but you do get to pick the die.

The Lucky feat is a great example of an exception to a general rule. The general rule I have in mind is the one that tells us how advantage and disadvantage work (PH, 173). The specific rule is the Lucky feat, and we know that a specific rule trumps a general rule if they conflict with each other (PH, 7).

[RSP]

Hm? I've been operating under the assumption that Lucky gets to break that specific rule for 7 years now, ever since that was confirmed in Sage Advice back in 2016.

Could well be RAI (though I dislike that term and it has flaws), but I’m not sure it’s a great example of specific over general.

The wording of Lucky doesn’t at all make it clear it’s over riding the general rule; while the general rule is set up to specifically restrict what the SA states is happening in Lucky.

The plural in Lucky (“You choose which of the d20s is used...”) could refer to the 2 d20s as per the “general rule”, or the 3 d20’s as per the SA.

So the SA claim is it overrides at least two “general rules”, one of which is very specific in only applying to this one situation (the other being the Disadvantage rule), yet the wording doesn’t make that clear (if it indeed was intended when written that way, as SA states).

[Chronos]

Honestly, by this point, the fact that Sage Advice takes one position on something is almost an argument for the opposite position, given how often it gets things wrong.

[Gurgeh]

The issue is that the pre-emptive wording around advantage/disadvantage explicitly deals with re-rolls, while the Lucky feat doesn't use those and instead uses a parallel mechanic. I could see some tables ruling the Lucky feat's resolution as "handle the advantage/disadvantage roll and reduce it to whatever value you'd get, then roll an extra d20 and use that instead of your original result if you want", but that doesn't seem any better supported by the text of the Lucky feat than the super-advantage interpretation.

(the 5e writers also deserve a bonk on the head for using the exact same name for two completely different features printed in the same book)

[Psyren]

The Sage Advice ruling (https://www.dndbeyond.com/sources/sac/sage-advice-compendium#Lucky) is all that matters to my table/DMs; Lucky = super advantage.

[Gurgeh]

It's also entirely valid to say that it's a bad rule that makes the game worse and deliberately choose not to use it - either way, it's definitely part of the rules of the game and there's nothing to be gained by arguing otherwise.

[Chronos]

If we rule that that rule in the Advantage section doesn't apply, though, then we simply don't know how Lucky interacts with disadvantage. The feat says "You choose one of the d20s", but it doesn't define the scope of "one of the d20s": One of which d20s? Nobody plays that it means no limitation on scope and that it can be any die sitting on the table at the moment. It's one of the relevant d20s. But which ones are relevant? Absent any other rule, that could mean "any of the dice you rolled for any reason as part of that resolution", or it could mean "that die, or the specific other die you're replacing". Even if the rule on page 173 doesn't precisely apply, it's the closest thing we have to a rule that applies.