For exactly the reasons Malcat outline, I think it's a lot more instructive to pick one of the dynamic withdrawal scenarios that never run out of money, like the "withdraw the lower of 4% of the initial stash, inflation adjusted, or 4% or the current stash", which sort of mimic how a real person would behave. Then look at what the minimum withdrawal rate was across the scenarios and see if you'd be OK with that number.
Instead of "you ran out of money in 5% of the realizations", it's a lot more instructive to know whether your $200k spend had to be reduced to $190k or $10k. If the "failure" is the former, you'd probably be OK with it. Not so much with the latter.
My argument isn't even to pick a variable WR, my argument is that most people will naturally vary their withdrawal rate given ongoing market conditions that threaten to erode their principle.
I don't recommend putting faith in any output from a FIRE calculator. Although, it is very helpful to compare fixed WR to variable to understand the magnitude of difference.
I don't really recommend anything beyond being realistic about what any of the math actually means.
I mean, there's some balance point between "f it, I'm sure it'll be fine" to "the calculator shows 0.64% risk of failure so it won't work". You need some tool to help tell you what the outcome would have been in different situations to give you an idea of whether you need to worry about a 2008-level recession or a 1930-level depression, or whatever. If you don't have that tool you're just as likely to react to minor market swings by excessively cutting spending as you are to thinking everything is fine until you've eroded your principal to an unhealthy point.
I feel like I'm saying the same thing over and over again and never managing to say it clearly.
First, I wasn't saying that I I'm against recommending a planned variable WR, I meant that I never personally recommend any universal approach to anything. I was clarifying my previous point that I wasn't previously making a recommendation. That's all.
Second, sure there's a balance. There's a balance between understanding what the calculator can do and what it can't.
The calculator can indicate the relative impact of a market drop, and the relative impact of a fixed WR from a variable rate, but it cannot actually output a predictive value.
I spent ages with these calculators and they were invaluable to me in terms of understanding how all of this works, but the more I fiddled with them, the more I understood that the percentage output is only meaningful as a comparison.
The percentages could be changed to any scale from red-->blue, or grades F-->A+, or even the pain scale faces at the doctor's office. They would still be just as informative, and probably more helpful because they would remove the illusion of predictive value.
The only thing an output of 100% says compared to 80% is "given the inputs that you chose, there is a significantly higher magnitude of positive impact on projected outcome compared to your previously selected inputs."
If I'm being totally unrealistic my ability and willingness to cut my spending in down years, it doesn't matter if the calculator spits out 100% compared to 80%.
Put more simply, if the calculator has no inter-user predictive validity, then it has no predictive validity period. Given the same inputs, it can't tell the difference between an unrealistic person and an overly cautious person. One might be very likely to end up needing to go back to work and the other might be very likely to die rich, and the calculator might spit out the same 90%.
So is a variable WR a good idea? Sure, for some people.
Is a bond tent a good idea? Sure, for some people.
Is a sub 4% WR a good idea? Sure, for some people.
Is being prepared to have to generate money in retirement a good idea? Oh absolutely...for some people.
That's why I don't make generalized recommendations beyond very, very vague concepts, because the factors that determine which people those "some people" are are very personal and individual in ways that calculators can't effectively account for.