Yes, that is why I wrote :

"For the same power, AND SAME RPM, more force needed..."

I know that some people will avoid the additional force needed by increasing RPM (also true on fixie :-), that is why I also discussed RPM right after...

This is true but its not the whole story.
First, if rpm is the same, the pedal will move slower on a short crank. This will result in muscle contracting slower which increases the muscle's ability to produce force due to force-velocity characteristics. See the figure I linked to in the Rohan Denis thread.
Second, if you raise your saddle so as to maintain maximum extension, the leg will be more extended throughout the rest of the cycle. A more extended leg is stronger. You probably know you can do more weight with a partial squat than a deep squat. In my field this is called biomechanical gear ratio.
For those two reasons your muscles are capable of greater force production. And, as you mentioned, you can always change gear too.
But all these are in the nuance. McDaniel and colleagues showed clearly that crank lengths between 145-195mm have no effect on metabolic cost / efficiency.
Cheers,
Jim

Interesting, thanks

Not just "some people". I'll stop short of saying "all people", but I would certainly argue "most people"....and probably everyone here. We shift when our legs tell us "that's too hard, we can't keep this up for very long". Case in point, here is some data from me. Two rides, one on the road bike with 175mm, and one on my TT with 150mm. Both are 20min 90% IF efforts:

road (175mm): 5 September 2019, AvgPwr = 214 watts, AvgCad = 91 rpm, Speed = 21.1 mph, position=On-Hoods
TT (150mm): 9 October 2019, AvgPwr = 217 watts, AvgCad = 92 rpm, Speed = 24.7 mph, position=Full-Aero

I have no idea what gears I was in each time. I don't ride to a set cadence---Often I don't even have it displayed on my head unit. I simply select the gear that feels good at the time, for an effort level that my brain tells me I can hold for the remaining duration. I don't even ride to a set power---I ride to an effort level, and expect power to be in a certain range.

At 25mm difference, my TT cranks are 14% shorter than my road bike. That's a pretty extreme change by most peoples' standards. Yet, my power and cadence differ by 1/10th of that (~1.5%).

Yes, that's just one example---which does not a proof make. But, its a pretty extreme example of changes in bike, position, and crank length....which are indistinguishable when looking at power output and RPM.

I made a decision,

I am going with a 50/34 160mm length crank and an 11-23 cassette on the rear.

I am super stoked to run this starting in December.

This both makes sense objectively and tallies with my subjective experience when I switched from 172.5 to 160 cranks.
I haven't changed cadence significantly and have maintained the same power. Part of the reason I changed was that the load on my knees sometimes felt uncomfortable at the top of the stroke - possibly die to a bit of instability in my knees. Since dropping the top of stroke position by 25mm, the angle of my knee at that point in the stroke is straighter and thus it's both more stable and able to produce higher force. This increased ability to produce force seems to easily compensate for the reduction in crank length and I can produce as much or more torque compared to before, while seated. My knees are feeling much more comfortable for the same power output and my stroke is smoother.

Cool, thanks for sharing.
The issue of tight hip angle at the top of the pedal stroke was a major part of my recent presentation at ISCO. Everyone knows you should set maximum leg extension. Why not set minimum leg extension (by chancing crank length) too?