That is a common misconception that is completely false.
Check out the Marlin owners 444 sub-forum and the Beartooth Bullets articles on the 444.
The trick is to use WFN bullets that are short for their weight and drive em as fast as possible keeping a safe pressure.
Beartooth Bullets > Tech Notes > .444 Marlin- America's Most Versatile Big-Bore Part III
Bullets as heavy as 355 grains have been shot into 1" 100 yard groups using the 444S.
Longer than that and the 444P or other 1/20" do stabilize better.
I am having a 1/25 twist being made for the Encore to play with these.
I am currently shooting 280, 290, and 300 grain bullets in 2 different 1/38 guns successfully.
The Hornady 300 XTP also shoots well in the 444S according to many shooters.
No one really knows who started this rumor (a long time ago) but they obviously used cast bullets that were too small.
There weren't any heavy jacketed 44 bullets at the time.
It's been repeated by every gun writer that obviously never shot a 444 (most of them).
The 444 likes cast bullets .432 or .433 or you will get poor results.
The 444 has been misunderstood ever since it was introduced using a 240 grain Remington FP.
This is still fantastic deer medicine and the round really hit it's stride with the addition of the Hornady 265 FP which was designed for the 444.
The current Superformance factory ammo using the 265 is excellent and really can't be topped by the handloader (with that bullet).
They are now driving the 265 100 fps faster than Rem drove the old 240.
I know my shoulder could not take much more anyway.
Here are a couple of 40 yard targets I just shot with my 1980 444S
The first is using a bullet they say WILL NOT stabilize as it is over an inch long and the box says 1/20 only.
The second is my first load using a 280 out of the gate no development ladder yet.
The fact is just about any bullet shorter than .9" will shoot in a 1/38 given enough velocity.
Weight helps stabilize, it's the length you eventually run up against.
Check out the Miller stability formulae and play around with em.
https://en.wikipedia.org/wiki/Miller_twist_rule
Turns out the plastic tip does not really count towards the length when computing stability.
Miller has yet to include this but JBM does
http://www.jbmballistics.com/cgi-bin/jbmstab-5.1.cgi