In addition to poor design, there is simply specific design philosophy. Much of the trouble with balancing was a product of the Drag Freight era when *nobody* was trying to move freight quickly. Their ideal was to move every car in the yard with one engine. It moved slowly, but it got there. Exceeding 30 mph on these trains was a rarity.
One example: the Frisco spot-series 2-10-2s. They were ponderous and massive, and the yardmasters emptied the yards on their drawbars. They weren't great steamers, but they did the plodding drag jobs just fine. BUT: every time they exceeded 35 mph, they twisted or broke rails. The Operating Department loved 'em; the Track Department loathed 'em.
By the '40s, all but two had been rebuilt into modern, high-speed Mountains. Those two, #19 & #40 lingered on as the primary power on the Ft. Leonard Wood branch in Missouri, and they saw heavy use because of high wartime traffic in and out of the base. Unfortunately, the demands of the traffic caused the engineers to push the speed limits. One of them--#40, IIRC--kicked the rails out from under herself on a curve atop a tall, timber trestle over the Little Piney River. The track gang had to go out on that trestle and jack up the engine and relay the track under her while suspended in mid-air. Not long after, the two drag-era orphans ended up the nearest deadline, probably because the Army failed to see the humor in the situation.
The problem here was not really one of bad design but of using the wrong tool for the job. An engine designed for one job fared poorly trying to do another.
Edited to add: Source Collias, Joe. Frisco Power.