Longer key travel is helpful for almost everyone (for accuracy, speed, comfort, and avoiding injury). Ideal is probably at least 1.5mm before actuation, and at least 1mm after actuation. Maybe more. No recent laptop keyboard comes anywhere close to a truly fast and accurate keyboard, because the space constraints are simply too tight to fit in a mechanism optimized for ergonomics. All the best keyswitches are from <1990. They had many small moving parts and were very expensive to produce relative to cheaper alternatives like rubber domes.
(Laptop keyboards also don’t have the space to properly raise the keycap tops on keys in further rows, because they don’t have space for it; this makes reaching for the top 2 rows of the keyboard noticeably slower and less comfortable)
The best keyswitches actuate about halfway through the stroke, have a distinct drop in force at the actuation point, and have a significant mechanical bounce to them on the upstroke, launching your fingers up toward the next keypress. My all-time favorites are Alps “plate spring” switches used on certain Japanese IBM computers and some IBM luggables in the late 1980s (I can type about 20–30 wpm faster on these than on any laptop keyboard). You can also get this effect (with a very different overall feel) from Topre switches (still produced), or from IBM beam spring keyboards (from the 1970s). Some other keyboard enthusiasts have different preferences, e.g. for simple springs with no additional tactile mechanism (“linear switches”) optionally with a clicker or buzzer to give audio feedback of key actuation. Such switches are slower to type on than a snappier alternative, but some people prefer the feeling.
Nobody should be hitting any type of keys “way too hard”. You want to use just enough force to actuate the key and not push the key hard all the way through the bottom of the stroke. This is much easier to judge when the switch has relatively dramatic tactile feedback and a lot of post-actuation travel which ramps up in resistance toward the bottom.
What you’re describing are literally the most horrible keyboards I have the displeasure to use.
For example: the IBM Model M is one of the single worst typing experiences imo. The only way to type on that thing is use about 5x the force I normally do.
No wonder our industry is plagued by RSI - it lionizes horrible things like this due to misguided nostalgia.
The IBM Model M is a great typing keyboard after you get used to it, but you are right that it is on the stiffer side. The previous Model F keyboards took about 10 grams less to actuate, and are widely preferred. I find even those a bit stiffer than my taste. Neither of those keyboards has my preferred feedback or force curve, though I don't mind typing on them. The biggest objection I've ever heard to either is the noise.
In any event, actuation force is an orthogonal concern to travel distance, tactile feedback, post-actuation force curve, upstroke bounce, keycap shape, etc.
Relatively fewer people typing on a such IBM keyboards got RSI compared to other keyboards, because the extreme tactile snap makes it very obviously unnecessary to type with much force more than barely required to actuate them.
But the cause of RSI is largely down to posture and typing style more than keyswitch design, though the latter can contribute (the worst are cheap rubber dome boards from the 90s-2000s, which often force a hard smashing typing style). We would have much less RSI if computer users kept their wrists straighter, pulled the keyboard closer, tilted it parallel to their forearms, and stopped slouching as much.
Large key travel is only good for those hitting they keys way too hard imo