Okay, so since it's been brought up I suppose I should chime in with the clarification on exactly how the Model M buckling spring breaks down.
The thing to remember with a Model M is that the spring is a wearing part. The constant action does have an effect on the metal and does fatigue it over time. This fatigue results in a
significantly lighter action then original. (Which is why it still amazes me that these things are good for literally billions of keystrokes. Seriously. I have some with over 2 billion, still work fine.)
Out of the box - as in new - the Model M varies between 75cN and 80cN typical. Usually +-5cN to my recollection. For comparison, the Complicated White ALPS is stated as being 75cN +-25cN. Which is a horribly wide range. The Cherry MX Green is rated at 80cN +-10cN. So yes, the IBM Model M spring is in fact, by the numbers, the stiffest of the lot. It has a narrower range as well and much tighter tolerances.
But as I said: they wear down as a matter of course. This wear results in a lighter actuation feeling and less bounce-back. The end result is that your typical Model M with we'll say 2-3M keystrokes aggregate (all keys) is around 65-75cN on the springs. There's also the fact that the Model M has less sprung weight than a Cherry MX and a longer travel distance. In a Cherry MX switch you have multiple elements riding on top of and interacting with the spring, which is shorter and moves in a straight compression. So from that 80cN on an MX Green you need to subtract switch stem weight and keycap weight, presuming no stabilizer interaction.
The buckling spring does as the name says - it buckles sideways. It also has NO weight other than the keycap itself. There are no interfering elements and the striking element (the foot plate) is assisted by gravity in its quest to strike membrane. This means the spring interaction is much more direct, resulting in a stiffer feeling. The stabilization method also matters more than people realize - an ungreased Teflon slider (white type) versus a greased plastic slider (e.g. black and some blues) on an M feels significantly different. The post stabilization method also means that stabilized keys maintain a more direct spring interaction because the stabilizer does not interact with the spring in any fashion. If you take that post and stick it in a stabilizer slot it will just fall down. There's no upward action from the stabilizers.
So generally speaking when you're on an IBM Model M these days, you're looking at around 70cN direct interaction with no downward force and more direct interaction. On an 80cN Cherry MX you're actually probably looking at somewhere around 70cN as well after adjusting for sprung element weights. (... hm, I wonder if that's why Alps Electric rated the White as +-25cN: to derate for sprung weight. Might be.)
Boy, that turned out long.
