If the "normal" rate of lung cancer is X, the observed rate in nonsmokers who get secondhand smoke is X+Y, and the observed rate in smokers is X+Y+Z, if you compare nonsmokers and smokers it looks like smoking increases your rate by Z when it's actually Y+Z.

No, that's where you're wrong.

You are only interested in that independent risk.

I, and many others, are interested in how much smoking changes that risk.

Picking random numbers, let's say smoking gives you a 10% chance of lung cancer. It's fine for you to only care about that 10% number, you get to care about what you want to.

But for the rest of us, when making informed decisions based on risk, it matters whether smoking changes it from 9.9% to 10%, or 0.1% to 10%.