You probably have to differentiate risk. I assume for some risks associated with smoking there is a linear correlation with exposure, where it's on/off for others, or e.g. following a logarithmic function.

Nicotine, CO and radioactive toxicity have a long half-life, where primary radical damage or reactive carcinogen burden is dose dependent stochastics.

Also people forget, you are much, much more likely to die from cardio-vascular damage or COPD as a consequence of smoking, than lung cancer.