Regression Analysis, Dummy Variable with 0, 0.5 and 1?

Dear all!

I have an urgent question for a paper I have to write in econometrics - I would be extremely grateful for your help!

The question revolves around the regression equation
anyevent = β0 +β1*smoker +β2*diabetes +β3*age +β4*gender +u

where...
...smoker is a variable that can take the values 0 (smoker), 0.5 (former smoker) and 1(non-smoker)
...diabates and gender are dummy variable that can take the values 0 (diabetes / man) and 1 (no diabetes / woman)
...age is measures in years

I am confused about the smoker variable, I don't know how to interpret a dummy variable that can take 3 values?
We are now asked to look at the definition of "smoker" and state what assumption we are implicitly making when inserting the variable into the model as above. Then we are supposed to estimate a regression that does not make this assumption.

Does anybody have any ides, or could help me understand the interpretation of the smoker variable in general?

Thank you a lot!!