I actually agree. I should have said “just a habit.” Calling heroin a “habit” is putting it in a class which includes doodling, tapping your foot, talking a p

I actually agree. I should have said “just a habit.” Calling heroin a “habit” is putting it in a class which includes doodling, tapping your foot, talking a particular route to work, or putting your jacket on the same way every time. In other words, it is technically a member of a class but very different than the median member of that class. This is why naive use of centroid clustering to reduce dimensionality and distance quantization leads to bad results. I think Johnny Thunders would agree.