Vote weighting is possible, but maybe it's too extreme yet blunt for the problem at hand. Vote weighting would affect all articles, but really there's a specific problem of popular-but-trashy articles that you want to target.
For example, you could add a penalty for negative-rated comments. A trashy article is more likely to cause the sort of argument that gets comments downrated.
Or you could try to classify the comment contents and organization to get a sense of how "argumentative" vs "quality" they are. Bayesian if you want. You could use article contents, too, if you had a scraper.
Either way the ranking algorithm will have to get more complicated as time goes on. Logically you should have more information with more users, so the quality of articles should actually get better. The problem with vote weighting is that by downweighting votes you are discarding this extra information rather than using it more cleverly.
For example, you could add a penalty for negative-rated comments. A trashy article is more likely to cause the sort of argument that gets comments downrated.
Or you could try to classify the comment contents and organization to get a sense of how "argumentative" vs "quality" they are. Bayesian if you want. You could use article contents, too, if you had a scraper.
Either way the ranking algorithm will have to get more complicated as time goes on. Logically you should have more information with more users, so the quality of articles should actually get better. The problem with vote weighting is that by downweighting votes you are discarding this extra information rather than using it more cleverly.