You'd presumably plan to do that so long as the Republican was in first or second place, but if polling started to show the Republican candidate in third place, you'd want to switch the Democratic candidate's score down to 0.

In the end, range voting boils down to approval voting, but with a trick to penalize people who are bad at math; and approval voting itself penalizes people who don't closely follow election polling.

On the other hand, I'm not sure that voting weights based on mathematical aptitude and knowledge of current events are necessarily bad things, and even if they were they're still probably not nearly as bad as the hysteresis effects of plurality voting.