I think filters might work. I'm. Not tech savvy, but if there is a way to allow for just votes that were done through the website to be displayed on the site (thus, reducing hiding the bulk of auto upvotes people get from readers who aren't interested in science), it might go a long way to help.
That way, only votes from people who are interested in the post counts..
It's really dicey though, because if I were a casual reader and i found something I'm interested in, I probably could just vote because I found it meaningful anyways, I don't have to use StemQ... But then again, without something like this in place, it would always seem like the more popular people are better at science than others (as will be reflected on the votes they accumulate).