we are talking about the choice between two agents/proxies one of which has a closer value system and the other is more effective at achieving his goals.
If the two environmentalists had a debate about this subject, each could start the debate by saying they want to do whatever is best for the environment. And then each could present a series of facts suggesting that nuclear power either is or is not good for the environment--a factual disagreement about what the right instrumental goal is for achieving the terminal goal of helping the environment.
If you think anti-nuclear environmentalists possess lack of nuclear plants as a terminal value, imagine what would happen if one was convinced of the factual belief that nuclear power is actually good for the environment. If your model is correct, we can imagine that they would continue to be anti-nuclear environmentalists because that's their terminal goal (while acknowledging that nuclear power is actually the best option for the environment). But we have counterexamples like Stewart Brand who switched from anti-nuclear to pro-nuclear after doing research & having their beliefs change.
Actual human beings' goals don't divide neatly into instrumental and terminal, and actual humans can be inconsistent. So you can have someone who has instrumental goals (that can be changed with evidence showing that they don't meet a terminal goal), terminal goals (which cannot), and inbetween goals like nuclear power that are harder to change than the former category, but easier to change than the latter.
I just saw that Donald Trump is running for president. Which led me to the following thought: would any of the big names in tech have a chance at being elected president of the US? Elon Musk? Sergey Brin? Jeff Bezos? Reid Hoffman? Peter Thiel? Edit: Bill Gates?
Some follow up questions/thoughts:
Edit: hypothetically, if one of these big-name tech people were to try to gain political power, how should they go about doing so?