I believe we already have this measure in place, via the help center:
You lose reputation when:
We can not expect new users to read the help center, but they should; the main articles, at least. We can all admit that we would have a lot less cleanup regarding new posts if they did. The above mentioned is one of these main articles.
At the end of the day, I feel that an additional warning in the review queue has some good merit to it; but I see other problems with this feature, more specifically, the extent of the effect it will actually have.
- If we put up an additional warning, who are we actually targeting? First time reviewers who are literally on the borderline for having that privilege. I will leave this to someone else to gather analytics regarding how frequently this actually happens; I feel safe speculating that it does not happen very much, at all.
- Any additional warnings will not bear any effect if the user does not actually read the warning; we have to take that into consideration. We can not assume one way or the other, but we already know that in the few cases where such a warning would assist the user, they have already chosen against reading about our basic practices in the help center.
- I know that some of our sister sites move a lot quicker than others, in regards to reputation gains. However, in my experience, by the time you reach the minimum reputation requirement to partake in reviewing you should already have enough practice to understand that when you click the down-vote, you lose a point of reputation.
- As you have already pointed out, the few times where this user case would lead to user frustration, it is literally the matter of a single reputation point. Assuming the user continues to conduct themselves appropriately, they should be able to reclaim that reputation point fairly quickly.
- Implementation could be tricky; A clear warning on the screen, itself, pushes towards interface overload. It's like having all of the buttons visible at any given time; it looks bad, and it lessens the user experience. The only other implementation I could think of would be a pop-up, but then we have the additional question of will users pay attention long enough to notice the pop-up, and if they do, will they go back and check it or gloss over it immediately? I feel it is uncommon to expect users to hover their mouse over the downvote button for an extended period, for example. If I choose to vote on a question or answer, my mouse hovers over the button for such a split second that it took me a very long time to even notice the current pop-ups that display.
As such, I vote against this proposition. Not because it is a bad idea, but because it is not feasible in comparison to how it would affect typical users, the workload, and the effect on the actual interface.