The idea is excellent. In this way, at least in theory, the vast community of Steam users can help bring informative and helpful user reviews to visibility, while bad and unhelpful reviews will sink to the bottom.
The problem is that people are not actually voting on whether the review was informative and helpful, but whether they agree with the review or not. I have noticed that it seems that if somebody disagrees with the "recommended / not recommended" verdict of the review, they are very likely to thumbs-down the review, completely regardless of its actual contents. In other words, even if the review is extremely informative and brings useful information to potential buyers of the game.
Not all users, of course, but too many. For example, I wrote a "not recommended" review of the
Since this seems to be a very popular visual novel, I got an unusual amount of feedback and votes. 61 in total. Of those 61 votes, only 9 "found this review helpful".
Of course I can't know the mindset of the other 52, but I'm willing to bet that the majority of them downvoted the review for the simple reason that they like the visual novel, and disagree with my "not recommended" verdict. That is, rather than seeing that hey, this may indeed be helpful information for potential buyers, who have no way of knowing the nature of the "game" before they buy it. (None of the descriptions on Steam make it clear that this is not an interactive game, but a pure digital comic.)
(One person wrote a comment on my review praising the fact that it gave this useful information, and lamented that people were downvoting it for all the wrong reasons. So at least some people understand the nature of such reviews, which is excellent. Too bad that so few people do.)
I have another example. While this one is arguably less unambiguous and clear, I think it's still illustrative, and also here I think that the majority of downvotes were simply because the voter disagreed with my verdict, rather than the usefulness of the review itself. Namely, I gave a negative review of the hugely popular game Shovel Knight (which I found needlessly hard, for no good reason, up to the point of sheer tediousness and frustration.) 2 out of 15 people "found this review helpful".
I think it's a pity that the "helpfulness" vote is in practice an "I disagree with the verdict" vote, especially when the review goes contrary to the majority opinion. It lessens the usefulness of the review system. I really think that one can appreciate whether a review is helpful to potential buyers even if you disagree with the final verdict. Unfortunately the vast majority of people don't seem to think like that.