Voters who don't read post, don't get any satisfactions, but just upvote provide any information about content itself.
Imagine that I'm a fiction writer, and I want steemit to be a more welcoming place for other fiction writers, so I write a bot to upvote every post in the fiction category, whether I read it or not. Maybe I put in some basic checks for grammar and image files and length of article. Over time, I begin refining my bot to become more sophisticated about the content that it votes for. Even though I didn't read those articles, I still derive satisfaction from those votes. I'm also representing the views of some other group of non-voters, who would be attracted to the platform (if my votes prevail). It's the curation rewards that tell me (and steemit) how much of the community shares my interest (leaving aside the [n2] distortion). It's also the curation rewards that enable the whales (major longer-term stakeholders) to shape the platform content. Anyway, even if curation rewards didn't provide those services, shouldn't I get rewarded for my acts of representation?
However, I expect their dissatisfaction from voting bots (e.g. concentration of votes on same authors) will be reduced and the overall they have more benefit.
pointed out, the bandwagon voters don't actually influence rewards distribution anyway. And what are you going to do when you take away curation rewards and people continue to vote in ways that you don't like?And there is the crux of it. You don't like the way people are voting, so you want to force them to vote differently or stop voting. Even though, as @sigmajin already
If you want to more understand my statement, please look at the number of votes and number of views on this post.
Those votes don't count. They're just here for the profit. ; -)
Edit: To put it simply, using this framework, your proposal might decrease Type-I voting errors, but it would do it at the cost of a huge increase in Type-II voting errors. IMO, that trade-off is contrary to the health of the platform.
Your framework borrowed from statistics is interesting, and thanks for your example. But please take my point: the issue is mixture of motivation. If you want to upvote every post in fiction tag, that's not because you want to maximize your curation reward but because you have some satisfaction from doing that.
@sigmajin's post is interesting too. But if no curation reward exist, the $160 might go to other posts by for-content curators. So they already meaningfully influence reward distribution, as well as voting counts.
The problems I perceive are, our filtration system for good contents are distorted by hybrid motivation. At least 500 bots (clues from votes on @steemvoter's posts. Maybe less but still hundreds I think) are round and casting votes for "profit", not for "quality of contents".