This is an interesting idea – but there is an inherent problem with it.

It's not a bid. It has nothing to do with bidding. Bidding is the process of transferring some sort of value from a pool you control to a pool someone else controls as a wager on the outcome of another event. While I have repeatedly described the functionality of the Steem blockchain as "a bidding game," the bids are not made via of the proximal mechanisms of votes in that sense. I would say that the underlying point is that they are implicitly relative.

That's not to say that you aren't on to an important facet of the failure of the design by pointing out that it is repeatedly sold as the idea of "this content is not worth more/is worth more than X value" while the vote itself doesn't actually communicate that at all, not least reason being that "X value" is not a stable expression within the context of valuation of a post. (After all, the actual reward value of a given post isn't the only thing that changes – the underlying value of the token can change rather aggressively from moment to moment, so just saying "I don't think this is worth more than $10" doesn't communicate enough in the context of mechanisms available on the blockchain.)

What you're talking about isn't bidding, however. It's a far more elemental part of communicating value – pricing.

What you're talking about is explicitly telling the system what you think the value of a post should be. And then your idea is that the system should mechanistically apply your voting power to push the value toward that desired price. Which is an interesting idea, except for the fact that in actual use it would be pretty cumbersome. Rather than simply signaling to apply a portion of your voting power to changing the value as you saw it of a given post, it would require you to specify a value for every post you see, and that kind of cognitive overhead and mechanical overhead just leads to people not engaging with content.

Imagine reading Reddit and instead of upvoting or downvoting content, you had to put in a desired/intended number of total votes. You just wouldn't engage with the system because it's too much mechanical overhead to deal with.

In a system in which you were only expected to interact with 10 or 15 pieces of content a day, and assigning a value to half of those, it could work. But that is not how social media consumption happens, and because of that it's not really a feasible design. Interesting, but not feasible.

At this point, I maintain that there is already too much cognitive overhead when dealing with upvotes and downvotes on the blockchain as it stands. "Is this currently under or overvalued?" "How much of my voting power should I use?" "Is it within seven days of creation?" "How much of my voting power do I have left for the day?" If I were going to re-work the voting system, part of the core of doing so would be to remove most of these questions entirely. Simply keep a sliding pool of voting power, let people vote for as many or as few things as they feel like, and divide that voting power equally over everything that they have voted for, up or down, over that particular window. Oh, yes, and eliminate the seven day voting requirement and reward content if it had a vote during that cycle. Done. Massively less complicated and much easier to deal with.

Unfortunately, we will never see this.