I, myself, have written about this in the past (almost 2 years ago now, terrifyingly) – but succinctly, the problem with "Proof of Brain" is that it isn't.

We have a vast amount of experential proof which shows us empirically that the articles which get the most votes are in no way the most intelligent, the most creative, or the most immediately valuable except as representations of a bet. The Steem blockchain is architected as a betting pool, not as a mechanism of content evaluation.

When you place an upvote, that does not communicate to the system "I think this is a good piece of content." If it did, the system would then go out of its way to find more stuff like that and put it in front of you because that's the sort of thing you like. Instead, you are signaling, "I think that other people are going to vote this up." If you are wrong, you lose the potential value of that voting power. If you are right, you are rewarded with resources that can increase your voting power in the future. As your voting power increases, you become one of the people who everyone else is laying bets will upvote certain pieces of content.

Note that the actual content, it's quality, or its nature itself is not mentioned in this description – because it simply doesn't matter. It literally doesn't matter. The only thing that matters is whether or not you can predict what most other people are going to do and do it slightly before they do.

What happens in that kind of environment? Well, exactly what we see on Steem. There is a general collection of one specific type of content which is most likely to get voted up. Because people are lazy and masturbation is obvious, on the Steem blockchain the content most likely to get voted up is content about the Steem blockchain. Everyone knows that talking about the platform is more likely to get activity, focus, and votes than any other type of content. As a result, as a creator, that is exactly what kind of content you try to make in order to attract the attention of the implicit vote value.

The more of it that gets created, the more dilute the particular type becomes, the less likely any particular creator is likely to give voted up, and you see most the votes going toward a very small set (comparatively) of creators, because that is where the votes have always happened in the past.

No brain necessary.

This summarizes the first problem with "Proof of Brain" as an assertion.

The second problem is worse:

There is no "wisdom of crowds." It ain't real. It's never been real. Or at least as it is presented in places like the Steem blue paper and popular culture, it's not real.

The reason that it's not real is an essential misapprehension of understanding about the nature of markets. Markets are very good at finding and expressing the most efficient product or products in a niche. However they are terrible at finding the "best" product in a niche in particularly bad at finding the "best for me" product in a niche.

For a market to be properly communicative, I need to be able to view the output of "people like me." I need to be able to slice through the entirety of everything being freely traded and just look at things important to me. The market as a whole needs to be large enough to support a variety of things such that I can find something that appeals to me, even if it is not what "everyone else" wants. This is especially and particularly important when it comes to intellectual goods and creator generated content.

This is exactly what the Steem blockchain does not do. In fact, it goes out of its way and has always done so, to make it difficult for me to find things that appeal to me, and it really doesn't care about me as an individual or the signals that I'm sending. All he cares about is the pool of active voting power as a singular whole.

Now, there are arguments that the upcoming Community system is going to provide part of the architecture which will allow us to segment all content into things that we are interested in, but look at the points of friction which are defined in engaging with Communities. They actually cost "money" to set up. They cost "money" to open up slots for people to be a member of the community. They seem to be tailor-made to push SMTs and not to actually add useful functionality for the actual use case for social media platform.

Again, they don't care about me (or you) as a user, or as a consumer, or as someone who wants to find things that appeal to them – which is the whole point that any of us are here for.

An essential disconnection at such a fundamental level, both on Proof of Brain requiring none and the Wisdom of Crowds simply not existing suggest to me that, all while your suggestions for changes are well-intentioned, they would be doomed to affective failure. The intention is clearly to make the signals which the user throws to engage with content more individual and less connected with the ultimate betting pool nature of what's going on behind the scenes. You want to build an actual, functional, cut down Reddit-like on top of the Steem blockchain as the backing database.

Unfortunately, that ship has sailed. We are probably a good three years beyond the point at which that would have been effective. And it's sad, but it is true.

The answer to your ultimate question is relatively simple. Because nothing ultimately matters at a content level and the underlying betting pool mechanism is inherent to the architecture, there is no chance in Hell that downvotes are going away and unless something magical happens and someone unrelated to Steemit Inc. figures out something worth selling using the Steem blockchain, it's going to remain just an abstract representation of some kind of value which continues to dwindle.

It's a harsh world.