The solution is simple, of all the people I've told it too, only Kevin Wong seems to be on board. Indeed, I'm usually the person he's quoting, including in his comment above.

You increase curation rewards to 50/50 to close the gap between bid botting and curation.

That's probably not quite enough to change the status quo, so you also need to hand out a small amount of free downvotes. Roughly 2 downvotes daily vs 10 upvotes will likely suffice. That way, there is at least a viable threat to upvoting garbage content as the ones that stand out will more than likely be hammered down now that there's no cost to doing so within limits. This will further push the balance towards actually curating content that you think is appealing in terms of rewards.

Finally, you can't downvote what you can't detect. If someone splits their self voting up into thousands of different account, the cost to hunting them down is too high. Therefore, a small level of superlinear is necessary. Even something as simple as half rewards up until around 5 steem, then linear after that will do. That way, you might as well curate properly than spam micro votes as you're only getting half the rewards under a certain threshold. After that, it'll be much easier to detect and downvote with their limited free downvotes.

Increasing curation, allowing free downvotes, and having superlinear all have downsides. This is why excessive measures such as 90% curation, as much free downvotes as upvotes, and n^2 are bad ideas. The minimum amount of these measures should be used in tandem. Together, they compliment each other to plug all the leaks in this sinking ship at the lowest.

With the current economics, the platform is completely self defeating.