The large scale crony voting risks literally turning Steem into a Ponzi scheme, as I've further explained here: https://steemit.com/steem/@troglodactyl/is-steem-a-ponzi-scheme

Partly this can be solved (or at least managed) by continually building consensus on definitions of abuse and what should be downvoted. In the long run, I think we'd be well served by a few protocol changes:

1. Downvotes should get curation rewards. Steem is based on the premise that most of the stakeholders will vote in the network's best interest to increase the value of their stake. This should be just as true regarding downvotes as upvotes, but currently dealing with spam and abuse is uncompensated. When a post or comment is downvoted below 0, the voters who downvoted it should be rewarded, just as upvoters are rewarded for curating good content.

2. There should be curation penalties for losing voting wars. Currently, shortsighted stakeholders see no incentive for avoiding abuse. They feel they have nothing to lose, because each of them think their own actions are too insignificant to impact the price of Steem. If there were curation penalties (negative curation rewards) for downvoting something that settled to net positive votes or upvoting something that settled to negative, then stakeholders who more often than not oppose what others consider best for the network would have reason to sell their stake and move on.

3. Increase the length of the vesting schedule. The level of abusive voting indicates that many voters don't have the longterm interests of the network in mind. The length of the vesting schedule should be gradually increased after #2 is implemented to give abusive stakeholders time to see what's happening and sell out gracefully.

4. Return to a non-linear reward curve. Linear rewards reduce the incentive to generate consensus on what is desirable and what is undesirable. If the R^2 curve is deemed excessive maybe we should go with R^1.5, but R^1 is a step in the wrong direction. Combined with #2, curation penalties and rewards should be calculated symmetrically on the same curve, with exponentially higher penalties and higher rewards for posts settling with stronger consensus.