Rep based on followers. An interesting approach. Though it could create the hunt for followers; Such hunt could go into: I give you Steem/US$/Euro and you follow me; ie Gaming the system. But I guess many technology only algorithm has its flaws and leave room for gaming. Also, do you think minnows will start de-following others, eg Whales, when such Whale is posting 'wrong' and 'bad' posts and comments, and when a Whale is flagging? Keep in mind, the flagging by some of the Whales are sometimes good, and sometimes not so good, it really depends how you look at things.

In the science world the number of references to your article establishes a kind of reputation. Google uses similar way of ranking search results. I know it is not in our culture on Steemit to reference articles, but maybe we can start doing so when we build article references into the rep formula.

I opt for a supervisor board with enforcement tools. I know this goes against many of our fellow Steemians, but I think any new system should take the good things from the old system and use it. Supervisor boards may not be always implemented in a correct way, and can also be corrupted, but I'm sure we can come up with a way of implementation that will work better than technology alone: Think of members being voted for by the community; Think of limited time for each member in the role as supervisor; Think of rotation of supervisors.

Some will say: this is a police force. My argument is: When that is a police force, then all the rules implemented by technology is also a police force. The only difference is that the enforcer is either human or technology. Such supervisor board could get tools to ban certain users, or to mute a certain user for some time, or whatever measurements we as a collective decide upon shall be taken to whatever misuse of the service, the community etc. Some of us think that we shall have all the freedom we want, but in a world where we live together with others we have to respect each other. When someone is not respectful anymore in such a way it is damaging the community we are part of, some measurements should be possible to take to prevent further damaging the community. To decide if someone is going to far, some judgement needs to happen and some enforcement shall follow. I do really see no harm whatsoever to involve humans in such process; I actually think it is better, since not all situations are the same, even under the same conditions. In the end we are humans, and humans are not robotic in the sense we are all alike. Any situation is also unique, some maybe similar, but rarely they are exactly the same. At this stage technology is not able to make the nuance required to determine what is good and what is damaging for the community.