A more even distribution of delegations across validators will provide a more healthy ecosystem in the long-run. This is not to say that the existing structure is unhealthy, but rather to say that it has resulted in some unhealthy practices including 0% commission fees.
I have looked through some forum posts to try to aggregate views and see if there are any people think are useful and could do the job.
Raise the commission floor that validators can charge above 0% (suggested 5%)
By creating a minimum commission that generates income for validators, more parties will be willing to run a node as they are able to cover their costs. By increasing the validator population, it is expected that the distribution of SCRT will equalise over time. However, there is an issue here that validators may propose their own refunds to delegators to retain price discrimination in the market.
Greater levels of slashing for larger validators
Validators that have a larger fraction of the total voting power are slashed at a higher level. For example, a validator that has 1% of the network could be slashed 5% but a validator with 10% might get slashed at 10%. This will disincentivise delegators from delegating to this validator but it will also have a greater cost to validators for operating. This will be too complicated to implement without allowing validators to stop accepting delegators as if an individual still stakes, delegators who have already staked will now have an additional cost without having done anything. If implemented at a standard level, it will very easily result in existing validators splitting nodes.
Variable bonding and un-bonding times that will have longer lead times for larger validators.
The higher the proportion of total SCRT delegated to a validator, the longer it will take for a new individual to delegate SCRT to a validator. This will reduce the return to a delegator in the short-term but will tend towards the natural APR in the long-run as the amount of SCRT missed is a smaller percentage of SCRT gained. A similar mechanism will occur on the way out with longer un-staking periods for delegators. The solution here passes the cost on to delegators if they choose to stake with large validators and is soft enough that validators won’t be incentivised to split nodes.
It appears that a good solution needs to incentivise delegators to stake to smaller validators but it cannot do it by attacking validators. The third option seems relatively healthy but it would be great to see views on these options and other views that you know of too!