Thought experiment:
Someone has 1000 SP. They vote for max number of witnesses. The witnesses each receive a 1000 SP vote.
Someone has 1000 SP. They split it amongst two accounts. 500 SP for one, 500 SP for the other. Each account votes for max number of witnesses. The witnesses each receive 1000 SP of votes.
Sybil attack is not possible with witness voting due to the "proof of stake" part of DPoS.
Here's an example of how extant 30x multiplication of stake weight inflates the value of substantial stakeholders in governance. User A has 1M Steem. User B has 100 Steem. User A votes 30x for witnesses, and casts 1M votes for each. User B does also, casting 100 votes for each.
In total user A casts 30M votes. User B casts 3000. The difference between the fiat value of their respective stakes is 999,900. The difference between the weight of their votes effecting governance of the blockchain is 29,997,000. This dramatically multiplies the impact the founder's stake has on governance, and contributes to Tron's ability to control consensus witnesses. If Tron is able to vote with 75M Steem for witnesses, it is able to cast 2.25B votes.
Each Steem should be counted once in governance/witness elections, to provide equitable weight to all stakeholders in governance of the blockchain. The extant voting method has now proven to be an immense security risk to the blockchain, whether you call it a Sybil attack or not.
To be clear, I agree with the idea of 1 SP 1 vote.
And I'm saying it can't be attacked vial Sybil attack.
So I'm saying it's a good idea that is maybe more secure than some think.