First of all, technically, integer is preferred (1, 2 or 3 for instance). Secondly, Sybil issue is not relevant because there is no difference between one big vote from an account or aggregated small votes from multiple accounts.
As @sigmajin well pointed out, the basic assumption of n^2 is equally distributed stakes, however this is not the case in the current Steem.
The only issue with linear system is self-voting. Regarding this, please see my comment above. IMO, self-voting is not a serious issue in the linear system.

>The second thing was the Sybil issue of people making multiple accounts. It is less profitable to spread your available steem power among multiple accounts than it is to concentrate it in one account, with n^2 voting.

No its not.  Under the current system, if you have 100 accounts with 10SP each, its exactly the same as having one account with 1000SP.  You gain no advantage and suffer no detriment.  

>I remember 2 other aspects from the White Paper. There IS a lottery effect from the current n^2 voting power formula, and it was a desired effect according to the White Paper. If you get a large payout once in a while, at an interval that you can't predict, then that plays on certain huann psychology which is also at play when we gamble.

I realize that, to a certain extent, i am trying to have my cake and eat it too by quoting the whitepaper in my post, then completely disagreeing with another aspect of it.

But i do and always have thought the lottery effect was a stupid idea (especially in the current system where the same people always win the lottery, but stupid generally.)  I think youll get many more interested engaged people by making a system where everyone gets something that approaches a fair share based on the value of their contributions.

>In fact, the exponent's value could be something the witnesses vote on, much like they currently vote on the SBD interest rate.

this is a super neat idea (though id have to think about it some before i came to any conclusions)