Thanks for your point of view.
Actually I think the bigger problem than the linear reward curve is the higher impact of fewer votes. Some people are now writing exactly 10 short articles per day (as they have 10 heavy 100 % votes available) just to upvote them all with 100 %. This is an example:
https://steemit.com/@sandrino

Apart from that I think the fast growing number of Steemit users would require the exact opposite of a few high impact votes (whereby big accounts tend to reserve their fewer but heavier votes for other 'big fishes' because of the expected heavy upvotes in return).
Instead of that many small votes for many different users would be a much better idea ...