Personally I think authors should get residual income, otherwise it's bullshit (just my honest opinion). Plus it can be a competition vector to say "hey, we made something like steemit, but we improve on it by also paying residual". I wouldn't leave that door open.

Now, I've gone over the github discussion. Two points

a) RAM issues. I don't understand how it is possible to have a blockchain sitting at 2gb and this requiring ...7gb of ram to handle it. Something is messed up in the software (?) and it should be fixable.

b) Absence of any mention of reward-scaling problem. What I mean by that is that the major hurdle of Steem is how to scale rewards along with the userbase. If we want to have, say, similar rewards to what we currently have, then that means 1000x userbase (70mn) requires 1000x marketcap (200bn USD). Now this is extremely difficult to happen because the explosive userbase growth cannot be shadowed by a similar marketcap increase (in all probability). So what we have instead is, say, a 1000x userbase and a 10-20-50x in marketcap, which effectively slashes the avg reward per person by 100/50/20. This problem is very severe and actually punishes Steemit for becoming successful due to the diminishing rewards. If you then start paying the old content, which as time progresses forward will dwarf the new content, then the payouts for new content would be so tiny that nobody would even bother. So there would have to be some type of percentage allocated (like 10-20-30% for old content) so as not to entirely drain the rewards for new posts. But even at 70-90% of the reward pool allocated for new posts, the problem will be severe if the userbase goes up significantly.