I agree that the current delegate pay system may not be ideal but imo the new approach seems to be a lot worse.
1. I still don't understand how companys would be incentivized to hire core devs.
2. How are you supposed to raise money with refferals before there is a finished product?
3. The stakeholders give up direct control of what gets funded. There are most likely projects which are very beneficial to fund for BitShares as whole but can not be sufficiently monetized with referrals. (similar to point 1, but more general) Said differently, refferals aren't a perfect metric for overall userfulness. It is also about pleasing long time users, security and overall service quality.
4. This depends on the implementation of the referral system but I see the following danger: As long as the ecosystem expands everything is fine but once the stream of new users stagnates the whole system collapses.
I remember Bytemaster said in the hangout something that the people making transactions should be the ones paying for the service and not stakeholders. Ultimately this is right, but I do think that stakeholders have a certain buffer role. There are times where expenses are going to exceed income, and yet it is necessary to spend those funds. Currently we are in this situation. If we had to cover all delegate pay with TX fees, they would be so high that no one would use BTS anyway.
I may be alone with this opinion but I still think that efficient, sustainable funding through dilution is a great idea and gives us an advantage over all the other crypto projects out there. For now growth is far more important than some percent stake. And once BitShares reaches a substantial size, and tx fees get adjusted we might be able to become deflationary.