The posts would automatically be structured with a reputation system like reddit, promoting interaction.

Scientists could follow interesting people in their field, and automatically get suggestions on which other scientists are writing about the same things. Promoting sparking ideas and start of new interconnected collaborations.

The profiles of a scientist would show how helpful their work is - what other scientists consider of their ideas and work, and which ideas that came from them originally, and opinions on their accepted publications (which are still a core part in Open Science). Only using number of quotations is slow and does not tell the whole story.