I think you might be right. Maybe a wait and see approach is best. It's still very early and not enough time is being given to the current algorithm to find out if it works or not when it scales. With enough people it might work better but with 10 people or 1000 people it might not work well.