I've seen and been involved in many discussions in the time I've been on the forum about Amsteel. In many of them the idea of knots comes up. The general consensus is that it's not a good idea but I wasn't seeing any hard data.
Today I was looking for something else and found this paper published by the Oregon Occupational Safety and Heath Administration. On page 17 (p22 of PDF) there is a table showing destructive testing data of several different knots. The numbers are worse than I ever recall anyone speculating. The worst knot tested was an improved (tucked) half blood knot which broke at 20% of catalog minimum. The best was a cow hitch (Lark's Head) which retained and average of 58%.
The take away point for me is that the average of their testing showed that Amsteel only retains and average of 32% catalog minimum breaking strength when knots are involved.
Ultimately everyone has to hang their own hammock. I just don't want to see anyone get hurt because they lack knowledge of the limitations of the equipment they are working with.