@ Jim,

I just purchased a bunch of literature to see which books you guys are actually talking about. One of them was the Sobon book of historic american timber framing which is where I also saw this drawing. ( Another reason was to compare Benson and Sobon since this comes up fairly often and I had no idea what any of this meant. Having looked at both I would definitly fall into the Benson category).

Back to the topic. Your/ Sobons explanation makes sense, at least parts of it. If you applied a load to the beam in the drawing it would indeed sag where it is not supported. But it does not actually go anywhere (neither up or down) above the post. It would crown up once you add the braces, in theory. But since everything is pegged together it won't. So if you have a full lenght beam going across the three posts shown and it sags midway between the posts it would put tension on the beam fibers above the post because it needs to "stretch/ become longer" to make up for the sag. Same would happen when the joint is put above the brace. Because the brace does not actually give. So it really supports the joint and holds it up as the post would. So in my opinion it should not really matter where you put the scarf joint. If in doubt I would add another post.

But Sobons drawing aside. Does anyone know of a lab test where this situation has been tested. There are test for pegged tenons and braces but I did not come across a test for the scarf joint.