You're right in that I don't disagree with most of your reasoning, but it's the entire assumption that "he's not as good against middle order" as other bowlers that is just not true.
In case you missed:
The entire discussion is based on a false statement
Look, my eyes told me starc was effective when the ball was swinging and not effective when it was not. I had no idea what the stats were. You posted some stats that supported this view so I commented on it.
You are also posting stats (which I did miss) for a different set of bowlers for different batting positions. You can prove anything you like through selective statistics like this. How about we don't quibble over a couple of % on stats like this, particularly since he is early career and his stats will change.
If we work with what we are seeing, I think we can agree that Starc is good with a swinging ball and worse with a ball that isn't swinging, and that the difference is more pronounced in him than bowlers with some combination of:
1.bowling seam in addition to or instead of swing
2 possessing a greater degree of variation, and
3. simply bowling a tighter line and length to nag a wicket and hold up an end until the ball starts to move.
Interesting stat on Steyn BTW. His middle order numbers being higher than Hazelwood and Philander don't surprise too much as they are mostly new ball bowlers, but I would not have expected him to have a higher % of middle order wickets than McGrath and Morkel. The percentage difference is too low to read anything into it, but it's interesting.