[W]e must reckon that the later Matthew knew the earlier Luke, took over parts which seemed to him appropriate, i.e. the content of which was promising, and in the process of course also altered his theological wishes accordingly. This is already indicated by the fact that as a rule the more original version is attributed to Q-Luke as opposed to Q-Matthew. In addition - and here Papias can put us on the right road - there were certainly also one or more 'Logia collections'. But - as I have already said - we can no longer reconstruct these adequately, especially as we cannot know what the evangelists changed or omitted in the sources, unknown to us, which they probably had in more abundance than we suppose."I think Hengel's relatively early date for Luke is in part a consequence of the stance he earlier developed on the historicity of Acts (e.g. in Paul from Damascus to Antioch) where he shows impatience for those with more radical perspectives. But the brief statement above is actually misleading. It is not the case that "as a rule", Luke is more primitive in double tradition material. If you work through the Critical Edition of Q and the volumes so far completed of Documenta Q, it's pretty much 50/50 in terms of alternating primitivity, hence the term alternating primitivity. I hate to say it of so great a scholar, and one so recently departed, but I don't see any evidence in Hengel's Gospels work that he has worked through the issues connected with the Synoptic Problem carefully.
James McGrath has an enjoyable and characteristically lively response to my post in Mark's Kung-Q vs. The Bolt from the Johannine Blue. With respect to my point about Matthean language, rhythm and imagery occurring in the double tradition material, James notes that "the bolt from the Johannine blue" (Matt. 11:21 // Luke 10:22) features language remarkably reminiscent of John in a passage that by most accounts pre-dates John. But within the context of discussing direction of dependence in material that is near verbatim identical (Matt. 3:7 // Luke 3:7; Matt. 3:10 // Luke 3:10), I was attempting to make the point that the direction of dependence is more likely to flow from Matthew to Luke than from Luke to Matthew.
What we have in the bolt from the Johannine blue is language that is similar to Johannine language, not language that is similar in a passage where there is verbatim agreement. It is an interesting point that this could be the seed out of which the Johannine discourses grow, but if so, what this gives us is an insight into the matrix of early Christian thought and its evolution. There is no verbatim agreement here between Matthew // Luke and John. In other words, no one seriously doubts that there is a literary link of some kind between Matthew and Luke at 3:7 (etc.), whether directly or mediated via Q. The question, once that literary link has been established, is where the direction of dependence goes.
With respect to my argument about editorial fatigue, James notes that Luke could simply be fatigued with Q. Two responses here. (1) I was responding to James's question "Did Matthew use Luke?" so pointing to Q effectively concedes my point. (2) Of course it could be that Luke was fatigued with Q; here is how I dealt with the issue in the article:
. . . The best way, therefore, to seek an answer to the question will be to bear in mind that if the Two Source Theory is correct, one will expect to see not only Luke but also Matthew showing signs of fatigue in double tradition material. Those who believe in the existence of Q will have to look for their own examples of editorial fatigue in Matthew's versions of double tradition material. I have looked for examples and cannot find any. On the Q theory it does strain plausibility that Luke should often show fatigue in double tradition material and that Matthew should never do so, especially  given Matthew's clearly observable tendency to become fatigued in his editing of Mark. (57-8).The Synoptic Problem is fascinating, isn't it? Can't imagine why so many apparently find it so dull!