I'm currently writing a lesson on the twelve disciples that focuses on the significance of Jesus calling twelve disciples and the evidence preserved in the gospels (and Acts, Paul, and even Revelation). Of course, quite a bit of time deals with the different lists of disciples preserved in Matt 10.2–4||Mark 3.16–19||Luke 6.14–16||Acts 1.13. E. P. Sanders (Jesus and Judaism, 98ff.), of course, emphasizes the differences between the lists, while Bauckham (Jesus and the Eyewitnesses, 96ff.) minimizes these. Both accept that Jesus was remembered as surrounding himself with disciples, to whom he referred as the Twelve, but Bauckham's thesis requires that the Twelve refers to twelve specific, named individuals who could take (or be assigned) responsibility for the Jesus tradition.
The problem, unless I've missed something, is that the lists aren't identical. This is especially a problem for Bauckham. The differences aren't simply a matter of order; Matthew and Mark both name someone called Thaddaeus, while in both of his lists Luke mentions another Judas, this one the son of James. Critics have gone back and forth asking whether Thaddaeus was also known as Judas; I'm happy to notice that none of our texts suggest this. But there doesn't seem to be any redaction-critical motivations for changing (on the assumption of Markan [or Matthean] priority) Thaddaeus to Judas. In fact, I would be more willing to accept the opposite: That sometime after the first Easter, any remaining disciples of Jesus named Judas would have looked around for another moniker. For that matter, there doesn't seem to be any redaction-critical motivations that could explain the variations in order evident among the four lists.
For that reason, I think, both Sanders and Bauckham invoke oral tradition (or memory) to explain the variations for which we need to account. Sanders says, "In the earliest period (evidenced by 1 Cor. 15.5) noses were not counted. That it was some time before they started being counted is clear from the lists of names" (Jesus and Judaism, 101). Of course, there really is only one name from the Twelve about which there are any questions, so the extent to which the delay between an awareness of the Twelve and a list of who those twelve were is actually not very "clear" at all. Bauckham provides a more helpful theory:
It is quite intelligible that a list of this kind should be remembered as consisting of three groups, with the first name in each group a fixed point in the memory, but with the order of the other three names in each group variable. (Jesus and the Eyewitnesses, 98)
Admittedly, I'm not sure which source-critical assumptions, if any, Bauckham brings to his analyses. Sanders, if I'm not mistaken, is probably a neo-Griesbachian, but again I could be wrong. Regardless of which solution to the synoptic problem (a peculiarly literary problem, after all) we prefer, Matthew, Mark, and Luke (in some combination or another) should be copying from another, and the list of disciples would be a place that we would expect the most careful copying (if the names of the disciples weren't already indelibly stamped on the tradition). In other words, on the basis of some strikingly similar passages (e.g., Mark 2.1–12 parr.), we postulate that some evangelists copy from other gospels. But if, for example, Luke had Mark's list of disciples available and in front of him, why should he vary either the order of the disciples or, most significantly, substitute Judas the son of James for Thaddaeus?
So the list of disciples in the gospels, and their counterpart in Acts 1.13, suggest that the historical processes source critics have envisaged as descriptive for our gospels' composition are unlikely at least and anachronistic at worst. Even if Luke could read/had read Mark's gospel, and even if the latter was very influential over the former's composition, how do we explain the differences between their lists of apostles? Sanders can't be convincing here, because even if the precise names comprising the Twelve wasn't fixed in early Christian tradition, Luke (who replaces Thaddaeus with Judas son of James) should have known Mark's list!
So on standard source-critical readings, Luke knew his lists differed from Mark's and either didn't care or thought he was correcting Mark. And if he thought "Judas son of James" was another name for Thaddaeus, we would have expected him to tell us that this was his other name (see, for example, Acts 1.23; 13.9). We can't plausibly explain Lukan redaction of Mark here. And we can't assume (as do Sanders and Bauckham) that Luke wasn't familiar with Mark (unless we're willing to jettison source-critical assumptions completely, a move which I increasingly favor). The lists of disciples demonstrate rather trenchantly, I think, that our conception of the Jesus tradition as a textual phenomenon has misled us. Instead, we should appreciate that the Jesus tradition was a narrative world (or, in more sociological terms, a symbolic universe) enveloping, contextualizing, and giving meaning to the written texts. In this perspective, even if Luke had read Mark, Luke was more richly and dynamically rooted in the tradition itself rather than the textual expression of the tradition found in Mark's gospel.
[Update: I should have read further. Bauckham (Jesus and the Eyewitnesses, 98) does try to explain the variation in order, at least for the first four names, in redaction-critical terms. Notice how utterly lacking in explanatory power is Bauckham's argument. Why should Matthew and Luke, who keep both sets of brothers together, be more original? Why should Mark have any interest in binding together the three nicknamed disciples? And, most problematically, why should Luke rearrange the first four names in Acts 1.13 simply because Peter and John are prominent characters (3.1–4.31; 8.14-25) and James's death is narrated (12.2)? All of this suggests that we have misconceived what the Jesus tradition is and so missed the dynamics by which it was entextualized in the written gospels.]