What does author position say about credit?

There's been a lot of discussion online lately about the relative importance of the position of an author name. Is it more impressive to be a first author on a report? If so, how much? John Lynch made a graph of Guillermo Gonzalez's publication record as a way of illustrating his argument that Gonzalez didn't deserve tenure. But there's a twist to the graph: it not only indicates articles on which Gonzales was an author, but also articles on which he was first author.

As the average number of authors in journal articles increases, does that mean that the contribution of individual authors is valued less? Brain in a Vat and Nautilus comment on a recent article by a team led by JD Wren, which found that for medical journals, the first and last author positions were both highly valued by department chairs, but for different reasons.

In general, they found that the last author is seen as having a supervisory role, while the first author is regarded as the one who did most of the work. The researchers were also careful to control for the position of the corresponding author. When the middle author of a five-author paper was listed as the corresponding author, then survey respondents rated the supervisory role of the last author lower, and the middle author's importance rose.

So what about all those other author positions--2 through 5 in a six-author paper? Are they seen as valuable at all? Here's how Brain in a Vat describes the situation:

An even more concerning trend is "author inflation." Wren et al. asked survey respondents about the prevalence of honorary authorships, in which additional researchers are included in the byline merely to pad their resumes. 40% of respondents indicated that this practice was commonplace, and 50% reported that this practice made it harder to judge who deserved promotion.

Indeed, the average amount of credit respondents gave to the fourth of five authors was less than ten percent for "initial conception," "work performed," and "supervision."

Also take a look at Brain in a Vat for an interesting graph showing the trend in number of authors on papers between 1966 and 2006. In 1966, the most common number of authors on a paper was one. By 2006, the mode for number of authors had risen to 3-4.