What makes a paper mind-alteringly good?

Sorry about the overly-dramatic title, but after reading what I think is a very well planned and executed paper on the phylogenetic and geographical dispersion of drought tolerance in plants earlier, I have been thinking about the power that single articles can have over our outlook on our work.

If the world's journal archives were about to be obliterated, the one article I would grab would be Davidsson and Janssens' 2006 paper in Nature on how climate-induced changes in decomposition may feed back. Actually, this is the only paper I kept from my undergraduate, and I have been reading the same dog-eared, scribbled-on copy for the past five years. It is also the only paper for which I have gone through and read every paper cited in it.

But why did this article appeal to me initially? I didn't understand most of it the first time I read it. Or the second. It took a week of staring at Box 1 to understand what it was talking about, and some of the other arguments in the paper seemed flawed to me because the lines of logic they were following weren't laid out, and disagreed with the facts I was aware of the time. But from what I could decode, I knew this paper would be really important for my understanding of the carbon cycle under climate change. And in my young inexperienced state, I thought that it must be good and right, because it was published in Nature.

Despite my interest in the paper's important topic, it was the dense challenge of the paper, working through the hidden complexities of the carbon cycle, that got me. Every time I read it, I get something new out of it, and it helps me frame my work in the bigger picture and remind my why I love my work.

But what article do you keep returning to?

Basic psychology would tell us that for most of us, our most powerful article will be one of the first we read on a topic; our experiences early in life (whether research or real) shape how we perceive subsequent events, and therefore we will find ourselves returning to the point (paper) which established our mindsets. I would think that review articles would also be favored over primary research articles, because they put the research in context and are generally written by people with respected views. Of course, regurgitating what is known doesn't help, but putting a new spin we hadn't thought of (for example pulling in information from other disciplines) would make an influential article in my books.

I think this would mean that journals that want to be cited lots should favor interdisciplinary reviews. I believe I read somewhere that this already happens - does it? That seems like it would be a much too simple key to "success"!