I've just had the most wonderful day. I spent the entire time notwriting papers. (The only thing I like more than not writing papers is not writing grants and not going to staff meetings.) I cleaned my desk, collated some data, fed my cells (I love to watch them wag their little tails while they eat), reblotted that western in the vain hope that something would appear with this new antibody (twelfth time lucky) and learned to dance the meringue - merengue. I also made a meringue.
Sigh. No, I didn't do any of those things today. I'm lying. But here's what's so great: in science this sort of lying is completely okay! Not only that, we're actually encouraged to lie, as long as it's a certain sort of lie. But let's be clear about what I mean about lying. The following discourse contains the verb `lie' in the sense of `dissemble, deceive, prevaricate,falsify, fudge, fake, flummox, fib'. It also contains the noun `lie' in the sense of `a falsehood, a prevarication, a fib, a deception, a fabrication'. However, in no way should any of these statements be taken as endorsements of lying about, falsifying, fudging, fixing, fabricating or inventing data. Anyone who invents data is beneath contempt, which is to say lower than a Mole's belly. So what sort of lie am I talking about? The thing is, all of the things I mentioned above are things that I've done (even dancing the merengue and cooking a meringue). I just didn't do them today or all in a day or in that order. But I did them. That's why it's an okay lie. And it's exactly this sort of lie that is the secret of writing papers.
O rapture, you say. The Mole is going to give us his secret of writing papers. And you thought I was just going to complain. But no, I'm not going to give this to you, for free (“unless you've actually subscribedto this journal,” he said incredulously), until we reach an understanding about lying. And before we do that, we have to talk a bit about writing papers.
Let's set the scene. Several months ago you got a new antibody in the mail,and when you used it an experiment you'd been trying suddenly worked. So you went over your notebook and pulled out experiments that sort of worked or wanted to work. Some experiments do just want to work but don't know how, poor things. And, since then, you've been repeating them,changing variables and trying new approaches. And some of it is actually working – and so it's time to think about writing a paper. And this is really all about destroying Myth and Wonder. And then lying about why.
Here's what you do. You put the data into presentable form (not publication form yet, but something you can look at and make sense of). And you start to arrange it this way or that. Until, and here's a secret, a story starts to emerge. At this point it is only a very rough draft, so that you can see all the huge holes. Yes, you really do need that control experiment. And the co-immunoprecipitation that resembles a Rorschach blot test (but it also looks like two Sumo wrestlers kissing) has to be redone...and that one...and,yeah, that one too. Okay, back to the bench.
Right. So you've got all the data. Now, you could just write it up and publish it in a journal where nobody is ever going to notice it, but I know you want one of those high-impact things. Of course you do, because you've been working so hard. So, the next thing we need to do is decide how your results destroy Myth and Wonder. It goes like this, “we thinkthings work one way (and this is our Myth), but they don't”. Oh sure,eventually you're going to show that everyone really agrees with you,but that comes later. Right now, we're working on the old bait and switch. You say, “it seems logical, reasonable, almost obvious, that things work like this, but guess what – they don't! The logical, reasonable, almost obvious thing is a Myth! And (here's the really exciting part) we can prove it!”
Good, and now, to prove it, you have to destroy Wonder – by which I mean the stuff of wide-eyed awe, which is, of course, an important part of what we do, and the wonder we feel when confronted by the remarkable (I often stand in wonder when my colleagues tell me how they spent their morning at a faculty meeting to determine the color of the seminar announcement). And I mean wonder as in “I wonder how in the world that happened”. This wonder thing is all around our day-to-day life in science, but don't mention it in the paper you're writing. You have to remove all of those little bits where you say, “wouldn't it be lovely if things work the way that we say they do? Wouldn't that be exciting?” And you have to replace those bits with facts, controls, clean results, that sort of thing. And (do I have to keep saying this?) you've got to be really sure you're right –no fudging, fibbing, looking sideways, wishing. No Wonder. (Yes, if we delve really deep, we can rediscover Wonder, as in, isn't it wonderful that it does work like this, but, hey, bend with me). So, are your data reallygood enough? Or do you have to do a few more experiments to destroy Wonder?
Still here? Fantastic. Now all you have to do is write the thing, and that's going to take a different set of skills. Now it's time to start lying.
It goes something like this. “We wanted to know why certain stuff happens.” No, you don't say this, you replace “certain stuff” with an interesting biological phenomenon. Okay, so “we wanted to know this (but this is a lie: you really never thought of this problem until you had your first result and realized that it could explain something). Therefore, we performed our first experiment shown in Figure 1(lie! You didn't perform the experiment in Figure 1 until yesterday). We reasoned (lie!) that, if true, then it predicts the result in Figure 2 (no,you did that experiment to test a completely different idea, but you were wrong). Therefore we performed the experiment in Figure 3 (lie: you did this experiment because you had cells left over). Next... (lie, this was the first experiment). We concluded... (nope, somebody figured this out for you, but you'll slip her name into the acknowledgements where no-one will notice).
But it's okay. You're supposed to lie about this stuff.
Why is this lying okay? We know that we absolutely, positively cannot lie even a little bit about our data. Okay, we do fib about “the figure shows a representative experiment.” It inevitably shows the best experiment, but you know that's not what I mean. And we absolutely cannot lie about how we did it: the methods we describe really have to be the methods we used. But we can (and do and really must) change around the order in which the experiments were done and why. And how they were interpreted when we did them.
But everyone knows that this part is just a story. Sort of a rewrite– “How I would do all the experiments again, knowing what I know now.” But why? Why can't we simply let the data speak their wondrous truths? One could easily write a paper without such lies, simply stating each result in a completely impersonal way. Figure 1 shows an experiment that gives this answer but raises this question. Figure 2 sort of answers the question, but it gets better when you look at Figure 3. Figure 4 really nails it, except for the problem that is fixed in Figure 5. Figure 6 looks cunning in a whimsical ensemble for summer. No lying about when we did what and why. Just straight facts. Hard, cold, true science. It could be done like this, but it almost never is. We love to tell the tall tale of the research.
I think we make this fictitious foray for a reason. We put a human face on the work. Because the story we tell about how and why the work done is possible, and it really is in the spirit of what we did, leaving out all the stumbling around, mistakes, bad reagents, false positives, real negatives and plates that were dropped. And it just feels better to take responsibility for the conclusions than to assert that the data are the ones to give the credit to (or blame), because all of it could still be wrong. Yup, even if we're pretty sure it's right, or we really hope it's right, it could all be wrong, wrong, wrong. So we show that we really did try to get it right, and we take responsibility for choosing which experiments to show. The story explains why these are the ones that make the most sense. All those experiments that failed (and there were lots of them, right?), they didn't make sense in our story. So we need a story to explain why the data we pick seem to tell another story, the real story of the paper. The truth(whatever that actually is) may be hidden in the experiments we didn't pick,but our story didn't lead us that way.
I see you doubt me. I hear you saying, “hey, we do science to find out what's what, and when we do find it out, it's right. Why all this talk of fictions and stories and lying?”
So here's something to do on a rainy day – or a sunny day when you're too pale to go outside and risk exposure to all that UV. Go into the library(the building with lots of books – ask an old guy where it is) and pull out some ancient issues of some very high-impact journals – more than 10 years old... really old. Good. Now browse through the stuff that relates to things you know something about. You may find some pretty cool papers. But you are also going to find lots of stuff you never heard of, things that aren't even close. Stories that were told that sounded pretty promising but just didn't work out. Detours on the highway of truth. Lovely, scenic tours of the nearly right or the dead wrong.
And that's exactly what we do. We try to view the universe of how things work, and we take our colleagues along for a ride to see what we've seen. And we tell them a little story of how we got here, and why we think it's right. And hopefully we all enjoy the ride. That's the secret of writing papers: enjoying the ride.
