Research papers will be free to access, eventually – Nature’s Philip Campbell

What are the difficulties in getting research funders to pay for published papers?

‘The journals have to cover their costs and research, so different journals have different costs, but if you look at the big journals which have professional staff, they put a lot of effort into copy editing and putting papers up online and maintaining them. If you are going to cover all of those costs, you are going to charge a group of authors for a paper in a journal like Nature well over GBP 10 000 (EUR 14 000), whereas the most people pay at the moment and are willing to pay I would say is GBP 5 000.

‘In the total scheme of things it is not a lot of money, but at the moment we are slightly stuck on that one, and actually there are whole disciplines that have no money anyway, like the social scientists do not have grants with funds attached that would allow them to pay for it.’

Do you think researchers and research funders will agree in the end to meet the full cost of publishing papers in journals like Nature?

‘Yes in the long run I do. I think that publishers will find ways of doing things more cheaply than they currently do and still maintain the quality, so that might bring the cost down, and also scientists will see the advantages and so will the funders. It is partly a matter of moving existing money that is currently spent on buying journals and subscriptions.’

What will be the impact of open access and open science?

‘To me it is the open data that matters just as much as the open text of the paper. By the data I mean the data that the researchers have independently collected separately from the paper. If you can get your hands on that, then you can really get in there much faster to check what is in this paper. In some complex pieces of work that can be immensely time consuming, but without it, it is quite hard to validate what the paper is saying.

‘I would love all the literature to be open access. I would love funders to find the money to pay what it takes to put the literature up there and the data, it’s a large sum of money but it’s a small sum of money compared to the total research spend.’

“

‘To me it is the open data that matters as much as the open text of the paper.’

Research papers which are submitted to journals like Nature are checked by other academics, the so-called peer review process. Does that need changing?

‘The capacity of the science community to peer review all of the research that is out there is very strained because there is a growing quantity of research. And even though the number of researchers themselves has also grown, somehow or another it doesn’t seem to be keeping up. I think the other thing that can happen with peer review is that people who are experts from their own perspective can take a paper and judge it only on their own perspective, rather than stepping back. In those circumstances, it is crucial for people running the peer review, whether they are editors or fund managers, to be able to have knowledge themselves. So we solve that at Nature by sending people out into the labs, getting them to know the areas, and we make our own judgements. We will overrule referees on occasion, from the point of view of whether it is interesting or not. If the referee has got a technical problem, then we will of course abide by his or her advice. So that question of breadth of knowledge and breadth of outlook and imagination for a peer reviewer can limit the quality of what you get back.

‘The final thing is that, I’m afraid to say, peer reviewers can use the process competitively and can speak negatively about papers and grant applications in order to hold back their competitors. The only thing one can hope for is that the editor is using more than one peer reviewer and so you can control for that, and that they also have their own knowledge and instincts about the field.’

In a recent paper published by Nature, Glenn Begley and Lee Ellis found that they could successfully replicate the scientific experiments used to support just 11 % of so-called ‘landmark’ biomedical research papers published in high-profile journals. Is this a fault of the peer review process?

‘A researcher may have mistakenly done something in the lab, or may have even subconsciously picked the best data to show, or even consciously picked the best data to show. All of these things happen and you just simply can’t pick that up when you are looking at a paper, because a referee has a day at most. All they can do is take on trust what the paper says. So, if there is stuff going on in terms of errors behind the data then it’s very hard for a referee to pick it up.’

Does it mean that many of the papers published by Nature are wrong?

‘All scientific papers are contingent. Every science paper is just the best statement that the authors can come up with. Some have outstanding discoveries whose interpretation turn out be to invalidated by subsequent work. There is, I am sure, a worryingly high proportion of biomedical papers in Nature and other journals that turn out to be wrong. There are all sorts of reasons why it is only over time that any particular paper is shown to be right or wrong, there are natural errors that occur, there is misconduct, which is a very, very small proportion of the whole, and the number of papers that are retracted because they are wrong is tiny, that’s something like 0.1 % of the entire literature. Most papers in Nature are certainly a reasonable approximation to the truth, let’s put it that way.’