Scientific publishing in China is in a quandary. Many articles in the country's 5,000-plus science and technology journals go unread and uncited, calling into question the value of the research. It also raises doubts over the effectiveness of China's scientific publishing — which, after all, is to disseminate details of research for others around the world to build on. One Chinese scientist has referred to the majority of China's publications as “pollution”.

Yet when it comes to publishing in international journals in English, Chinese scientists are second by volume only to those in the United States. Now, librarians and government officials in China are beginning to question why their own journals publish so few of these quality papers. The country's General Administration of Press and Publication (GAPP), which regulates all publishing, is to make reforms to strengthen its home-grown industry. This makes sense. And publishers in China could no doubt beat their Western counterparts at their own game. But GAPP has so far given few details of the reforms, causing confusion among the people most closely involved: the publishers. How should it be done?

GAPP should be aggressive — as it has promised (see page 261) — in evaluating its journals, improving the strong and killing off the weak. The resources and publishing rights currently allotted to eliminated journals could be transferred to the growing number of scientists and publishers who are familiar with the international publishing landscape and are finding niche areas for new products. Many of these journals will be in English, and additional resources will be needed to help ensure that articles read well and are peer-reviewed fairly.

Clearly, there is a strong demand for more information on the best science in China. This is especially true in fields in which the country excels, such as optics and materials, but also in areas such as public health, where data from China have been overlooked (see Nature 430, 955; 2004). If done well, these new journals could bridge a gap between the stronger Chinese literature and foreign scientists. A publisher of optics and photonics journals at the Changchun Institute of Optics, Fine Mechanics and Physics, for example, already plans an English-language publication to replace its weakest optics journal. It is a response to increasing demand from those researchers who have read abstracts in English and want a full translation. The journal will publish reviews that put Chinese experiments into the wider context of global trends.

The best opportunity to revive Chinese publishing, whether in Chinese or English, probably lies in an open-access platform — increasingly popular in Western journals. Many Chinese journals already charge authors a publication fee, so should be able to make a smooth transition to the open-access model, in which they are supported by fees rather than by subscription revenues. Making content freely available would help to popularize journals, and would encourage them to develop an online presence. Too many operate without one, enjoying a captive audience at their home institutions and lacking any competitive spur to bring themselves up to speed on Internet publishing. The government could provide the interest, investment and expertise to bring these publishers into the twenty-first century.

It would, however, be a mistake for government agencies to give themselves too strong a role in this transition. GAPP has mentioned the creation of five to ten strong publishing houses that would concentrate on science and technology. This could work well, provided that they are able to move freely and openly, and can compete both with each other and with foreign publishers.

Most importantly, GAPP needs to consult quickly with its publishers if reform measures are to be put in place by next January, as intended. The lack of details mean that resistance to the reforms from publishers seems unavoidable. GAPP needs to make its expectations and evaluation methods transparent and bring in its reforms consistently. So far, that does not seem to be happening.