The Intergovernmental Platform on Biodiversity and Ecosystem Services (IPBES) is adopting its initial work programme for 2014–18. This is a good time to assess whether the “rules of engagement” outlined for its success (E. Turnhout et al. Nature 488, 454–455; 2012) have been put into practice. We find that there are still some serious shortfalls that, unless addressed promptly, could undermine the transparency and credibility of the output from the IPBES.

Some of the nine rules (R1–R9; see go.nature.com/guc1gk) are now being followed, at least partly: reporting on progress is to be staggered (R8); the call for a decentralized approach that is sensitive to local knowledge, needs and conditions (R1) is now reflected in several deliverables; and most mandated functions are being addressed simultaneously (R2). Cautious optimism is also justified for R3 and R5: expert groups for discussing terms and methodologies are to be established, and non-scientific knowledge systems will be included.

However, non-elite actors are not yet properly involved, and decisions on stakeholder engagement and communication strategies have been postponed, which will affect R3 and R5; this also applies to the inclusion of “trusted civic organizations and networks” (R7).

Furthermore, stakeholders are only marginally involved in nomination procedures for the Multidisciplinary Expert Panel (MEP) and for experts who will work on IPBES deliverables, and are not involved at all in the selection process (R4). Of the MEP members selected, the mix is unbalanced (22 natural scientists, 2 economists and 1 social scientist, and women are under-represented). It is also unclear whether minority dissenting views will be incorporated into the IPBES process (R6). Re-evaluation and improvement of IPBES processes (R9) will be hard because IPBES members have declared that they are not prepared to renegotiate after issues have been formally agreed.