Context: This conceptual paper tries to tackle the advantages and the limitations that might arise from including second-order science into global climate change sciences, a research area that traditionally focuses on first-order approaches and that is currently attracting a lot of media and public attention. Problem: The high profile of climate change research seems to provoke a certain dilemma for scientists: despite the slowly increasing realization within the sciences that our knowledge is temporary, tentative, uncertain, and far from stable, the public expectations towards science and scientific knowledge are still the opposite: that scientific results should prove to be objective, reliable, and authoritative. As a way to handle the uncertainty, scientists tend to produce “varieties of scenarios” instead of clear statements, as well as reports that articulate different scientific opinions about the causes and dynamics of change (e.g., the IPCC. This might leave the impression of vague and indecisive results. As a result, esteem for the sciences seems to be decreasing within public perception. Method: This paper applies second-order observation to climate change research in particular and the sciences in general. Results: Within most sciences, it is still quite unusual to disclose and discuss the epistemological foundations of the respective research questions, methods and ways to interpret data, as research proceeds mainly from some version of realistic epistemological positions. A shift towards self-reflexive second-order science might offer possibilities for a return to a “less polarized” scientific and public debate on climate change because it points to knowledge that is in principle tentative, uncertain and fragmented as well as to the theory- and observation-dependence of scientific work. Implications: The paper addresses the differences between first-order and second-order science as well as some challenges of science in general, which second-order science might address and disclose. Constructivist content: Second-order science used as observation praxis (second-order observation) for this specific field of research.