An infamous case of a narrative not matching real life occurred a few decades ago, when an international forest research outfit launched its revised map of Indonesia’s forest cover. The scientists’ pride was obvious as they elaborated on their GIS prowess in modelling this feat. Yet, local experts noted numerous serious errors in the results and one of them inquired about the sites visited to verify the map. The muffled giggles following the answer – but a dozen sites across a handful of provinces – has been a constant reminder for me that any narrative stands on the prowess of its field verification.

More recently, foreign experts included the 1988 land system maps by the Regional Physical Planning Programme for Transmigration (RePPProT) in the 2008 high conservation value toolkit for Indonesia (Consortium for Revision of the HCV Toolkit for Indonesia 2008). They argued the RePPProT maps are accurate proxies for endangered ecosystems. A plethora of errors in the RePPProT model has emerged since, which regularly resulted in expensive revisions of high conservation value assessments. Hence, while the RePPProT maps were the best science had to offer at the time, it results in narratives that have little bearing on real life forests and other endangered ecosystems during local assessments.

Another example of false narratives based on pseudo-science are the accusations by Greenpeace that Sinar Mas Agro Resources and Technology (a.o.) converted deep peatlands. After extensive field verification (Anonymous 2010) it was painfully obvious that this narrative was based on dodgy models, and even more dodgy field verification. We regularly encountered peat where the models claimed there was none, and vice-versa. Yet, a representative of the organisation that developed the model rejected all field observations that didn’t conform to this model. It remains my most vivid memory of how blind faith in models is detrimental to real life conservation activities.

Lately, a few colleagues and I successfully challenged a model that concluded that land that was intact rainforest “yesterday” can be planted with oil palms “today” and certified as sustainable “tomorrow” (Cazzolla Gatti et al 2019, Cazzolla Gatti & Velichevskaya 2020). This model visually assessed Google’s ‘Pretty Earth’ imagery, without considering the limitations of this imagery nor verified the model’s results through field visits or through comparable datasets (Assen et al 2021). Once these issues were incorporated in the model, it produced significantly different results and uncovered a critical flaw in visual assessments. (We coined this the Pretty Earth Fallacy: an untrained observer perceives dark green areas as intact rainforest.) Quite a damning finding for narratives based on visual assessments!

My ongoing analysis of the accusations over deforestation in Indonesian oil palm plantations and forest concessions by Chain Reaction Research (CRR) suggest it regularly falls prey to the Pretty Earth Fallacy. An associate “researcher” described CRR’s “model” as visual checks based on a tree loss model (Anon 2021) to declare deforestation occurs ‘when we think it were high carbon stock forests’. My consequent inquiry about how CRR’s model incorporates a.o. land clearing 3-4 years prior to the assumed deforestation remains unanswered to date. Similarly, the plantation permit of a recent CRR top “deforester” (Natividad 2022) overlaps with open mining (see a.m.o. Arifin 2022 and Rochmad 2022a-b). Yet CRR does not account for the role of this activity in its report. It’s a great example of how flaws in a model mixed with the Pretty Earth Fallacy result in false narratives about deforestation.