The Truth is out there (somewhere!)

New research from the Cambridge Centre for Carbon Credits provides an interesting contribution to the debate around how best to use carbon finance to fund nature conservation at scale, in a way that reflects the actual cost of protecting forests by local communities in particular, and host countries (mostly in the developing world) in general. Academia has a part to play in understanding the different approaches, both in terms of their scientific robustness of how we model deforestation threats, and in terms of scrutinising the effectiveness of the financial incentives for communities implementing the governance processes that will ensure long term protection of these forest areas (permanence is central to this paper's treatment of the topic).

However much of the academic research published around the reliability of forest-based carbon credit estimates is pervaded by a concerning level of bias and partiality. Specifically I refer to the flaws in the single paper on which most of these headline grabbing claims ("Revealed: more than 90% of rainforest carbon offsets by biggest certifier are worthless, analysis shows", The Guardian, 18th Jan 2023) are based. The paper, titled "Action needed to make carbon offsets from forest conservation work for climate change mitigation" is a paper by West et al which explores how AI might provide an alternative way to model deforestation loss in a way that removes some of the perceived subjectivity that informs the conventional baselines used to verify credit issuance under one or other of the established international carbon credit Standards.

It is good that a group of academics are interested in helping everyone find new and better ways to estimate the effectiveness of forest protection schemes by using the new toys available to the world in the form of enhanced earth observation technologies and AI powered analysis systems, but the category error that has been committed by commentators, journalists, observers and the general public misunderstanding that this is simply an academic scientific process. Science never pretends to give us 100% certainty around any specific area of knowledge, it merely seeks to reduce the levels of uncertainty about specific questions.

But because there is one peer reviewed paper that concludes that using one approach to analysing forest loss which contrasts quite dramatically to the various other peer-reviewed methodologies for assessing the same phenomena does not automatically mean that a new truth has magically been arrived at! Why do these academics cite only the recent paper, and omit to reference the large canon of peer reviewed work by forest scientists published over the past 30 years? It smacks of, at best, a lack of intellectual curiousity, and at worst, ideological bias.

I cannot speak for the reliability of West et al's analyses of all the carbon projects included in their paper, but in the case of one of Carbon Tanzania's projects, the synthetic AI engine has made an obvious error. The "control" site to which it chose to compare the Makame landscape (an acacia-commiphora savannah rangeland inhabited by Masai pastoralists) is shown as being on the shores of Lake Victoria! The authors of the paper state that “We used real-world comparison sites to show what each REDD+ forest project would most probably look like now, rather than relying on extrapolations of historical data that ignore a wide range of factors, from policy changes to market forces”. It does not take a trained socio-economist to realise that the "policy changes and market forces" at work in the Masai Steppe of northern Tanzania where communities of Masai pastoralists compete with shifting agriculturalists seeking new land for their farms is a world away from the shores of Lake Victoria where people have lived, farmed and fished a landscape that was deforested and modified many centuries ago.

So let’s continue to use these new techniques, including this new one proposed by the Cambridge Centre for Carbon Credits, to scrutinise the accuracy of our models for understanding the potential patterns and drivers of deforestation across our critical and wildlife-rich forest landscapes, and let’s allow academia to debate the relative validity and value of each approach, but let’s also not consistently cite one paper with a singular and clearly flawed methodology for understanding the topic as providing us with some cosmic “truth” about the issue.

The truth, if you take the time to talk to the people on the ground who actually bear the responsibility and costs of protecting these resources, is that they see deforestation happening on their doorstep everyday, and deal with the very real effects of illegal encroachment, land conversion and poaching. When you tell them that some scientists in a far away land using a computer have decided that there is no value in their forest protection project because the threat of deforestation to the area is deemed to be overstated (or non-existent) by that computer, they just laugh and say "tell them to come here, and we will show them".