During the COVID-19 pandemic, many governments have adopted responses revolving around the open-ended use of non-pharmaceutical interventions (NPIs), including “lockdowns”, “stay-at-home” orders, travel restrictions, mask-wearing, and regulated social distancing. Initially these were introduced with the stated goals of “flattening the curve” of hospital demand and/or the eradication of the virus from the country (i.e., “zero covid” policies). Over time, these goals have shifted to maintaining sufficient NPIs in place until such time as population-wide vaccination programmes have achieved an appropriate level of herd immunity to allow lifting of these measures without excessive hospital demand. Supporters of this approach have claimed to be “following the science”, insisting that criticism of any aspects of these measures is non-scientific or even “scientific misinformation”. This idea that only one set of scientifically valid opinions on COVID-19 exists has encouraged the media, social media and even scientific journals to suppress and/or dismiss any differing scientific opinions as “erroneous”, “discredited” or “debunked”, resulting in discouragement of open-minded scientific inquiry or discussion. Accordingly, in the current article we identify two distinct scientific paradigms to analysing COVID-19 adopted within the medical and scientific community. Paradigm 1 is primarily model-driven, while Paradigm 2 is primarily empirically-driven. Using these two paradigms we have analysed the epidemiological data for 30 northern hemisphere countries (with a total population of 882 million). Remarkably, we find using each paradigm leads to diametrically opposite conclusions on many policy-relevant issues. We discuss how these conflicting results might be reconciled and provide recommendations for scientists and policymakers.