293 - Adoption Context

When discussing the impact of a technology's adoption, it is critical to keep in mind both the absolute and relative shifts in value, as well as the absolute values of what you're talking about. For example:

• Technology A improves life for the majority by a value of 0.2 on some scale.
• Technology A improves life for minority B by a value of 0.1 on some scale.

This might appall some if examined in isolation as an absolute shift in value, except that the shift in values might translate to a new absolute value of:

• Technology A improves life for the majority from 0.8 to 1 on some scale
• Technology A improves life for minority B from 0.9 to 1 on some scale

Without taking absolute values and the relative and absolute shifts into account even equality can look disparate. It can even skew in the opposite direction relative to the naïve inspection of only one factor. Considering all of these is also critical for purposes of validation and consequently for debunking.

One of the methods famously studied and used by Daniel Kahneman to reduce cognitive bias is to:

1. Build a structure within which each element is clearly defined and scored on a scale.
2. Complete all elements of that scoring process for all options and/or candidates, with the opportunity to make note of anything falling outside of those metrics at the end of each assessment.
3. Directly compare any 2 or more candidates jointly, looking at how they scored.

Within such systems, both absolute values and both metrics for shifts may be considered in context, alongside other factors such as volume and cost, allowing for cognitive bias to be reduced and reasoning steps to be broken down into more easily quantifiable, auditable, and human-comprehensible steps.

I was reminded of this when reviewing a paper for peer review that landed at the intersection of AI, Medicine, Cognitive Bias, and real-world applications. What set that paper apart is that they actually did a good job of considering all of this, to the point that my review to the journal noted that if the paper was rejected I'd permanently blacklist them from publishing any more of my work. That is about the most glowing endorsement I can give a paper and one that I've never had the occasion to give before.

I've observed that doctors often overlook simple but critically important factors, such as sodium vs potassium ion channel balance, or the corresponding conditions of imbalance and their often well-studied interactions and complications with common medications. Even simply reminding a doctor to ask about those conditions and symptoms where known complications are present could greatly improve many patient outcomes.