How do We Get Out of this Mess? (II) Complexity Development and Scaffolding Your Models

Guangzhou butchery — young bullocks in one of the Wet Markets you keep hearing about

Though it certainly won’t be the last, the academic literature is starting to turn to the obvious failures of the epidemiological science surrounding COVID, the conclusions, and their dramatic failures in over-predicting harm, which was then used as rationale for global civil liberties suspensions around the globe.

I’m going to discuss this in the context of a paper that came flying across my Twitter feed. The title of the paper is: Imagination and remembrance: what role should historical epidemiology play in a world bewitched by mathematical modelling of COVID-19 and other epidemics? The authors, George Heriot from Monash University, and Euzebiusz Jamrozik from Oxford, have the right credentials at least to be believable.

It’s a decent enough paper, and explores the ramifications of a refusal of the COVID mathematical modeling community to ground themselves in past pandemics with data as well as historical newspaper coverage. For those of us with a COVID hobby/problem, I highly recommend reading it. The authors point out with particular historic pandemic cases (they draw heavily, for example, on the 1889-1891 pandemic, with its similar behaviors in terms of low morbidity for children, and obvious seasonal variation.) Here’s the killer paragraph from the paper:

“Specific epidemiological correlates between the 1889–91 and 2020–21 pandemics include the low morbidity among children, the lack of the shift in excess mortality to younger age groups usually seen with pandemic influenza, the magnitude and distribution of peak excess mortality ratios in metropolitan settings, and the rapidity of epidemic propagation within communities (Valleron et al. 2010; Campbell A. and Morgan E. 2020; Nicoll et al. 2012; Nguyen-Van-Tam et al. 2003; Honigsbaum 2010; Smith 1995). While downscaling this synoptic analogy to make short-term forecasts of COVID-19 activity in any given place 130 years later is clearly foolish (short-range forecasts from well-observed local data being very much the preserve of computational modelling), the historical record may provide a richer and more useful understanding of the range of medium- and long-term consequences of a pandemic of this epidemiological pattern on human societies than even the most complex mathematical model.

If you had to have a quick takeaway, the authors say “look, this has happened before… and will likely happen again — just get your timescale right.” (For all Battlestar Galactica fans, cue the appropriate music..)

Looking at the structure of knowledge in the paper, the work itself fits well on the concept of what I’ve named Intellectual Flatland. And what is Intellectual Flatland? It’s that map that academics keep of their disciplines, that maintain there is no evolutionary pattern to information complexity. Everyone has their own little island, and that can definitely accrete, or erode as time goes along. Connectivity and precedence just isn’t there. The case the authors make is had this static pattern-matching happened before the start of the pandemic, we would have moved our understanding far faster, far more quickly with regards to modeling, if we had just paid attention to the past. If we had only visited Past Pandemic Pattern Island.

And, at some level, they’re correct. But the deeper “why” isn’t just “send all epidemiological modelers to pandemic history class,” though that might have helped a little. A confession — I’m always a bit suspicious of learned knowledge, because having delivered it for quite literally 37 YEARS!!! I just know how little people listen if their brains are not prepared for the message, or its complexity.

The real answer is to get people to do more asking “why?” And that’s harder. There are tools I highly recommend, like the Toyota Design Process “Five Whys”. But it really helps more, if your brain has the circuits, to understand the knowledge complexity scaffolding process.

I’ve written quite a bit about this. The short version is that we start with lessons learned from the simplest human social structures (Survival Bands) and then move up — through Tribal/Mythical knowledge, Authority-Driven Power Structures, Legalistic/Algorithmic processing, Heuristic thinking where individual decisionmaking matters, and then on up into higher empathy synergizing of individual viewpoints, and ending up Reflecting back on the whole can of beans.

The big insight here — is this whole knowledge structure thing is nested — lower levels are not thrown away — they instead are embedded in higher forms. It’s not “either/or”. It’s a multiple “Yes/And”.

Here’s the Knowledge Structure chart. This thing is gold.

The good stuff — Canonical Knowledge Structures and the basis for all information

In a perfect pandemic world, all people executing at a given knowledge structure would be well-scaffolded with at least some representative examples of the information underneath. What that might mean, in the case of modeling, is that expert analysis of historical data would be integrated as far as weighting functions for any contemporary modeling. That would mean that the historical information referred to in this paper would have been used, in some kind of translated aggregate, to map the pattern of this pandemic to the past. If this was wrong — that for whatever reason, there was no historic precedent — that would have shown up as well as data collected for the broader pandemic, as time progressed, would be incongruous with the model.

It’s worth it to take a minute and discuss the meaning of “history” in all this. Of course, anyone that’s ever looked at history knows that any numbers from history are literally fraught with peril. The authors of the paper claim to have analyzed historical data, and mention newspaper clippings as well. My guess is they’re playing the empirical research card for status assertion — they are academics, in academic hierarchies, after all. Yet any statistical researcher worth their salt knows that we really have only improved with data collection. When it comes to illnesses and death (especially of the poor) such numbers from history are mostly nonsense. No one really knows.

But even with that said, history as myth is vitally important. History as myth, or narrative, often captures deep information that actually happened — or the myth memetically was not likely to have persisted. Yes, BS does survive. But myths last because, fascinatingly enough, they contain information that are the result of validity grounding, often from large-scale catastrophe. In the lower v-Meme set (below the Trust Boundary and Legalistic/Absolutistic v-Memes) there is precious little validity grounding. Authorities, experts, and processes say “we know stuff, and you should believe us because we are smart/thorough/etc.”

Myth is different. One of my favorite examples of how this works involves Native American reverence for nature. There is a whole religious edifice built in tribal societies saying “preserve Mother Earth.” A good question is “where did that come from?” Obviously, in terms of Western and Eastern philosophy, for the most part, we don’t seem to care very much about the Earth.

If one, however, understands the context of the Pleistocene extinction, where tribes cross the Bering Land Bridge, and found a whole continent of giant, tasty mammals, this all starts making more sense. The first humans dined away on almost all of these, as few of the animals were evolved enough to avoid the newly transited predators. Many of the species that vanished were “giant” versions of earlier species. And giantism occurs as an evolutionary response in a given species from a lack of evolution along the lines of inter-agent coordination. The few large species that survived — like bison; and mammoths were some of the last to go for reasons of inter-agent coordination.

The incumbent likely starvation after the Giant Pleistocene Barbecue knocked those tribes of humans on their butts — and led to the deeper validity grounding of “don’t just kill off nature or it won’t be there to eat.” These types of deep history events likely constructed the mythos of nature worship.

There’s not much difference between the story above, and the need for appropriate myth generation in the epidemiology community, with solid narratives that form the basis for sound modeling. A selection of narratives might form the basis for model construction, as well as reflection upon why, if a model did fail in its predictive capacity, exactly why it failed.

My guess is you’re going to see serious narrative generation in microcosm about five years from now, when the reality of the societal transgression is fully understood. Epidemiologists and immunologists will be looking at a Survival-level event, with the incumbent trauma and neuroplasticity that accompanies all these things. The In-group supporting all this is going to come un-done, because the recommendations of the various NPIs have been so ineffective and fundamentally anti-human behavior. Like it or not, we are social animals, and all the NPIs are profoundly anti-social, and especially anti-empathetic. And also importantly, anyone who can read a time series can clearly see they didn’t work.

A better question is this — how did the epidemiological and immunological communities get so ungrounded? My suspicion is this has come as a price paid of the specialization and sophistication of the technical communities. Those models, even if they’ve been wrong, are not trivial. There are a host of skills required in order to create them, ranging from statistics, data analysis, coding and so on. There seems to be some magical belief that people learn everything in a field when they get a Ph.D. Nothing could be further from the truth. And in the heightened pressures of the current research milieu, you have to learn enough to avail yourself of funding. You’re far better off learning a statistical package than reading historical accounts of the bubonic plague. And worse — this turns out to be a Survival level choice if you’re a grad student. The longer musings are going to come far later in the career game, after the rigid hierarchical social structures inherent in research groups have a chance to hammer on your brain, and actually prevent you from caring on big-picture notions outside your immediate wheelhouse/silo.

And that, of course, will distort how you see the world, or history and your place in it.

One thing that is important to remember is that such sophistication comes with it sophistication in self-delusion as well. We can create elaborate rituals to elevate our status, and certainly the epidemiological community has done this. Two years ago, the prospects for various stars rotating through permanent advisory chairs in networks like CNN and MSNBC never occurred to them. Finally, at last, they were receiving the notoriety that their titles surely implied they deserved.

But a lack of validity grounding will get the best of any civilization. And the only way to avoid that is to understand knowledge scaffolding, and use it.

One thought on “How do We Get Out of this Mess? (II) Complexity Development and Scaffolding Your Models

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s