Wicked Problems and usefully true forecasts necessarily require open systems

The contest between models when not recognizing they are competing models.  

This is from Collapsing Empire by John Scalzi, a reasonably entertaining sci-fi story.  In the novel, there is a natural phenomenon called Flow that allows interstellar space travel at speeds beyond the speed of light.  In the novel, there is some physics handwaving and the underlying science is glossed over.  It doesn’t matter other than that the existence and path of the Flow determines the health and prosperity of the Empire and the individual planetary systems, particularly Hub, seat of the Empire and End, the remotest populated planet.  

Scalzi, in telling his story, structures a point that is too often underrated in national discussions.  

Page 245.  Nadashe, Amit, and Ghreni are three aristocratic conspiratorial siblings seeking opportunity.  The Interdepency is the Empire of trading planets.  

“This is the future,” Nadashe said.

Ghreni got up and walked closer to the monitor studying the new map.  “Where did you get this from?”

“I have a friend from university who grew up to be a Flow physicist,” she said.  “My friend was casting about for something to do her doctoral thesis on and she came across a monograph about a potential long term shift in the Flow.  The person who wrote the monograph never did anything with it.  She tracked down his information and he'd become a tax collector for the Interdependency.  So she followed up, worked the data, and came to the conclusion that after more than 1000 years of relative stability, the Flow streams are about to shift, probably to this map.”  

“When?” Amit asked.

“She says the data indicate it's already starting.  First slowly but then quicker and quicker.  It'll probably start in the next decade.”  Nadashe pointed at the monitor.  “This map is likely to be what the Interdependency looks like in 30 years, she says.”

Ghreni furrowed his brow. “’Likely?’ What does that mean?”

“She's modeling what she sees as the most probable pattern of collapses and shifts, based on her data set.  She says this pattern has an 85% probability of being what things will be once the pattern stabilizes.  And when it stabilizes, it will likely hold for another thousand years.”

Ghreni pointed.  “And she's sure End is going to be the place all those streams are going to focus on.”

Nadashe nodded.  “She says that’s actually the most predictable part of the shift.  It's happened before, apparently.  The shift of the Flow streams.  Her data suggests the locus of the Flow activity switches off between Hub and End every thousand or two thousand years.  There's a less than one chance in 100,000 that some other system will be the focus of the Flow streams.”

So Nadashe, Ghreni and Amit (siblings) see the approaching change in the Flow as a distinct opportunity for profit making.  If the locus shifts to End, that is where money will be made in the future and Hub will become the backwater.  Their opportunity is to control End before anyone knows about the prospective change in Flow.

They have an 85% confidence in the forecast of the new patterns of Flow.  85% sounds like good odds during an existential change. 

But is that the decision model to be used?

Much later, the Tax Collector, who is actually the Flow Physicist under a guise provided by the Emperox, has a different conversation with Lord Ghreni.

Page 262.  Count Claremont is the Flow Physicist appointed by the Emperox to secretly study the Flow under the guise of an appointment as the Imperial Auditor based at End.  Marce is his son, also a Flow Physicist.  Lord Ghreni is one of the three conspiratorial siblings from the quote above.  Ghreni has captured Claremont and believes he has control over him.  

The count laughed, weakly.

“Why are you laughing?” Ghreni demanded.

“Lord Ghreni, you have no idea what's coming in the next five years,” the count said.

“On the contrary, Claremont, I do.  Changes are coming.  End is going to become the heart of the Interdependency.  All paths will lead to here.”  

“No.  No paths will lead to here.  In five years we'll be alone.  It's a physical certainty.”

Ghreni began to feel uncomfortable and realized it was the count’s last sentence that did it. “What do you mean?”

“Why do you think I sent my son away, Lord Ghreni?  At this specific time?”

“To escape the fighting here, and to complain to the Emperox about me kidnapping him.”  The latter was why Ghreni wanted Marce out of the way if he couldn't be retrieved. Ghreni wasn't sure how much pull the Count of Claremont had at the imperial court, but he knew Nadashe and Amit would not appreciate a report from End about his actions making their lives harder.

The count shook his head.  “I had him leave now because if he didn't, it would be impossible for him to ever leave.” 

Ghreni was puzzled. “Are you talking about the Flow stream?”  What would an imperial auditor know about Flow streams? The count’s specialty was taxes, not phys - 

“Oh my God,” Ghreni said, and openly stared at the count,  “You're him.”

The Count of Claremont seemed puzzled but amused.  “Who am I, Lord Ghreni?”

“You're him! The Flow physicist! The one whose work Hatide Roynold based hers off of.”

Claremont continued to look puzzled for a moment, but then Ghreni saw a sort of slow realization come over his face.  “I know that name. I remember that name. She sent me some of her work and a list of questions years ago.”

“And you didn't respond.”

“No, I didn't.  I had been ordered by the Emperox not to discuss my work with anyone.”  Another expression popped onto Claremont's face then. Concern.  “You think her work is accurate, don't you?  You think the Flow streams are moving to End.  That's it, isn't it?”

Ghreni’s mouth gaped.

Claremont slapped the side of his bed.  “That is it!  That's actually it!”  Claremont started laughing, a loud, almost agitated noise.  One of the marines opened the door and poked his head in to investigate.  Ghreni angrily waved him away.

Eventually Claremont got control of himself, wiped a tear away from his eye, and looked at Ghreni.  “Oh you sad, ambitious fool,” he said.

What do you know?”  Ghreni asked.

“I know that Hatide Roynold was sloppy with her math. I know if she didn't check some of her base assumptions, she's probably iterating wildly away in a direction that has no basis in reality.  Has any of her work you've seen had peer review?”

“No,” Ghreni said.

Claremont nodded.  “Of course not.  She's like me - snapped up by a patron and working alone.  Peer review is important, Lord Ghreni.  Until Marce was old enough to start checking my work, I was flying blind.  Made some stupid mistakes I just didn't see.  Roynold was making them too.  I know, I saw them.  She probably never corrected them.”  Claremont leaned forward and weakly poked Ghreni in in the chest. “And you, you ignorant grasping poltroon. You didn't know any better.”

There are actually two things going on here.  From my perspective the first is the one least discussed publicly.

Models are maps of reality but they are never the terrain.  The more of a Wicked Problem a model is intended to represent, the less likely it is to produce a usefully true forecast.  

If you are relying on a model to forecast a Wicked Problem, it is a fool’s errand.  There might be some intellectual entertainment involved and possibly some refinement of skills but fundamentally, the more wicked the problem and the further out in time, the less likely that there will be any empirical validity to the forecast.

And what are examples of wicked problems?

Economic problems

Market prices

Environmental problems

Political issues

Global Climate Change

Public Healthcare

Epidemics

Criminal behavior

Social problems such as homelessness and substance abuse

War

Basically, anything that is natural or social in nature has a reasonably high probability of being a Wicked Problem.

We do reasonably well when we define problems down to discrete issues with material knowledge about their cause-and-effect nature.  We have to experiment a lot, involve stakeholders, maintain transparency, but with a small enough problem and those sorts of conditions, we can make progress.

Something like Climate Change?  No way are the models producing anything that is usefully true in the near or long term.  The entire climate system is a Wicked Problem for which we have little reliable data on the time scales we are interested in and even less cause-and-effect knowledge.  

Certainly, we know bits and pieces, but we do not understand how the whole system works together.  We are still not even completely confident we know the key long-cycle effects.

In public discussions, people have a tendency to wave the results of models and claim that the science is obvious.  The model may be more or less transparent (usually much less) but it is not science.  All models of Wicked Problems, of complex systems, necessarily involve making a large number of assumptions.  It is those assumptions which shape the nicely precise graphs and statistics of the model.  But unless there is clarity about the assumptions being made, agreement on the assumptions, and great confidence in the validity of the assumptions, then the forecasts, pretty and precise as they might appear, are inescapably not accurate and not useful.

Nadashe, Amit, and Ghreni think they have a sure thing with a model that is estimated to be 85% accurate and a less than one in 100,000 chance about being wrong about the broad outlines.  But it is a flawed model and they don’t know that because they don’t know the assumptions and they don’t know the maths.  All they know is the apparently precise odds it produces that fit their plans.  

But Count Claremont makes the point that the map is not the terrain.  The model is only as good as its constituent assumptions.  And in their case, all the constituent assumptions are made by a single person confident in their work regardless of her errors.  

And that is the second key point of this story and Claremont is explicit about it.  All modeling of Wicked Problems has to be done openly with clear assumptions that are agreed to by all stakeholders, rigorously reviewed for basic errors, and frequently updated with new data.

That is why the Classical Liberal model will always overcome closed authoritarian systems in the long run.  The open competitive market of property and ideas is constantly updating and ideally transparent, motivated by everyone benefitting from a system that orients towards efficiency and effectiveness.

Blindly believing secret models which are highly disputed is no pathway to progress.