By Jon W. Hansen, FCIPS | Procurement Insights · November 2026
This week, David Shillingford shared a reassessment that deserves wide attention. Chris Gaffney and the Georgia Tech Supply Chain and Logistics Institute went back to the simulation they built in early 2026 for a closure of the Strait of Hormuz, and asked what it got right and what it missed.
The model got the physical cascade right: naphtha and LPG first, then petrochemicals, then fertilizers, then finished goods, with Asia feeling the effects first, then Europe, and the United States later, through diesel. What it missed, in Georgia Tech’s words as David summarized them, was “human ingenuity at scale under pressure”: the shadow tanker fleet growing faster than projected, alternative routes absorbing more than their design capacity suggested, insurance markets restructuring instead of retreating, China’s strategic reserves masking demand, and Iran calibrating its limited capabilities to restrict flows.
In my comment, I said that what Georgia Tech calls human ingenuity is agent behavior: strands changing the conditions under which the other strands operate. David replied with a line that has stayed with me:
“We tend to ‘close the book’ on events too early.”
He is right. And the reason it matters is that the events end, but the strands they crossed do not.
The storm faded. The strands it crossed did not.
When Hurricane Isaias made landfall on 9 October 2026, it was the story. Within days it had faded. But the strands it crossed kept moving. A diesel deal with Russia, announced hours before landfall, became a policy strand. The sanctions waiver followed a week later. The war in Iran carried the load throughout. And the effects kept traveling outward, toward food prices in import-dependent countries and toward brewers whose fuel and energy hedges are now beginning to roll off.
That is what strands do, and following them is the discipline we call Strand Commonality™. Its origins go back to 1983, and it was proven in our 1998 work for Canada’s Department of National Defence, where tracing the strands, rather than the documented process, revealed why next-day delivery was failing. In practice, it means recognizing which strands run through an operation, where they cross, and what they have in common when they produce an outcome.
That is what strands do. Some are primary: they run continuously, like fuel prices, policy and supplier capacity. Others are intermittent: they fade in, cross the primary strands, and fade out again. A storm is one. Close the book when the intermittent strand fades, and you miss everything it set in motion.
Strands fade in and fade out after Hurricane Isaias. Bright and thick: a strand carrying the load. Faint: fading out, or not yet arrived. Illustrative, not measured data.
Keeping the book open
Keeping the book open is not a figure of speech for us. It is how our labs work.
On 17 October 2026, Hansen AI Labs™ opened the Isaias Reverberation Lab, to follow the strands after the storm had passed. Before any outcome was known, the lab locked its baseline: eight strands (a ninth was added the same day), seven dated expectations, the rules for scoring them, and a fingerprint so no one can quietly change what it said. Since then, the lab has added a dated entry at every check-in. It changes nothing it said before; it only adds.
The second animation shows that record as it stands. It is not a finished exhibit. It is continuous, human-set tracking: built only from what the lab has actually recorded, and updated at regularly scheduled check-ins, with the check-in periods for monitoring each strand set by the human orchestrator, not the system.
The Isaias Reverberation Lab: active tracking, through Entry 7 (29 October 2026). Solid points are recorded events and readings; hollow points are approximate or have not yet happened; the shaded area has not yet been recorded.
Two things in that record matter as much as the readings.
The record catches its own misses. At the first check-in, the lab found that the energy ceasefire it listed at baseline had already broken two days before the baseline was locked, because the conflict strand had not been swept that day. At the second, it found that it had relied on an incomplete copy of an earlier lab’s record. Both are recorded as corrections, and both changed how the lab works: every locked document and lab record is now checked against its fingerprint before it is cited.
The record admits its limits. Tracking on a schedule is not the same as watching every moment. When Iran’s Revolutionary Guard declared the Strait of Hormuz closed on 26 October, the lab recorded it on 29 October, three days later, and logged the lag as a finding. That honesty is what makes the rest of the record worth trusting.
This is what Georgia Tech did in public, and what most models never do: make the misses as visible as the successes.
Where the book opens: the archive
A lab that starts tracking on the day it opens is still closing the book on everything before that day. The book has to open earlier, and for us it opens in the archive.
Procurement Insights has been published for nineteen years without sponsorship, and its posts are published once and never edited. That archive plays four roles in the tracking.
It is where the strands start. Before the Reverberation Lab tracked anything forward, it traced its strands back: through a 2020 interview with Export Development Canada’s chief economist, a 2020 webinar on Middle East supply, a March 2026 post naming the strands visible before the war, and our March 2026 paper, When the Strait Closes. One of the lab’s dated expectations descends directly from that March post.
It proves when something was recognized. Because the posts are dated and never edited, anyone can check what was said, and when. That is the difference between recognizing a strand before it crosses and explaining it afterward.
It preserves time- and date-stamped data integrity. The archive is checked before any claim about the past is made, checking regularly for errors as well as confirming what was said. Finding and recognizing past errors, which are usually facts that were true at the time, is incredibly important in a contemporaneous archive, both to identify trends and to achieve optimal accuracy. Nothing in it is changed; any correction is added as a new, dated entry in the current post.
It holds primary evidence no one else has. That includes physical records, such as a cassette of a January 2001 Ottawa radio broadcast on which Nortel’s chief financial officer told listeners the future was very bright, four weeks before the company warned of a loss.
Anyone can build a model that tracks strands from today forward. Very few can trace those strands back through two decades of dated, unedited record. The tracking shows the method working now. The archive shows it has been working, on the record, for a long time.
What this means for supply chain leaders
David closed his post with advice for supply chain leaders. I would add five steps to keep the book open. Together, they are Strand Commonality™ in practice:
Trace back before you track forward. Start from the outcome you are living with, and follow its strands back through your own records, not only your forecasts.
Name the strands before the next crossing, including the ones that seem unrelated.
Set the cadence yourself. Decide how closely each strand should be watched. A fast-moving strand, like a conflict, needs more frequent checks than a slow one.
Record what happens without rewriting what you expected. The outcome is new evidence. It is not permission to revise the original understanding.
Keep the misses visible. A model that learns is one that shows where it was wrong.
Georgia Tech deserves credit for closing nothing early. The storm ended. The strands it crossed did not, and neither should our attention.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
-30-
References
David Shillingford, LinkedIn post on the Georgia Tech reassessment, November 2026. [Link to David’s post]
Georgia Tech Supply Chain and Logistics Institute, reassessment of its 2026 Strait of Hormuz simulation. https://lnkd.in/gm3tzFnf
Georgia Tech Supply Chain and Logistics Institute, Modeling COVID-Style Shortage Risk from a Prolonged Iran Shock, May 2026. https://lnkd.in/g_Xqc9Sp
When the Strait Closes, Hansen Models™, March 2026. [Link]
Right Decision, Wrong Result: What a Hurricane, a Sanctions Law, and a Diesel Deal Mean for Every Supply Chain Going Forward, Procurement Insights, 17 October 2026. [Link]
The New Risk Management Mindset: Stop Focusing on Disruptions and Start Following Their Strands
Posted on November 2, 2026
0
By Jon W. Hansen, FCIPS | Procurement Insights · November 2026
This week, David Shillingford shared a reassessment that deserves wide attention. Chris Gaffney and the Georgia Tech Supply Chain and Logistics Institute went back to the simulation they built in early 2026 for a closure of the Strait of Hormuz, and asked what it got right and what it missed.
The model got the physical cascade right: naphtha and LPG first, then petrochemicals, then fertilizers, then finished goods, with Asia feeling the effects first, then Europe, and the United States later, through diesel. What it missed, in Georgia Tech’s words as David summarized them, was “human ingenuity at scale under pressure”: the shadow tanker fleet growing faster than projected, alternative routes absorbing more than their design capacity suggested, insurance markets restructuring instead of retreating, China’s strategic reserves masking demand, and Iran calibrating its limited capabilities to restrict flows.
In my comment, I said that what Georgia Tech calls human ingenuity is agent behavior: strands changing the conditions under which the other strands operate. David replied with a line that has stayed with me:
“We tend to ‘close the book’ on events too early.”
He is right. And the reason it matters is that the events end, but the strands they crossed do not.
The storm faded. The strands it crossed did not.
When Hurricane Isaias made landfall on 9 October 2026, it was the story. Within days it had faded. But the strands it crossed kept moving. A diesel deal with Russia, announced hours before landfall, became a policy strand. The sanctions waiver followed a week later. The war in Iran carried the load throughout. And the effects kept traveling outward, toward food prices in import-dependent countries and toward brewers whose fuel and energy hedges are now beginning to roll off.
That is what strands do, and following them is the discipline we call Strand Commonality™. Its origins go back to 1983, and it was proven in our 1998 work for Canada’s Department of National Defence, where tracing the strands, rather than the documented process, revealed why next-day delivery was failing. In practice, it means recognizing which strands run through an operation, where they cross, and what they have in common when they produce an outcome.
That is what strands do. Some are primary: they run continuously, like fuel prices, policy and supplier capacity. Others are intermittent: they fade in, cross the primary strands, and fade out again. A storm is one. Close the book when the intermittent strand fades, and you miss everything it set in motion.
Strands fade in and fade out after Hurricane Isaias. Bright and thick: a strand carrying the load. Faint: fading out, or not yet arrived. Illustrative, not measured data.
Keeping the book open
Keeping the book open is not a figure of speech for us. It is how our labs work.
On 17 October 2026, Hansen AI Labs™ opened the Isaias Reverberation Lab, to follow the strands after the storm had passed. Before any outcome was known, the lab locked its baseline: eight strands (a ninth was added the same day), seven dated expectations, the rules for scoring them, and a fingerprint so no one can quietly change what it said. Since then, the lab has added a dated entry at every check-in. It changes nothing it said before; it only adds.
The second animation shows that record as it stands. It is not a finished exhibit. It is continuous, human-set tracking: built only from what the lab has actually recorded, and updated at regularly scheduled check-ins, with the check-in periods for monitoring each strand set by the human orchestrator, not the system.
The Isaias Reverberation Lab: active tracking, through Entry 7 (29 October 2026). Solid points are recorded events and readings; hollow points are approximate or have not yet happened; the shaded area has not yet been recorded.
Two things in that record matter as much as the readings.
The record catches its own misses. At the first check-in, the lab found that the energy ceasefire it listed at baseline had already broken two days before the baseline was locked, because the conflict strand had not been swept that day. At the second, it found that it had relied on an incomplete copy of an earlier lab’s record. Both are recorded as corrections, and both changed how the lab works: every locked document and lab record is now checked against its fingerprint before it is cited.
The record admits its limits. Tracking on a schedule is not the same as watching every moment. When Iran’s Revolutionary Guard declared the Strait of Hormuz closed on 26 October, the lab recorded it on 29 October, three days later, and logged the lag as a finding. That honesty is what makes the rest of the record worth trusting.
This is what Georgia Tech did in public, and what most models never do: make the misses as visible as the successes.
Where the book opens: the archive
A lab that starts tracking on the day it opens is still closing the book on everything before that day. The book has to open earlier, and for us it opens in the archive.
Procurement Insights has been published for nineteen years without sponsorship, and its posts are published once and never edited. That archive plays four roles in the tracking.
Anyone can build a model that tracks strands from today forward. Very few can trace those strands back through two decades of dated, unedited record. The tracking shows the method working now. The archive shows it has been working, on the record, for a long time.
What this means for supply chain leaders
David closed his post with advice for supply chain leaders. I would add five steps to keep the book open. Together, they are Strand Commonality™ in practice:
Georgia Tech deserves credit for closing nothing early. The storm ended. The strands it crossed did not, and neither should our attention.
Truth Is Believing. Accuracy Is Knowing. Outcome Is Proof.™
-30-
References
Share this:
Like this:
Related