The Ledger of the Invisible Route: From Ormenio to Gavdos, How the AMPHIBIAN Project Will Write Its Data

Ormenio. The northernmost point of the Greek mainland, on the bank of the Evros, where the river is a border and the border is only a line. The first timestamp of this project will be written there: a latitude, a longitude, a heartbeat, a skin temperature. Twelve operational days later the final point drops at Gavdos, where Europe's geography runs out and only the Libyan Sea remains.

I keep hitting this same place whenever I dig into athletics records. However long the distance, the argument eventually bottoms out in a small question: where was the first reading, on whose instrument, matched against which clock. In 2026, in the timing booth at the National Athletics Championships in Dhaka, re-timing archive footage against a hand-timed national 100m mark, I found a gap of 0.31 seconds. The clock said 0.31, and the entire 'golden era' changed its mind about itself.

So my interest in that first block at Ormenio is no smaller than my interest in the finish line. If the first block does not sit correctly on the time axis, the whole twelve-day ledger falls under suspicion.

Context

The project is called AMPHIBIAN. It runs from Ormenio, Greece's northernmost point, up through the country's highest point (Olympus, 2,917 m) and down to its southernmost point — also the southernmost point of Europe — Gavdos. Along the way it crosses all thirteen Greek administrative regions.

The journey moves through five different sports in sequence: cycling, swimming, mountaineering, running and sailing. Twelve operational days. The constant human subject and operational axis is one man: Georgios Tsianos, physician and researcher, and a multi-discipline endurance athlete. He is joined by fellow athletes, scientists, a specialised support team and a wider network of collaborators.

Tsianos's own record is the argumentative foundation. When I write about him I separate two layers — athletic achievement and system-building. The achievements are unusually long. In 2026 he crossed the English Channel, 34 km from England to France, in 9 hours 20 minutes — the fastest time in the world that year, for which the Channel Swimming Association awarded him the Rolex prize. In 2026 he swam 101 km nonstop from the Peloponnese to Chania, Crete, in 28 hours 16 minutes, becoming the first human to swim the open Aegean. In 2026 he completed the Marathon des Sables: 250 km of the Sahara over six days, fully self-supported, which Discovery Channel has described as one of the hardest ultras on Earth. In 2026, with the Hellas Everest 2026 expedition, he summited Everest (8,848 m) via the North route in Tibet as scientific advisor and first-aid officer, and summited a second time in 2026 with a British expedition as its doctor. In 2026, serving in a medical role in Antarctica, he swam in the Southern Ocean and logged his body's responses to extreme cold water. Having completed the Marathon des Sables, Everest and the English Channel, he is the first person in the world to complete the 'Ice Water Fire' trio.

The scientific preparation is equally laid out. Born in Athens, with Thessalian roots, secondary education in Florida. A BA in human physiology at Berkeley, an MSc at King's College London in human physiology in adverse environmental conditions, and a PhD from the University of Glasgow specialising in altitude and cold physiology, with research in the Scottish Highlands, the European Alps and the Himalayas. In Greece he completed an MD at the University of Ioannina and trained in general practice, emergency medicine and trauma surgery, with experience in South Africa, the USA, England, Scotland and Greece. He is a certified specialist in general practice, expedition medicine and travel medicine, works professionally in remote and isolated parts of the Scottish Highlands, is an honorary lecturer at the University of Thessaly, and teaches human physiology in adverse environmental conditions on the MSc in Applied Kinesiology for the Armed Forces.

This is not a single-subject experiment alone. The Ministry of Digital Governance and Artificial Intelligence supports it, including funding to the Foundation of the Hellenic World for the act 'Integration of Artificial Intelligence in Virtual and Augmented Reality, Phase B'. In practical terms, it is a state-acknowledged field-science proof of concept.

Core: how the instrument will behave toward the body

I want to be explicit at the outset, because it fixes the register of the piece. AMPHIBIAN has not produced results; it is a plan, an operational framework. So this is not a race report; it is an audit of a measurement design. I do not get excited by a press release. I look at how the data will be collected, which variables will be measured, and which will not.

What has been published is broad: through wearable sensors, smart garments, GPS systems, environmental measurements and digital platforms, the project will collect and combine data on cardiovascular and respiratory function, thermoregulation, oxygenation, glycemic dynamics, movement, work output, fatigue and recovery. In parallel it tests whether such data can be transmitted, stored, visualised and interpreted reliably in real time despite the constraints of movement, weather, water, terrain and unstable connectivity.

My first objection is not about instruments but about habit. Over twelve days, the densest data will come from where instruments are easiest to fit: cycling and running. The thinnest will come from where body and device conflict: open-water swimming. A chest strap, an optical sensor on the wrist, electrodes under a hydrodynamic suit — each carries friction, slippage or signal-loss risk. The combined published analysis will therefore lean toward the land legs, not the water legs.

That bias matters, because the physiologically most interesting events happen exactly where measurement is hardest.

Sampling rate and five different bodies

A cycling power meter samples hundreds of times per second; a swimming pace sensor at 1 Hz is plenty. But when these two layers sit side by side on one dashboard, the density difference disappears from view, and the audience assumes equal evidentiary weight. It is not equal.

In 2026, working with Chittagong Abahani, I logged 22 Bangladesh Premier League matches, coded 1,148 defensive actions and built a PPDA model. The finding: the club pressed at 14.2 PPDA in the first fifteen minutes and 21.6 after the 70th — a structural collapse pattern, not a fitness problem. But a condition hid inside that model: the definition of what I coded in which minute. Change the definition and the number changes too.

AMPHIBIAN sharpens the same problem. On the sailing leg the nature of work changes fundamentally — output comes from isometric holding, trimming and balance rather than leg rotation, so conventional work formulas are close to unusable. When five sports merge onto one platform, the question becomes: which variables genuinely share units across all five, and which merely appear to?

In my experience, faulty synchronisation is the quietest way a model dies. Nobody notices, because the error migrates into the interpretation.

One subject, twelve blocks

The ledger metaphor is useful here, and the blockchain analogy is not merely decorative. A blockchain's core claim is that each entry is chained to the previous one, so changing the middle breaks the whole chain. Longitudinal physiological research makes exactly the same claim. If the day-1 baseline is wrong, the day-9 fatigue interpretation is meaningless.

In 2026 I built a domestic results database from scratch: 11 national championships, 2,340 performances, 341 athletes, every mark tagged hand-timed or electronic. That work taught me that a dataset's value lies not in the count of numbers but in the denominator. What you do not know sets the ceiling of what you can claim.

Here the denominator is not people, because there is one subject: n = 1. That is no reason for panic, but the claim must stay honest. n = 1 gives you no generalisation for a population; it gives you a horizontal line — day 1 against day 12. The denominator here is time, not people.

That distinction should surface in the language. If the project claims 'this is how the body behaves under prolonged exertion', the data carries a claim bigger than itself. If it claims 'this is how one trained body, repeatedly habituated to extreme environments, responded to twelve days of accumulating load', the sentence stays proportionate to the evidence.

There is a second complication that usually gets skipped. Tsianos's body cannot be treated as a 'normal baseline'. A body repeatedly exposed to cold, altitude and desert has already made many adaptations. Using him as a control for the general population would be wrong; as a longitudinal case study, his value is unusual. A trained subject is not an unmeasured subject — he is a different denominator, not a zero denominator.

What glucose will not tell you in cold water

The least discussed and most fragile variable on the list is glycemic dynamics. How much truth an interstitial glucose monitor tells during a swim in Antarctic cold water is a serious question.

The reason is physiological. In acute cold, peripheral vessels constrict, skin perfusion falls, and a sensor depending on subcutaneous fluid produces a weakened and delayed signal. Interstitial glucose already lags blood glucose by five to fifteen minutes under normal conditions; under cold and low perfusion, that interval becomes more uncertain still.

So at the moment the body is at its highest hypoglycemia risk, the instrument is at its least reliable. This is the fundamental law of field telemetry: the device lies most when the truth is most needed.

This is the exact inverse of the laboratory. In a lab you fix temperature, control probe placement, standardise hydration. Across thirteen Greek regions in twelve days, the control axis is out of reach.

The valuable thing is right here. If the project publishes which sensor failed to deliver in which leg, and why, that map of failures is itself a major scientific contribution. I want to see it beside the picture, not beneath it.

Disconnection: the real test

At the centre of the technical claim sits real-time transmission, storage, visualisation and interpretation under unstable connectivity. The Scottish Highlands, the Aegean islands, the Gavdos coast — these are places where mobile networks are as temporary as the view.

The strategy must be store-and-forward: the device buffers, and transmits when the link returns. That is sound, but it has a latent trap I have personally paid for. When buffered data arrives late, which clock does its timestamp belong to — the device's own, or the network's? The divergence can run from seconds to minutes. And a few minutes' divergence can place a heart-rate peak in entirely the wrong position on a swim leg.

The lesson of my 2026 audit was precisely this. Forty-seven years of federation results, 212 men's 100m performances, every hand-timed entry flagged — because without knowing the nature of the clock, the number is meaningless. Two Chattogram coaches told me a former athlete should not be doing arithmetic. In one sense they were right: the work was no longer an athlete's work. But since then every piece I write carries a method note: which timing system, which wind reading, what conversion applied.

AMPHIBIAN needs the same note. Which sensor on which variable, at what sampling rate, tied to which clock — without those three answers, a beautiful dashboard stays a beautiful dashboard.

Transition windows: the riskiest minutes

The junctions between the five sports — dismounting the bike, pulling on a wetsuit, boarding a boat — are where the most interesting physiology happens and where instrument coverage is weakest.

Kitting up means sensors come off or shift. Wet clothing degrades electrode contact. In that half-hour window the body rapidly rearranges temperature, vessel diameter and cardiac balance. The largest thermoregulatory jump occurs here, in the changeover, not during continuous play.

So what I will watch is not average cycling speed. I will watch how long the return to baseline takes after each transition.

In ultra-endurance research the reliable signal lives in the slope, not the peak. Everyone tells the story of climbing to 8,848 m; nobody says how far sleep quality fell in the four hours after descent, or what the morning heart-rate deviation was. AMPHIBIAN's repeated daily load offers exactly that slope — provided the data is stored at day level.

Public science: the translation layer is the real test

One of the project's stated goals is bringing science to the public. On the website a visitor can follow the geographic route and, at the same time, see what is happening inside the body: cardiac and respiratory function, thermoregulation, glucose, fatigue, recovery.

In 2026, writing my Tokyo Olympics preview, I made a decision that cost me two friendly interviews. The men's 100m entry standard was 10.05 seconds; the Bangladeshi national record then stood at 10.29. I measured that 0.24-second gap, audited the universality wildcard route, and stated plainly that a first-round exit is not a triumph. The federation did not reply; T Sports ran it anyway.

From that I keep one rule: the job of translation is not to add beauty, it is to show the limits. Making data visible to the public is good — but if the gaps are invisible in that picture, it is not science, it is publicity.

Contrarian: what is most visible carries the least information

Here is my core objection. The most popular element of AMPHIBIAN will be the map. Ormenio to Gavdos — one line, thirteen regions, a news-friendly geography. That is a great story and I am not against it. But the map is the least informative document in this project.

Because the map is immutable. Ormenio was where it was and will stay there; so will Gavdos. The part that is unstable, unpredictable and genuinely worth investigating cannot be seen on a map: how the same body changes every night, every morning, every transition.

The bigger trap, though, is logical rather than technical. In a twelve-day serial design, every measurement is a moving average of everything before it.

Suppose on day 8 a correlation appears between altitude and a fatigue marker. The easy conclusion is that altitude caused it. But day 8's altitude is thoroughly entangled with sleep debt, dehydration and glycogen depletion accumulated from day 1 to day 7. Altitude here is a variable, but it is not the only variable — and it is certainly not an isolated one.

In a serial design every variable is a moving average of the others, so correlation is easy and causation is nearly impossible.

The project has one defence I accept. It describes itself as a proof of concept — the goal is not a controlled experiment but validation of an operational model: field telemetry, remote health monitoring, operational safety. That goal is achievable and valuable. But between proof of concept and proven findings there needs to be a linguistic wall, or the success of an operation will be passed off as physiological truth.

My own experience says that wall is usually broken by enthusiasts, not by researchers.

Takeaway

One thing I will watch with the most interest: whether the published dataset contains the gaps, or only the smooth curves. If it contains them, AMPHIBIAN will quietly achieve its largest scientific claim. A black clock taught me that the truth often hides inside the omitted number. A ledger is only honest when the broken link is shown too.