Springboks Make 13 Changes for Australia: Erasmus Tests Marx at Flanker in a Perth Scrum Audit

September 22 and a team sheet I had to read four times

On September 22 — the original report does not specify the year, and I will return to that detail at the end of this piece — South Africa's team for the test against Australia in Perth was published. I read it four times. Not because it was hard to understand, but because one line in it refused to sit still.

Thirteen positions in the starting XV changed from the previous outing. That previous outing, the Springboks beat New Zealand 43-28.

The detail that stopped me was an option head coach Rassie Erasmus left open: if Ox Nche comes on, Jan-Hendrik Wessels could be pushed up to hooker, and Malcolm Marx — the best lineout thrower South African rugby has produced in a decade — could shift out to play as a flanker.

A fifteen-man team in which the front row is no longer a fixed block. The entire story of this match lives there, and it was written before the ball was kicked.

In my trade, people tend to read a team sheet the way they read a scoreboard. I read it the way I read a balance sheet. Thirteen changed lines do not tell you whether a team is strong or weak. They tell you what the coaching staff believes it can risk without paying for it.

Context: from Baltimore to Perth

To understand why a team at its peak would dare change thirteen players, you have to read the earlier part of the story.

According to the original report, South Africa came into this match on the back of a series win over New Zealand by 3-1 in the fixture block the media branded Rugby's Greatest Rivalry, staged in Baltimore. The most recent match in that sequence finished 43-28 in the Springboks' favour. These are facts supplied by the source, not yet cross-checked with independent data, so I flag them as requiring verification.

The match against Australia is a one-off test, played away in Perth. The opponent is coached by Les Kiss, newly installed as Wallabies head coach. And Kiss has a prior working history with Erasmus.

Those three facts — the recent series win, the one-off nature of the fixture, and a newly appointed opposing coach — combine into a very specific configuration. One side arrives as a team holding the highest prestige and holding the licence to experiment. The other arrives as a team in transition that needs a marker.

I follow international rugby the way I follow the football transfer market in Turin: I do not read statements, I read structures. A coach can say anything in a press conference. But when he signs his name to a team sheet with thirteen changes, that is an action, and actions have a price.

What is the price here? The risk of losing control at the scrum in a match where the opponent wants to plant a flag. Erasmus knows that. And he signed anyway.

What a hooker actually does: dissecting the most misunderstood position

Before discussing the option of Marx at flanker, the hooker position itself has to be reconstructed, because most readers who follow football and read about rugby will skip this detail. And skipping it renders the whole story meaningless.

A hooker is the man in the middle of the front row, wedged between two props. In the scrum, he absorbs direct compression from behind as all eight forwards drive forward, while also hooking the ball with his foot when it is fed into the tunnel. In the lineout, he is the thrower — a movement whose margin of error is measured in centimetres across ten metres, executed under time pressure and in noise.

In other words, the hooker is the only position on the field that combines three unrelated skills: absorbing compression at extremely short range, controlling the ball with the foot in a chaotic situation, and throwing accurately at medium range while stationary. No other position in the sport demands such fragmentation of skill.

Malcolm Marx performs all three at world-leading level, and has for years. That is why pushing him out to flanker is not a gimmick designed for attention. It is a statement about the kind of player Erasmus wants on the field in the sixtieth minute.

A flanker, on the other side, is one of three loose forwards. This is the most mobile group in the pack: they join open play, contest the breakdown, cover the fringes of the scrum, and cover ground comparable to the backline. A good flanker in this sport runs further than a football centre-back, and absorbs more collisions.

So what is the Marx-to-flanker conversion really? It is taking a player trained to absorb compression inside one square metre and placing him into thirty metres of space. Physically, it is a test of converting explosive energy into repeatable endurance. Tactically, it is South Africa manufacturing a fourth forward capable of contesting the breakdown without losing scrum capability.

And here is the point I want to hold onto: that conversion is only worth anything if the scrum remains contested. If reconfiguring the front row forces South Africa into uncontested scrums, they have not lost a player — they have lost a weapon.

The doctrine of positional elasticity: Erasmus's signature

According to the original report, this is not the first time Erasmus has spoken about training players in multiple positions. He is credited with a long-standing cross-position development philosophy. In this team sheet, traces of that philosophy appear in several places: Marco van Staden is mentioned for his ability to switch between loose forward and hooker, while Wessels — a prop — is considered for the hooker role.

I call this the doctrine of positional elasticity, and it has very clear accounting logic.

A national team is constrained by how many players it can call up and how many training sessions it can hold. Under those conditions, every multi-position player is a double saving. First, he occupies one squad slot but supplies two options. Second, he allows the coach to change structure mid-match without spending a substitution on restructuring.

At national-team level, substitutions are the scarcest asset during a match. Positional elasticity is how you buy extra substitutions without paying a fee.

But this doctrine carries a cost that is rarely discussed. A player trained in two positions often fails to reach maximum sharpness in either. In football, I have seen this hundreds of times in the transfer market: full-backs who can play both flanks are usually worse defensively than those who play only one. Flexibility is an insurance policy, and every insurance policy carries a premium.

The right question for Perth is not whether Erasmus is clever. He is. The right question is: what premium did he pay for flexibility this time, and at what minute is it settled.

The arithmetic of thirteen changes

Thirteen positions changed in a fifteen-man side. Two positions stayed the same, or nearly so.

The popular reading is: this is a match being taken lightly. That reading is correct on outcome but wrong on operation.

What actually happens when a national team changes thirteen players is a depth audit. A depth audit does not aim to find the strongest XV. It aims to answer a different question: if three core players are injured over the next six months, who steps in, and how wide is the gap between that man and the man he replaces.

That question can only be answered by a real match. No training session reproduces the pressure of an away test match.

The second notable point is bench structure. According to the original report, South Africa left Malcolm Marx — a frontline hooker — on the bench. In the modern operating model of rugby, the bench is no longer where unfinished players sit. It is where second-half weapons sit.

Leaving Marx on the bench means Erasmus is planning for a long match. He wants a fast starting XV to wear the opponent down, then to release a breakdown-contesting player as Australia's pack begins to slow. That is a design by time, not by lineup.

And here is my first signature line for this piece: Nobody calls Croatia a miracle when every one of them ran 400 km on Russian soil. People call a working structure a miracle when they have not read its running data. South Africa in Perth are the same. Thirteen changes are not arbitrariness. They are the output of a development system invested in for years.

Mobility as a design principle

In the quoted remarks, Erasmus speaks of a genuinely mobile side, and names front-row forwards with a movement-oriented profile. The front row is mentioned as Grobbelaar, Wessels and Porthen. Eben Etzebeth is retained as the anchor.

This structure is easy to read if you look at it as a system rather than a list.

Etzebeth is the physical anchor point. He is not the most mobile man in the pack. He is the man who guarantees that the scrum and lineout still have a stable axis to lean on while those around him move. In risk-management terms, Etzebeth is the low-volatility asset inside a high-volatility portfolio.

The other three names in the front row are the volatility. If they are more mobile than traditional props, they can join open play on the edges, support continuity attacks, and — most importantly — sustain a higher defensive tempo in the final thirty minutes.

This is where many people misread modern rugby. The strength of a pack is no longer measured by mass. It is measured by the ability to repeat high intensity over a long duration. A pack of 130 kg per man can dominate the first twenty minutes and vanish in the last twenty.

Erasmus reads it the other way round: if Australia plays high tempo and keeps the ball alive through structure, the opposing pack will tire first. To exploit that, South Africa itself must have a pack that can run.

This inverts conventional logic. People usually assume a scrum-dominant team imposes the scrum. Erasmus appears to think a scrum-dominant team imposes the tempo.

Australia, Les Kiss and the structure that keeps the ball alive

In the original report, Erasmus describes the key contest by referencing how Australia defends, how they fill the field, and how they keep the ball alive through structure.

This is the most notable passage in the entire report, and I want to read it the way I read a scouting report.

When a coach mentions three elements — defence, filling the field, keeping the ball alive through structure — he is saying his opponent does not win through the scrum. They win through continuity.

That is the kind of team that makes heavy packs pay. Every continuity phase forces the pack to travel from one contest point to the next, and the physical cost of that does not show on the scoreboard until it appears suddenly.

Les Kiss's recent appointment adds another variable. A new coach usually brings two things: a structure the opponent has not yet decoded, and a short-term psychological effect inside the squad.

The shared working history between Erasmus and Kiss complicates both sides. The two men know each other at the level of method. That means tactical surprise is reduced — but it also means both know exactly where the other's weaknesses lie.

In football, I see this pattern constantly. Two coaches who once worked together tend to produce matches that are suffocated of ideas, because both know where the trap has been set.

Which metrics are worth tracking, and which are just decoration

This is the section for those who genuinely want to read this match through data.

The original report supplies no performance metrics at all. No scrum success rate, no metres carried, no tackle completion rate, no turnovers won. That is a significant information gap, and I flag it as such.

But knowing which metrics need measuring is a skill independent of having the metrics available. For this match, I would track four groups.

Group one is the scrum. Specifically, win rate on own scrum, and the number of scrums driven backwards. If South Africa changes front-row structure mid-match, this is the only metric that tells the truth about whether the experiment worked.

Group two is the lineout, specifically the throwing accuracy of the thrower. If Wessels has to throw while converting from prop to hooker, that rate is a direct measure of the premium I mentioned above.

Group three is the number of continuity phases Australia generates per half. If that number rises over time, Erasmus's conditioning plan has failed.

Group four is breakdown turnovers, split by half. This is the metric that directly evaluates whether leaving Marx on the bench was correct.

And this is where I have to be blunt about something this industry repeats far too often: some metrics look beautiful and say nothing.

In football, it is distance covered. A midfielder who runs twelve kilometres a match may be a man running to the right places, or a man running twelve kilometres to the wrong ones. In rugby, the equivalents are metres carried and tackle counts.

A player who repeatedly carries twenty metres may be a man breaking defensive structure, or a man receiving the ball in a position the pack had already opened up. A player with twenty tackles may be an outstanding defender, or a man so consistently out of position that he has to make covering tackles.

Neither metric answers the quality question on its own. They have to be read alongside field position, timing within the match, and the success rate of the ensuing phase.

That is why I never write a piece on a single metric. One metric is a question. Three metrics read together are an answer.

The contrarian angle: correlation is not causation

This is the part I enjoy most when writing about any sport, because it is where most analysis goes wrong.

The popular reading of this match is: South Africa just beat New Zealand 3-1, they are at their peak, they changed thirteen players, therefore they have depth and will win comfortably. That chain of reasoning has three links, and all three are correlation, not causation.

Link one: beating New Zealand 3-1 leads to the licence to experiment. The truer version is the reverse — a long and intense series creates the need to rest frontline players, and that need is the cause of rotation, not a reward for winning. A good result merely makes rotation easier to explain in public.

Link two: many changes mean a weaker team but it does not matter. This is the most dangerous link. Thirteen changes do not merely weaken the side individually, they weaken it in cohesion. In rugby, cohesion in the pack is built through hundreds of hours of scrumming together. Assembly time is a variable that cannot be replaced by individual talent.

Link three: a more mobile team beats a structured team. This is an assumption, not a law. The mobile team wins when the match is played at high tempo. The structured team wins when the match is dragged down into set-piece warfare and contest.

And here is the variable the original report exposes but does not emphasise: motivational asymmetry. Erasmus said Australia will certainly want to put down a marker. That is the polite phrasing of something very specific — the visiting side is preparing for a long cycle, the host side is preparing a statement.

In sporting history, that configuration — a strong side rotating against a transitioning side with a new coach at home — is the configuration that produces the most upsets. Not because the strong side is weaker. Because they are playing a different match from the one their opponent is playing.

Back to the story of the stadium. The empty stadium of 2026 left behind a warning sign that very few read in time: when context changes, every old forecasting model loses value simultaneously. Perth will not be empty. But its context differs from Baltimore, and anyone using Baltimore results to forecast Perth is blending two different data samples into one calculation.

The rules risk: front row and uncontested scrums

There is a technical detail in Erasmus's plan that I want to separate out, because it is a real risk and it is rarely discussed.

World Rugby's laws state that a team wishing to contest a scrum must have sufficient front-row trained players, and must have a replacement option for that position. If a team no longer has a qualified front-row player to bring on, the scrum becomes uncontested.

This is a technical regulation that sounds administrative, but its tactical consequences are large.

An uncontested scrum removes one of rugby's most powerful attacking weapons: the right to drive the opponent backwards, the right to win penalties at the set piece, and the right to control the tempo of the match from a fixed situation.

If South Africa shifts Marx to flanker while only one specialist hooker remains on the field, and if that man has to leave, they could fall into a situation forcing uncontested scrums. At that point, the entire advantage a team with a pack as strong as South Africa's holds would evaporate inside a single substitution decision.

I rate this a medium risk, not high, because Erasmus has clearly accounted for a fallback by leaving Marx on the bench. But it is a real risk, and it belongs to the category of risks people only notice after they happen.

In my transfer-market governance work, I call this a hidden structural risk. It is not in the player. It is in how the clauses are written. A badly drafted release clause can destroy the value of an asset in thirty seconds, and nobody sees it until someone picks up the phone.

On a data-infrastructure error

There is a detail from the very process of analysing this article that I want to record, because it relates directly to my trade.

The original report was misclassified at the first processing layer, labelled as football while the entire content concerned rugby union. That is a labelling error. But its consequences are not small: if that error propagates into an automated system, it will generate a football analysis of a rugby match, and nobody will catch it because the label looks plausible.

In ten years of working with market data, I have learned that most serious mistakes do not come from wrong analysis. They come from a wrong label at the bottom layer, faithfully inherited by every layer above it.

That is also why I always label my own statements. Which part is a measurement. Which part is a qualitative observation. Which part is inference. Those three must not be mixed in one sentence, and absolutely must not be mixed in one metric.

In a meeting room full of men in Italy in 2026, I learned that the market trades in seating positions too. And I learned something else: when a room of similar people reads the same number, they will misread it in the same way. Diversity in a data room is not a moral issue. It is a quality-control measure.

Contracts, agents and the hidden cost

I cannot write about international rugby without addressing the subject I address in every football market piece.

In rugby, there is no transfer window in the football sense. But there is an equivalent bargaining structure, and it is far harder: player release between national unions and clubs in England, France and Japan.

That negotiation has no transfer fee. It has time. And in elite sport, time is the only currency that cannot be printed.

When a national-team coach has to build a squad for a one-off test in Perth, he is not merely managing fifteen players. He is managing a network of release agreements, club fixture lists, and injury risk that the owning club will bill to his account.

That is the largest hidden cost in this sport, and it almost never appears in the news.

A beautiful transfer contract usually begins with a phone call in which both sides stay silent before naming a price. With national teams, that call often begins with an email asking about the condition of a knee that nobody wants to name.

The blind spot of cohesion

There is something in this match that no metric can measure, and I would argue it decides the outcome more than any positional experiment.

That is cohesion between men who have never played beside each other.

In rugby, understanding within the pack is not an abstract concept. It is knowing exactly at which angle the man beside you will drive, at which beat he will release, in which direction he will rotate his hips in a collapsing scrum. Those things are learned with the body, not with video.

Thirteen changes create a great many new pairings. Every new pairing is a point where communication has to be established from zero.

This is why I am always sceptical of analyses that simply count changes. Counting changes is counting independent variables, whereas in reality they are interacting variables. Thirteen individual changes are not dangerous. Thirteen changes arranged in a configuration that has never trained together is dangerous.

And this is where my second signature line appears, in its rugby version: Nobody calls a pack a machine when it has never once driven together.

What is actually being tested

If I had to compress this entire match into a single question, that question is not whether South Africa wins.

The question is: can South Africa's multi-position development system produce a pack capable of changing structure mid-match without losing set-piece capability?

That question has value beyond Perth. If the answer is yes, South Africa will possess a structural advantage no other elite team holds: the ability to change the shape of the team mid-match without paying in set-piece power.

If the answer is no, they will learn something equally valuable: where the limits of the elasticity doctrine lie. And knowing the limits of a doctrine is the first step in turning it into a mature one.

In my valuation work, I never assess a player by asking how good he is. I ask in how many situations he is good, and in how many he is bad. The same logic applies to a coaching doctrine.

Signals to track in the next cycle

After Perth, there are five signals I will note in my book, and they matter more than the final score.

First, the officially announced front-row configuration. If Marx genuinely starts at flanker, that is a statement about the future of the South African pack, not just an option for one match.

Second, scrum success rate and the number of scrums driven backwards. This is the most direct measurement of whether the experiment left a bill.

Third, the minute Marx enters. If he comes on in the fifty-fifth rather than the seventieth, it means Erasmus misread the tempo and had to correct the plan.

Fourth, Australia's intensity in the first twenty minutes. If they open with a long continuity sequence, the motivation Erasmus feared has materialised.

Fifth, and this is the signal I care about most: Erasmus's post-match response. If he talks about what was learned, the doctrine is evolving. If he talks about the result, the doctrine is being defended.

The difference between those two answers is greater than any scoreline.

An unfinished thought

In a report thousands of kilometres away from me, with no year specified, someone recorded a small decision: leaving open the possibility of pushing a hooker out to flanker in a test match that does not decide a crown.

Small decisions like that are usually where sporting systems get rewritten. Not in finals, but in Saturday test matches away from home, where a coach holds enough credit to risk trying something nobody has tried at this level.

What I will carry away from this story is not the Perth result. It is a different way of framing the question for every team I follow: if a national side is deep enough to change thirteen players and still hold its structure, then its advantage does not lie in the fifteen men on the field. It lies in the entire system behind those fifteen.

And when someone calls that a miracle, I only have one answer.

There is no miracle here. Only a system operated long enough that it begins to look like one.