While the Skies Are Still Calm

• "Its role would be purely exploratory."
• It would not replace the board or the leadership team.
• It would not have governance authority over any aspect of the school's operations.
• "It would produce thinking for the board, not decisions on behalf of it."
• Its function: "to expand the range of possibilities the board is equipped to consider, and to do so with intellectual rigour, independence, and a longer horizon than normal governance cycles allow."
• The author explicitly rules out lighter-touch alternatives: "This challenge does not lend itself to a workshop, a strategy day, or an additional item on the board's existing agenda. It requires sustained, independent, structured inquiry."

• Size: six to eight individuals — "small enough to function as a genuine thinking community and diverse enough to avoid reinforcing existing assumptions."
• Suggested areas of expertise (as listed in the paper):
  1. Artificial intelligence and machine learning, particularly as applied to learning systems and knowledge work
  2. Higher education, including admissions, curriculum design, and the future of credentials
  3. Labour market economics and the changing nature of professional capability
  4. Educational innovation, including experience with non-traditional learning models in an Australian context
  5. Organisational strategy and institutional transformation
  6. Access to a consultative voice from within the Somerset community — a staff member, parent, or alumnus invited to contribute context at the group's discretion, without formal membership or voting standing

• Independent, with credibility in institutional strategy or education policy, and "no current affiliation with Somerset College."
• Rationale given: "The Chair's independence is important: it ensures the group's conclusions are not shaped by the school's own institutional interests."

• Reports directly to the board, via the Chair, on a schedule agreed at establishment.
• The Headmaster serves as an ex-officio observer — present for institutional context and operational grounding, but "not a voting or directing member."
• Explicitly not filtered through school management: "the work be independent, rigorous, and connected directly to the board rather than filtered through the school's management structure."
• Cadence of reporting, scope of mandate and terms of engagement are all left as matters for the board to determine.

• None. Purely advisory/exploratory. Produces thinking, not decisions.

1. How might AI reshape the process of learning, the nature of teaching, and the assessment of capability over the next decade? What does this mean specifically for an institution like Somerset?
2. How might university admissions evolve in Australia, and what implications does that evolution have for the ATAR-based value proposition of independent schooling?
3. What are the most credible alternative models of education now emerging globally and within Australia? What can be learned from them, and what do they suggest about the direction of travel?
4. What strategic options should Somerset's board consider — "ranging from incremental adaptation to more fundamental reinvention, each examined on its merits"?

The author closes by minimising the commitment required: "The above is offered not as a finished blueprint but as a starting point for discussion… this paper asks only that the conversation begin." And: "It does not ask the board to change anything about Somerset College today. It asks only that the board create a serious, independent mechanism for thinking carefully about what Somerset College should be tomorrow. That, at a minimum, seems like the responsible thing to do."

Argument: Pre-emptive disarmament. The paper is "not a prediction about the future of education, nor a criticism of the current success of Somerset College." It is a contribution to a conversation about how institutions prepare for change.
Support: "Institutions that thrive during such periods are rarely those that assume they already understand the future. More often, they are the ones that create space to explore possibilities before change becomes unavoidable." Sets out the three-movement structure and flags the think tank proposal up front.

Argument: Somerset is genuinely successful — and that is the uncomfortable part.
Support: Strong enrolment demand, respected academic reputation, committed staff, supportive families, consistently strong outcomes, a waiting list. Then the pivot: "Institutions are often at their most vulnerable when they appear strongest." Christensen is introduced: market leaders "rarely fail because they are badly run. More often they struggle because they continue improving a model that was designed for a world quietly changing around them."
The central question posed: not whether Somerset is strong today ("It plainly is") but "whether the model that has made it strong will remain equally valuable in the world now emerging."

Argument: AI is not another delivery technology; it is the first technology that can simulate instruction itself.
Support: "Radio did not reinvent school. Television did not reinvent school. Personal computers did not reinvent school. Even the internet… largely wrapped itself around the traditional architecture of schooling." Class sizes, timetables, subject structures and age-based progression survived all of them intact. By contrast, "An AI tutoring system does not merely present content — it responds, adapts, diagnoses misunderstanding, and adjusts its approach in real time," making a skilled teacher's functions "replicable at scale and at negligible cost."
Four simultaneous implications identified: (1) how students learn and how efficiently core knowledge is acquired; (2) how assessment functions — AI both assists production of work and complicates evaluation of independent capability; (3) how universities select and employers evaluate; (4) the deepest one — "what is the purpose of school in a world where knowledge itself has become instantly accessible and where a personalised tutor of extraordinary capability sits in every student's pocket?"
Closing move: that question "does not need an answer today. But it does need to be asked" — and the fact it can be asked seriously "for the first time in the history of modern schooling" is itself the signal.

Argument: Disruption is a consequence of competent management, not incompetence — and it changes the category, not just the product.
Support: Christensen's mechanism laid out in four beats: (1) new models emerge at the margins, "simpler, cheaper, and less prestigious"; (2) incumbents correctly judge them inferior and rationally focus on existing customers — "That rational response is also, often, the fatal one"; (3) the new model improves "incrementally at first, then more rapidly" until it is "good enough" for a meaningful segment; (4) because incumbents have spent years telling a coherent story about why their quality justifies their premium, "the moment of recognition often comes later than it should."
Christensen's "most important insight" per the author: "disruption does not always occur because an existing product becomes worse. More often it occurs because the definition of the category itself changes. The market that an organisation has perfected is no longer the market that matters."
Two worked examples: BlackBerry/iPhone and horse-and-cart/motor car (see §7).
Application to schools: the stable decades-old proposition — "invest in premium independent education, receive strong academic outcomes, secure favourable university pathways, improve life chances" — is stress-tested with three "what ifs": AI dramatically improving outcomes outside elite environments; university pathways fragmenting away from ATAR; a few schools proving a radically different model gets equivalent results with a richer experience.
The hinge: parents stop asking "which school has the strongest academic record?" and start asking "what exactly am I paying for, and is it still the best use of this investment?" The author calls this "the dangerous question" — dangerous not because every family defects, but because "once enough families begin to think differently about the value equation, the assumptions underpinning premium independent schooling enter a period of genuine uncertainty."

Argument: The waiting list is a lagging indicator, not a forward signal.
Support: "A waiting list reflects decisions parents are making today… It is evidence that the existing value proposition is trusted — not evidence that the underlying drivers of that trust will remain stable." Historical parallels: "Video rental stores maintained high utilisation in the years before streaming became mainstream. Print newspapers maintained circulation while online readership was still thin." The warning: "By the time the change appears in the headline metrics, the window for strategic response has often narrowed."
The three assumptions the author says actually drive demand: that ATAR remains the primary pathway to university success; that employer value of independent-school credentials remains high; that no credible alternative model exists. If these soften, "waiting lists may prove to be a more delayed signal than reassuring ones."

Argument: Alpha School is not a model to copy; it is proof that the assumptions underpinning traditional schooling are already being seriously questioned.
Support: Alpha School, founded in Austin, Texas, now expanding into a national network, built on a "2 Hour Learning" model — core academics compressed from six or seven hours into a much shorter block using adaptive software and AI-supported personalised tutoring, with the rest of the day given to "projects, entrepreneurship, workshops, physical development, creativity, and life-skills development." Adults function "less as subject-delivery teachers and more as guides and coaches."
The provocative claim stated: if technology delivers core knowledge more efficiently, most of the school day can be reallocated to "communication, leadership, creativity, resilience, and the capacity to work on complex, ill-defined problems."
Notably even-handed on evidence: "Early reports from Alpha suggest strong academic results and unusually high levels of student engagement. These claims deserve healthy scrutiny — much of the published performance data originates from the organisation itself, and independent validation at scale remains limited." Legitimate questions flagged about "teacher roles, screen time, and socialisation." In August 2026, Alpha announced twenty-seven new campuses for the coming school year — "one of the largest single expansions in its history" — while "investigative reporting has sharpened questions about the independence of its performance claims and the governance of its network."
The author's framing device: "Neither development should surprise us. Rapid scaling and rough edges are precisely what the early stage of a disruptive model looks like." Alpha is "Version 1.0." Comparison drawn: "The first automobile was slow and unreliable. The first commercial aircraft was noisy, uncomfortable, and limited in range. What matters in the early stages of disruption is not whether Version 1.0 is perfect. What matters is the direction of travel, and whether the underlying logic of the model is sound." He concedes Alpha 2.0/3.0 "may look very different… But they will follow from it."
Relevance stated explicitly: "The relevance for Somerset is not imitation. It is awareness… Once alternative models become imaginable, the foundations of traditional premium schooling are no longer beyond question."

Argument: The university is the keystone of the entire value chain independent schooling is built on — and its erosion is already measurable, pre-dating AI.
Support — the Gallup series: US public confidence in higher education fell from 57% in 2015 to 36% in 2023 — "a collapse of nearly a third in under a decade." It "recovered modestly to 42% in 2025 before slipping back to 38% in mid-2026." The author is careful: "The path is not a straight line, but the direction across the decade is unmistakable." Cause given: not incompetence, but people "questioning whether the proposition — the time, the cost, the credential — was still worth what it once was."
The author's sharpest analytical point: "Almost all of that decline occurred before the current generation of AI systems existed… Artificial intelligence, in other words, did not cause this erosion. It has arrived on top of it — a genuinely new force acting on an already weakened proposition."
Support — public expectation of AI's effect: In 2026, for the first time, Gallup asked Americans whether AI will change the value of a university degree. 46% believe AI will make degrees less important over the next five years; only 20% believe the opposite. The author's inference: the paper's central question "is no longer confined to strategists and futurists. It has become the public's question."
The unbundling argument: a degree has been "a bundled proposition: knowledge acquisition, professional credentialling, employer signalling, and peer networks, packaged together and delivered over three to four years at significant cost." AI unbundles all four simultaneously — knowledge now near-free; credentials "fragmenting into micro-credentials, portfolios, and demonstrated competency"; employers (major tech firms and "increasingly, professional services companies") "quietly dropping degree requirements for roles where capability can be assessed directly."
Support — labour market: In the US, "unemployment among recent graduates now exceeds the rate for the workforce as a whole — a reversal with almost no historical precedent — with the decline concentrated in early-career roles most exposed to AI." In Australia, "researchers at the University of Sydney have found that junior lawyers are entering a squeezed market because AI now performs many of the tasks on which young professionals traditionally learned their craft." Verdict: "The bundle is coming apart."
Why it matters to Somerset: "The logic has always run in one direction: strong school, strong ATAR, strong university, strong career." If the destination becomes structurally uncertain, the risk is not merely changed admissions — it is that "the question parents are really paying to answer, what is the best investment I can make in my child's future?, may no longer have the same answer it has had for the past fifty years."

Argument: Australia's lack of an Alpha equivalent is not reassurance — it is a lag. And unbundling has already begun here.
Support — the absence argument: "Australia does not yet have a true Alpha School equivalent… That absence might appear reassuring. It should not be read that way." Regulation, cultural conservatism and parental caution may delay innovation "but it is unlikely to be prevented indefinitely." The key warning: "by the time a true Alpha analogue is clearly visible in Australia, the window for thoughtful strategic response will have narrowed considerably. The danger to elite schools is not that an Alpha School equivalent has already arrived in Australia. It is that boards may wait for a perfect analogue before recognising that the logic of disruption is already underway."
Support — the bundle: elite independent schools sell "academic excellence, personal development, wellbeing support, university pathways, peer networks, and social capital, all delivered through a single institution at a single premium fee." The threat "is not necessarily that a competitor will offer a better version of that bundle. It is that the bundle itself may begin to come apart."
Support — Australian exemplars:
- Big Picture Education Australia — a network across multiple states using "personalised learning plans, real-world mentorship placements, and portfolio-based assessment." High student-reported engagement; outcomes "different in character from traditional academic metrics" but "taken seriously by universities developing broader admissions frameworks." Author is careful: "Big Picture does not threaten Somerset directly. But it demonstrates that alternative models can earn credibility within the Australian context."
- Ravenswood (NSW) — shows "ambitious curriculum innovation and strong academic outcomes are not in tension"; a high-performing independent school can embrace different approaches "without sacrificing the academic results that parents value."
- Caveat stated: "Neither Big Picture nor Ravenswood is the full expression of the Alpha thesis."
Support — ATAR erosion: "Non-ATAR entry has quietly moved from exception to mainstream." In New South Wales, admissions centre data indicates more than a quarter of first-year university students now enter through non-ATAR routes. In Victoria, more than nine hundred courses were open to non-ATAR applicants for 2026 entry. The University of Queensland, Griffith University, and Bond University all offer alternative admissions pathways to a growing range of degrees. Consequence: "If ATAR becomes one of several viable routes… one of the most reliable arguments for investing in premium academic preparation weakens."
Support — fee economics: "independent school fees across Australia have continued to rise at rates significantly above inflation," now "a substantial proportion of median household income, and that proportion has been growing." The squeeze: "fees are rising while the exclusivity and necessity of the traditional pathway may be declining."
Measured conclusion: "None of these developments constitutes a crisis. But together they constitute a signal worth taking seriously."

Argument: Separate core values (enduring) from operating models (evolvable) — and never defend the second as though it were the first.
Support: "Great institutions… endure not because they preserve every feature of their current form, but because they distinguish clearly between what is essential and what is structural. Core values endure. Operating models evolve. The fatal mistake is to confuse one for the other — to protect structural features as though they were values, and in doing so to sacrifice adaptability while believing you are protecting identity."
Somerset's core, as the author names it: development of character and moral courage; cultivation of intellectual curiosity and rigour; depth of human connection between students, teachers and families; the experience of sustained challenge and growth; community, belonging and service; and the formation of young people capable of contributing meaningfully to the world.
The reassurance (and the strategic argument): "These things are not going out of date. In fact, they may be the most important argument for institutions like Somerset in the age that is coming." What AI cannot replicate: "a teacher who notices a student struggling and knows exactly what to say; the moment a young person discovers they are capable of more than they believed; the formation of character through sustained challenge, genuine community, and relationships built on trust and care… In a world of increasing automation, these qualities become more valuable, not less."
Somerset's structure, as the author names it: age-based cohort progression; fixed timetables with subject-specialist teachers delivering content in discrete blocks; the school day as the primary container for learning; ATAR as the dominant metric of student success; the primacy of teacher-directed instruction over student-directed inquiry. "These are not Somerset's core values. They are structural features — inherited architectures that made sense within a particular technological and social context."
The board question Collins implies: "not whether to change — but which parts of the school's design are genuinely core to its identity, and which parts are structural conventions that may need to evolve." And critically — a direct set-up for the think tank — "It is not a conversation that can be had quickly, or comfortably, within normal governance structures."
Collins' second principle invoked: great institutions "confront the brutal facts of their reality without losing faith in their capacity to succeed." Hence: "The appropriate response… is neither panic nor complacency. It is disciplined, honest inquiry."

Argument: Capable boards fail at this for structural, not personal, reasons.
Support: Draws on governance scholar David Beatty and decision scientist Annie Duke, combined with Christensen, to name four traps (detailed in §8). Duke's concept of "hidden consensus" is cited specifically: "people assume that because an idea sounds implausible to them, it is probably implausible to everyone else in the room. Dissenting views go unspoken."
The diplomatic passage (important for how this lands in the room): "none of what follows is a reflection on the capability of the Somerset board. Boards of this quality… are precisely the ones best placed to engage seriously." But: "A board that has governed a successful institution through a period of stability is, by the nature of that experience, less likely to have built the external perspectives, the specialist knowledge, and the independent thinking space that structural disruption demands."
The load-bearing conclusion: "The four traps described above are not character flaws. They are the predictable consequences of success… the range of futures a board can seriously consider from within its normal structures is narrower than the range of futures that may actually arrive. This is precisely why the thinking described in the remainder of this paper cannot be done entirely from within the institution — not because the board is ill-equipped, but because no board, however strong, should have to carry that work alone."

Argument: The title metaphor. Do the emergency thinking while calm, because under pressure you won't think as well as you expect.
Support: The author's own flying lessons. "Flying safely is not primarily about controlling the aircraft… The real discipline lies in situational awareness and preparation." Pilots scan the horizon constantly. But the specific discipline cited is the pre-take-off briefing: thinking through engine failure shortly after leaving the runway. "If the engine fails below roughly 1,000 feet, the safest option is almost always to land straight ahead in whatever open ground is available. Above that altitude, there may be sufficient height to complete a circuit and return to the runway." The principle: "the thinking is done before the emergency occurs… if the engine actually fails there is no time for leisurely analysis… And under those conditions, people rarely think as clearly as they believe they will."
The governance mapping: "Boards do not fly the aircraft day-to-day — that is the role of the leadership team. But they carry responsibility for scanning the horizon, asking difficult questions, and ensuring that the institution has thought carefully about what it would do if the conditions that currently support its success were to change more quickly than expected."
The title line: "it may be wise to do the thinking while the skies are still calm." And the key timing argument: "The governance traps described in the previous section are hardest to overcome during a crisis. The conditions that permit genuine strategic imagination are the comfortable ones — which means that now, when Somerset's position is strong, is precisely the right time to ask the harder questions."

Argument: A concrete, year-by-year narrative is more persuasive to a board than abstract principle. Framed carefully: "This is not a prediction. It is a plausible sequence of developments, each of which is already underway in some form." Full timeline in §9.

Argument: A no-regrets / insurance case. Even if the thesis is wrong, the work pays.
Support: An explicit climate change analogy — "not for the reasons it might first appear." There remains disagreement about precise mechanisms and timelines, but many of the proposed responses — "reducing pollution, improving energy efficiency, developing cleaner technologies, building more resilient infrastructure — are beneficial regardless of whether the most alarming projections prove correct."
Concessions made (a genuine strength of the paper): "It is entirely possible that the traditional model of premium independent schooling will prove more resilient than this paper suggests. Universities may maintain heavy weighting on ATAR for longer than anticipated. Parents may continue to trust established pathways. Alternative models may take longer to reach credibility than the early experiments suggest. These are not implausible outcomes."
The asymmetry argument: even then, the work — "thinking carefully about Somerset's core versus its model, tracking emerging developments in education and technology, stress-testing the assumptions underpinning the school's value proposition" — still strengthens the institution. "The cost of doing this thinking is modest. The cost of not doing it, in a scenario where the pace of change accelerates, could be significant."

Argument: The proposal itself. Full detail in §4 above.
Support — why a think tank specifically: the forces are "complex, multidimensional, and moving simultaneously… artificial intelligence, labour market transformation, evolving university admissions, alternative learning models, changing parent expectations, and shifting fee economics. Each of these developments is significant on its own. Together they form a system that is difficult to interpret from within the normal demands of school governance." Precedent cited generically: "Large institutions facing complex, structural change often respond to this kind of challenge by establishing a dedicated body whose sole purpose is to think deeply about the future."

Argument: The closing restatement — values sound, structures dated, timing is now.
Support: "Its values… are not going out of date. In an age of artificial intelligence and rapid technological change, those qualities may be more important, not less. The question this paper has attempted to raise is a different one. The values are sound. But the structures through which those values are currently expressed were designed for a world that is beginning to change."
The urgency framing: "the time to think carefully about it is now — not when the waiting list begins to shorten, not when a competitor announces a new model, not when university admissions systems have already shifted."
The final thesis: "The institutions that will navigate the next decade of educational change most effectively will not necessarily be the largest, the most resourced, or the most prestigious. They will be the ones that thought clearly about the future before the future arrived. That kind of thinking does not happen by accident. It requires structure, independence, and the willingness to ask questions that are, at least initially, uncomfortable."

Argument: Don't take my word that AI is different — go and experience it. Full detail in §10.

The paper's citations are named but not footnoted. Gallup figures, the NSW admissions-centre share, the Victorian course count, the US graduate unemployment reversal and the University of Sydney junior-lawyer finding are all asserted with an attributed body but no reference, date-stamp or link. The Alpha School data is explicitly flagged by the author himself as largely self-originated and lacking independent validation at scale — a credit to his candour, but it means the single most vivid exhibit in the paper is also its weakest-sourced.

Strong institutions develop "a coherent and compelling story about why they succeed." For Somerset the author suggests that story includes academic excellence, strong community, experienced staff, clear values and a long history of producing capable graduates — "These are genuine strengths."

The failure mode: "a powerful success narrative also creates a filter. Ideas or scenarios that challenge the model can feel instinctively implausible — not because they are wrong, but because they do not fit the story the institution has learned to tell about itself. The narrative becomes a lens that sharpens some possibilities and blurs others."

"Boards are particularly susceptible to this dynamic, and for reasons that are not straightforwardly a product of poor governance. Directors are typically thoughtful, accomplished, and mutually respectful. Meetings are structured around consensus and collegial exchange."

The failure mode: this environment — "in many ways a sign of good governance" — "can also make it surprisingly difficult for ideas that feel radical, unlikely, or uncomfortable to gain genuine traction."

Duke's "hidden consensus": "people assume that because an idea sounds implausible to them, it is probably implausible to everyone else in the room. Dissenting views go unspoken. The range of futures the group seriously considers quietly narrows."

"Disruptive change often requires imagining futures that do not yet exist. Human beings naturally project the future by extending the patterns they already understand. We assume tomorrow will resemble today, with incremental variations. But technological change does not always follow incremental patterns. Sometimes it alters the rules of the system itself."

The failure mode: "the hardest step is not implementing change. It is genuinely imagining that such change is possible in the first place." And the sting: "Institutions with long histories and strong identities often find this kind of imagination unusually difficult, because their own success is evidence that the current model works."

"In board settings, where relationships and collegiality matter, genuinely dissenting views can go unspoken — not because people have no concerns, but because they assume the group has already reached a conclusion, or because challenging a prevailing narrative feels like disloyalty to a strong institution."

The failure mode: "convergence on comfortable interpretations of the future rather than serious exploration of uncomfortable possibilities."

"when the issue involves complex, multi-dimensional technological and societal change, even capable, well-intentioned boards may find it difficult to explore the full range of possibilities within normal governance structures. The challenge is not intelligence or commitment. It is the structure of the conversation itself."

And the direct bridge to the proposal: a board that has governed a successful institution through stability is "less likely to have built the external perspectives, the specialist knowledge, and the independent thinking space that structural disruption demands… no board, however strong, should have to carry that work alone."

Note on traps 2 and 4. These overlap substantially — groupthink and polite agreement are close cousins, and "hidden consensus" is cited under trap 2 while trap 4 describes much the same silence. A sceptical reader may regard this as three traps stretched to four. See §11.

A two-page appendix (pp.24–25) inviting the reader — a director — to personally test the paper's central claim that AI systems are "capable of something qualitatively different from previous technologies," before forming a view about it.

• "The exercise takes less than three minutes."
• "Copy the prompt below and paste it into any of the following: Claude (claude.ai), ChatGPT (chatgpt.com), or Gemini (gemini.google.com). All are free to access and require no account for basic use."
• "Read the response carefully. Then send the follow-up prompt and read that response too."

"What you are about to experience is not a search engine returning links, nor a database retrieving stored answers. You are about to give a single paragraph of instruction to a system that will produce, in under thirty seconds, a piece of strategic analysis that a skilled consultant might take two days to assemble. The quality of that experience — not the description of it — is the point. It is difficult to read the response with genuine attention and come away believing that what these systems can do is irrelevant to an institution in the business of developing human knowledge and capability."

You are a senior strategic adviser with deep expertise in educational disruption, AI systems, and institutional strategy. I want you to conduct a specific exercise.

Imagine it is 2033. A well-regarded Australian independent school — high fees, strong ATAR results, long waiting list, proud tradition — has just convened an emergency board meeting. Enrolment inquiries for the following year are down 22% on the prior year. Three competitors have introduced AI-supported hybrid learning models in the past four years. Two Queensland universities have moved to portfolio-based admissions for their most competitive faculties. A new school has opened on the Gold Coast offering a fundamentally different learning model at fees 30% lower, and its first cohort has produced ATAR results within five points of the traditional schools.

The board is asking: how did we get here, and what did we miss?

I want you to write the retrospective strategic analysis that an honest adviser would present to that board. What were the early signals they should have acted on between 2024 and 2027? What assumptions did they hold that proved wrong? What decisions — or failures to decide — created their current vulnerability? And what are their realistic options now?

Be specific. Be honest. Do not soften the analysis to protect feelings. Assume the board is intelligent and capable of confronting difficult truths.

Now step back from 2033. It is 2026. The same board has just read your analysis as a hypothetical. What is the single most important thing they should do in the next twelve months that they are probably not doing?

1. It bypasses the imagination gap (trap 3) by substituting demonstration for argument.
2. It is a pre-commitment device. A director who runs the exercise has personally experienced the capability before debating whether it matters.
3. The follow-up prompt is a funnel. Asking an AI "what is the single most important thing they should do in the next twelve months that they are probably not doing?" will, with high likelihood, return an answer resembling establish independent, structured strategic inquiry capacity — i.e. the paper's own proposal, now arriving in the director's own browser rather than from a fellow director. This is rhetorically elegant and worth naming openly in the boardroom. See §11.
4. The scenario figures are fictional. The 22% decline, the 30% fee gap, the five-ATAR-point gap, the three competitors and two universities are constructed inputs for the exercise, not claims about the world. Directors running the prompt should not later recall them as data.

1. It makes a proportionate ask. The gap between the scale of the threat described and the size of the remedy requested is unusually modest. A think tank of 6–8 people with no authority is cheap, reversible and non-threatening. The paper never over-reaches into "Somerset must change X." That restraint is the strongest thing about it as a governance document.

2. The Collins core/structure distinction is the right de-escalation device. By ring-fencing character, curiosity, human connection, community and formation as untouchable core — and explicitly arguing these become more valuable under automation — the author removes the identity threat that usually kills this conversation. Directors can accept the analysis without feeling the school is being attacked.

3. It is genuinely honest about its own weakest evidence. The author volunteers that Alpha's performance data "originates from the organisation itself," that "independent validation at scale remains limited," that there are legitimate questions about "teacher roles, screen time, and socialisation," and that investigative reporting has questioned the governance of the network. Most advocacy papers bury this. He also concedes outright that the traditional model may prove more resilient than he suggests, and calls those outcomes "not implausible."

4. The Gallup "AI didn't cause this" observation is the paper's best original analysis. Noticing that almost all of the confidence collapse (57% → 36%) occurred before capable AI existed — "Artificial intelligence… did not cause this erosion. It has arrived on top of it — a genuinely new force acting on an already weakened proposition" — is a genuinely sharper insight than the usual AI-disrupts-everything framing. It also makes the argument robust to AI underdelivering.

5. The climate-change no-regrets analogy gives sceptics an exit. A director who thinks the whole thesis is overblown can still vote yes, because the proposal is framed as cheap insurance rather than a bet on a forecast. This is good persuasion architecture.

6. The scenario is calibrated, not catastrophic. Somerset survives 2033 with "solid" enrolment. The loss is optionality. Much harder to dismiss.

7. The flying analogy earns its place. It is first-person, specific (the ~1,000 ft decision height), and it does real work: it converts present strength from a reason to defer into the reason to act.

1. The Australian evidence is notably thinner than the American evidence. The heavy artillery — Gallup confidence, graduate unemployment reversal, Alpha School — is all US. The Australian material is comparatively soft: Big Picture (which the author concedes "does not threaten Somerset directly"), Ravenswood (innovation compatible with results — true, but not disruptive), non-ATAR admissions share, and a single University of Sydney finding on junior lawyers. The paper acknowledges Australia has no Alpha equivalent, then argues the absence proves lag rather than difference. That is an assertion, not an argument — and it is unfalsifiable in its current form.

2. The "Version 1.0" move is unfalsifiable. Any criticism of Alpha can be absorbed as "that's just what early-stage disruption looks like." But most Version 1.0s fail — the author says so himself ("Most fail") and then does not apply that base rate to Alpha. The framework is constructed so no evidence about Alpha can count against the thesis. A sceptic is entitled to notice this.

3. Christensen's theory is applied without its known limits. Classic disruption theory requires a low-end or new-market foothold that incumbents are structurally unable to pursue because serving it would damage margins. It is not obvious that Somerset is so constrained — a wealthy independent school with a waiting list can adopt AI tutoring tomorrow without cannibalising anything. The paper never addresses the strongest counter: that AI may be a sustaining innovation for well-resourced incumbents, who have the capital, the staff and the brand to deploy it better than a startup. Christensen's own theory has also been substantively contested in the management literature; none of that appears here.

4. Non-ATAR admission is not clearly the same thing as ATAR irrelevance. Many non-ATAR pathways are equity schemes, early-offer programs, portfolio entry for specific disciplines, bridging and mature-age routes. "More than a quarter of NSW first-years enter through non-ATAR routes" does not establish that a high ATAR has stopped being valuable for the competitive courses Somerset families actually target. The paper treats volume of pathways as evidence of ATAR decline without separating those cases.

5. The four traps are closer to three. Groupthink (trap 2) and the culture of polite agreement (trap 4) describe substantially the same mechanism — unspoken dissent. Duke's "hidden consensus," cited under trap 2, is arguably a better description of trap 4. The framework is padded.

6. No costs, no success criteria, no sunset. The paper says "the cost of doing this thinking is modest" but never estimates it. There is no budget, no proposed term length, no definition of what a successful first term would produce, and no termination condition. For a board paper proposing a new standing body, these are material omissions — and the "board to determine" framing, while appropriately humble, leaves directors voting on an undefined commitment.

7. The relationship with the leadership team is under-specified and potentially fraught. A body that reports directly to the board, deliberately bypasses "the school's management structure," is chaired by someone with no Somerset affiliation, and relegates the Headmaster to non-voting observer, while reserving the right to pull in staff, parents and alumni "at the group's discretion" — that is a structure that could easily be experienced by the executive as a shadow strategy function or a vote of no confidence. The paper does not engage with this risk at all.

8. The counterfactual is missing. The paper asserts that a workshop, a strategy day or an agenda item is insufficient, but gives no reasoning. It does not consider cheaper alternatives: commissioning an external report, adding a director with AI/edtech expertise, a time-limited board sub-committee, joining a sector consortium, or simply directing the Headmaster to report annually on these questions. The think tank may well be right — but it is presented as the only option without the comparison being run.

9. The AI-authorship disclosure cuts both ways. It is admirable that the collaboration with Claude is disclosed on the cover. But a paper arguing that AI produces consultant-grade strategic analysis, which was itself substantially drafted with AI, and whose appendix invites directors to prompt an AI into recommending the paper's own conclusion, is a closed loop. The author would strengthen his position by naming this explicitly rather than leaving a sceptic to find it.

10. Several figures are attributed but not referenced. Gallup, "admissions centre data," the Victorian course count, the US graduate unemployment reversal and the University of Sydney finding all lack citations, dates or links. For a board paper that will be quoted in minutes, this is a practical weakness — directors cannot check or update the numbers.

• "We are being asked to accept that our greatest strength is actually our greatest risk. That argument is structured so that no evidence of success can ever count as reassurance. How would I know if this thesis were wrong?"
• "Every one of the hard numbers is American. Where is the Australian evidence that any of this is affecting enrolment, fees or outcomes at schools like ours — today?"
• "Why is AI a threat to us rather than an advantage? We have the capital, the staff and the brand to deploy AI tutoring far better than a startup in Austin. Christensen's theory is about incumbents who can't respond. Which constraint actually stops us?"
• "Non-ATAR entry includes equity and early-offer schemes. Show me evidence that ATAR has lost weight in the competitive courses our families actually want — medicine, law, engineering at Group of Eight universities."
• "Alpha School's own data is self-reported, it's under investigative scrutiny, and we are told to discount all of that because it's 'Version 1.0'. That is an argument that cannot lose. Most Version 1.0s fail — the paper says so."
• "What does this cost? How long does it run? What does it deliver? How do we shut it down if it isn't working? None of that is in here."
• "How does the Headmaster feel about an externally chaired body that reports over his head and can summon his staff at its own discretion?"
• "Why a standing think tank and not a one-off commissioned report, a board sub-committee with a sunset, or recruiting a director with this expertise? We have not been shown the cheaper options."
• "The appendix asks us to prompt an AI with a loaded hypothetical and then asks it what we should do in the next twelve months. We will be told to set up independent strategic inquiry. The exercise is designed to return the paper's own recommendation."
• "The paper was drafted with Claude and argues AI is transformative. I would like to understand where the author's judgement ends and the tool's fluency begins."

As a persuasion document this is well above average for a board paper: structurally disciplined, intellectually honest about its weak points, rhetorically careful not to threaten anyone's identity, and asking for something small enough that refusing it looks worse than granting it. Its real weakness is not the thesis — which is defensible — but the proposal's under-specification: no cost, no term, no deliverable, no sunset, no comparison to cheaper alternatives, and no engagement with the executive-relations risk. A board minded to say yes should say yes with those parameters attached.

1. Do we accept the paper's central distinction — that AI is categorically different from radio, television, computers and the internet because it can simulate instruction rather than merely deliver content? If we accept it, what follows? If we don't, what is our counter-argument?
2. The paper argues our waiting list is a lagging indicator resting on three assumptions: that ATAR remains the primary university pathway, that employers still value independent-school credentials, and that no credible alternative model exists. Which of those three do we believe is most likely to weaken first, and what would be the earliest observable sign?
3. What evidence would change our mind? Can we specify, now, three to five concrete, observable indicators — in enrolment inquiries, competitor behaviour, Queensland admissions policy, or Gold Coast fee positioning — that would tell us the thesis is playing out? Equally: what would tell us it is not?
4. Is AI more plausibly a threat to Somerset or a sustaining advantage? We have capital, staff quality and brand. What would actually prevent us from deploying AI-supported learning better than a lower-cost disruptor?

5. Working from Collins' distinction: what is genuinely core to Somerset, and what is inherited structure? The paper nominates age-based cohorts, fixed timetables, subject-block delivery, the school day as the container for learning, ATAR primacy, and teacher-directed over student-directed inquiry as structure. Do we agree? Which of those would we actually be prepared to examine?
6. If a Somerset parent asked today, "what exactly am I paying for?" — could each director give the same answer? Is that answer still ATAR-dependent?

7. Which of the four traps — success narrative, groupthink, imagination gap, polite agreement — is most present in this board? Can anyone give an example from the past two years?
8. When did this board last seriously entertain a scenario in which the current model stops working? If the answer is "never," is that evidence of soundness or of the imagination gap?

9. What would this cost, in dollars and in board/executive time, and over what term?
10. What is the first-term deliverable, in what form, by when — and what would make us judge the first term a success or a failure?
11. What is the sunset condition? How and when do we wind it up or renew it?
12. Have we tested the cheaper alternatives — a one-off commissioned external report, a time-limited board sub-committee, recruiting a director with AI/education expertise, or a sector consortium? Why is a standing think tank the right instrument rather than the most ambitious one?
13. Is the Headmaster comfortable with the proposed structure — an externally chaired body reporting directly to the board, bypassing management, with him as a non-voting observer and with discretionary access to staff, parents and alumni? If not, what structure would preserve the independence the paper argues for without creating a shadow executive?
14. Who would we actually appoint as independent Chair, and do we have realistic candidates with "no current affiliation with Somerset College"?
15. How do we prevent the think tank from either (a) being ignored, or (b) becoming a de facto strategy committee? What is the mechanism that converts its output into board consideration?

16. Should we commission an Australian evidence base before or alongside establishment — Queensland admissions trajectories, Gold Coast competitor fee and model positioning, actual weight of ATAR in the courses our families target, and our own enrolment-inquiry leading indicators?
17. Should the school be tracking a small dashboard of the paper's named signals (Queensland non-ATAR share, competitor hybrid-model announcements, Year 7 inquiry volumes, fee-to-median-income ratio) regardless of whether we establish the think tank?
18. The paper is explicit that its figures are attributed but unreferenced. Do we want them verified and date-stamped before this is minuted or circulated more widely?

19. Will every director commit to running the Appendix A exercise before the next meeting? And will we discuss it knowing the follow-up prompt is likely to return something close to the paper's own recommendation?
20. How do we structure the next conversation so dissent is actually expressed? Should we formally assign someone to argue the opposing case — that Somerset's model is more durable than this paper allows — before we vote?