The 75% Signal: Why Winners Miss the Next Curve
A forensic look at why mature-market leaders keep missing the disruptions aimed straight at them — and the diagnostic signal that would have caught each one in time.
By Herbert Roberts, P.E.
I. The Milestone That Started This
In September 2026, Honda Aircraft Company announced completion of the first wing structure for the HondaJet Echelon test unit — a routine-sounding program update, five wing structures now moving through final assembly, mainline assembly targeted for early 2027. On its face, a manufacturing milestone.
Look closer and it’s stranger than that. This is the same company that, sixty years earlier, entered the American automobile market from the bottom with small, reliable, inexpensive cars while Detroit’s Big Three dismissed the segment as beneath serious competition. Honda is now doing something that rhymes with that move — in a completely different industry, with none of the original people, market conditions, or technology in common.
That’s not a coincidence worth a footnote. It’s evidence of something structural, and this article is the case file for what it is.
II. The Mechanism: S-Curves and the 75% Signal
Every mature technology follows a predictable improvement curve: slow early gains, a steep middle climb, then a flattening top where each additional dollar of R&D buys less and less. This is standard S-curve behavior, well documented in TRIZ methodology and broader innovation theory.
The flattening is the tell. Below roughly 75% of a technology’s practical ceiling, pushing the current paradigm harder is still rational — real, comparatively cheap gains remain available. Past that point, the economics invert: marginal cost per unit of improvement rises sharply while the return keeps shrinking. Continuing to push the same metric past this point stops being optimization. It becomes a symptom.
The diagnostic rule this article uses throughout: once a core technology passes roughly 75% of its exhaustion point, an organization should be actively fielding a second effort whose only job is finding a different curve — one starting near zero, where early gains are still cheap — while the first team keeps harvesting what’s left of the original.
What makes this signal hard to catch in practice is that it’s rarely visible from inside the system generating it. An organization measuring its own progress in the units of its own paradigm has no natural way to register movement on an axis it isn’t instrumented to see. That blindness — not a lack of intelligence, not laziness — is the throughline connecting every case below.
III. Case One: Detroit, 1960s–70s
By the mid-1960s, the American automobile industry was a settled oligopoly of three: General Motors, Ford, and Chrysler, competing almost entirely against each other on horsepower and styling. Each confirmed the others’ priorities were the right priorities simply by being the only other players worth watching. That mutual confirmation functioned as market validation, when it was really shared blindness.
Reliability and fuel economy — genuine must-be attributes for a car buyer — were left largely unaddressed, because the captive, brand-loyal American buyer of that era wasn’t demanding them loudly enough to move the metric incumbents were already winning on.
Japanese manufacturers, Honda and Toyota chief among them, entered from a different axis entirely: durability and operating cost, starting near zero on a curve Detroit wasn’t watching. They didn’t out-style Detroit. They made styling temporarily irrelevant to a growing population of buyers who wanted a car that didn’t need constant repair. By the time the 1970s oil shocks made the gap undeniable, the disruption had already been underway for years.
IV. Case Two: The Business-Jet Flagships, Today
A parallel oligopoly of three exists in ultra-long-range business aviation: Gulfstream, Dassault (Falcon), and Bombardier, each racing the others on cabin volume, quiet, and digital cockpits — Kano’s “attractive quality” dimension — because range and speed have converged into a solved, must-be layer that no longer meaningfully differentiates a purchase decision. The Bombardier Global 8000 and Gulfstream G800 now both offer comparable maximum range in the 8,000-nautical-mile class; further gains there don’t move the needle the way a better cabin does.
The tell that this segment has crossed its own 75% mark is visible in the market’s own language: the Dassault Falcon 10X is explicitly marketed as trading a slice of maximum range for what many consider the finest cabin in the sky — a manufacturer consciously spending its remaining development budget on delight rather than the core performance metric, because the core metric has stopped being cheap to move.
Leadership at both companies reinforces the blindness rather than correcting it. Gulfstream’s president, Mark Burns, joined the company in 1983 as a CAD operator and rose entirely from within over more than four decades. Dassault’s CEO, Éric Trappier, joined Dassault in 1984 straight out of engineering school and has spent his entire career inside the company. Neither profile is a failure of individual talent — both are accomplished, respected leaders. But four decades of promotion within a single institution’s internal logic is a structural way to lose the outside eye needed to notice a different curve rising elsewhere. Nationality is a red herring here — Trappier is French-trained, Burns American, and neither pedigree explains the pattern. Tenure inside one incumbent institution is the more likely variable.
The must-be layer this segment has left unaddressed isn’t a mystery: crew cost and operating economics. None of the flagship makers’ current customers are price-sensitive enough to ask for a fix, so it doesn’t register as a gap from inside their measurement system.
V. Case Three: HondaJet, Now
HondaJet’s own sequence, from certification in 2015 to the Echelon program today, follows the identical two-phase pattern Honda’s automotive division used sixty years earlier: solve the must-be layer completely and visibly first, then climb into delighter-tier attributes only once trust in the fundamentals is earned.
Phase one was pure structural engineering: an over-the-wing engine mount configuration — a genuinely novel structural answer that solved cabin noise and drag simultaneously — a composite fuselage that solved weight, and a natural laminar flow wing that solved aerodynamic efficiency. None of these were about comfort or market expansion. They were physics problems, solved to a standard the existing paradigm hadn’t matched, and proven over a decade of real flight hours.
Phase two, the Echelon, is the climb the fundamentals earned: single-pilot certification for a transcontinental-range aircraft, expanding who can buy and operate the jet without a second pilot’s salary and training overhead. Regulators do not extend single-pilot trust to an airframe without a decade of proven structural data behind it — the sequencing wasn’t a choice, it was the only available path.
The team that built this entered from outside the industry’s own gene pool at every level. Lead designer Michimasa Fujino held an aeronautical engineering doctorate from the University of Tokyo but built his early career in Honda’s automotive aerodynamics and NASA computational fluid dynamics research — not as a career business-jet-industry veteran. The company itself was built outside the established aviation cluster: rather than Wichita — home to Cessna and Learjet, with the shared talent pool and supplier networks that come with it — or an established California aerospace hub, Honda located its aircraft headquarters in Greensboro, North Carolina, chosen for available land, workforce, and airport growth potential rather than proximity to incumbent thinking. The parallel to Honda’s automotive history is exact: the company’s first major U.S. plant was built in Marysville, Ohio, not Detroit.
This is not incidental. Co-locating with an established industry cluster imports that industry’s assumptions by osmosis — through hiring, informal socializing, supplier relationships, and the quiet sense of “how it’s done” that any team absorbs from its surroundings. Physical and disciplinary distance is a way of protecting a team’s ability to think differently before day one, rather than trying to unlearn incumbent orthodoxy later.
The compounding effect of this sequencing is what makes the resulting gap hard to close. HondaJet’s advantages — engine mount, composite structure, laminar wing, weight distribution, resulting fuel economics enabling single-pilot certification — are interdependent, not additive. An incumbent cannot bolt any single piece onto an existing two-pilot, non-composite airframe; the architecture was built around a different set of trade-offs from the ground up. Catching up on one dimension means re-deriving the whole structure, not patching the old one.
VI. Case Four: GE and the Ceramic Matrix Composite Lineage — The Control Case
Every forensic case needs a control, and this one matters because it proves detection alone isn’t sufficient.
GE Research Center’s work on ceramic matrix composites identified a correct answer years before it could be acted on: a clear physics case and cost rationale — higher operating temperature tolerance, longer part life, higher turbine efficiency — long before the fiber supply chain existed at the volume needed, or a manufacturing base existed to produce at scale. This is a technology that read its own 0%-curve signal early and correctly.
What separates this case from Detroit and the business-jet flagships isn’t detection failure — it’s institutional override. A funded, technically sound program still faced cancellation pressure for institutional and political reasons unrelated to the underlying physics. The lesson this case adds to the forensic file: correctly reading the signal is necessary but not sufficient. An organization also has to grant the team that found the new curve permission to act on it, and protect that permission from the institutional forces that default to defending the current one.
VII. The Big Why: Why Detection Fails
Pulling the four cases together, detection failure is not one cause. It’s several compounding failure modes, each structural rather than a matter of individual competence:
Incumbents measure the wrong buyer. Feedback loops run against current, already-captured customers, who by definition never ask for the thing that will eventually displace the incumbent — because they are not the buyer the disruption is aimed at.
Internal incentives reward the wrong axis. Promotion and budget flow toward whoever improves the metric the organization already competes on. Nobody advances a career by flagging a small, currently-irrelevant competitor climbing a different curve.
Entry barriers create false security. Certification costs, capital intensity, and brand trust are real defenses against a frontal assault on the current paradigm. They do nothing against an entrant who makes the current paradigm’s battleground irrelevant instead of attacking it directly.
Insider-only leadership loses the outside eye. Long tenure inside a single institution builds deep expertise and simultaneously erodes the capacity to see that institution’s own blind spot — because the blind spot is built into the institution’s own operating logic.
The disruptor’s early product looks weak on the axis being measured. Early HondaJets, early Toyotas, and early Southwest Airlines flights all underperformed the incumbent on the incumbent’s own chosen metric. That’s not an oversight on the incumbent’s part — it’s the correct reading of incomplete evidence, using a measurement framework built to track only one curve.
Segment distance compounds all of the above. A disruptor rarely enters where an incumbent’s customers already shop. HondaJet’s Very Light Jet segment shares almost no customer overlap with the ultra-long-range flagship buyers at Gulfstream, Dassault, or Bombardier — so strong, visible growth in one segment simply doesn’t register as competitive pressure in the other, until the disruptor matures enough to climb toward the incumbent’s territory.
The synthesis: history repeats because market maturity manufactures the same four ingredients — a fully served buyer, an incumbent optimized to serve that buyer at rising margins, an underserved buyer at the edge willing to accept “good enough” at lower cost, and an outside capability unlock — every time a market matures, regardless of era or industry. Detection fails because the very apparatus built to keep an organization winning today — customer feedback, internal incentives, competitive benchmarking, leadership pedigree — is structurally calibrated to defend yesterday’s definition of winning, which is precisely the axis a real disruption doesn’t compete on.
VIII. The Antidote: Different Thinkers Never Stop Thinking Differently
Honda’s cross-industry repeat is the case that suggests a genuine antidote exists, and it isn’t mysterious. It looks like an organization that never stopped running a second team — hunting for a curve at zero — regardless of which industry it happened to be operating in at the time.
The practical version of this, assembled from the evidence above, is a short checklist any organization could actually apply, not merely aspire to:
Track the marginal-return curve on your core technology. Past roughly 75% of its ceiling, treat that as a standing trigger, not a warning to note and move past.
Field a second team explicitly tasked with finding a different curve, staffed independently from the first.
Recruit that second team from adjacent disciplines and distant geography — not from the industry you’re trying to out-think. The industry’s own veterans are the people most thoroughly trained not to see what you need them to see.
When a new curve is found, solve its must-be layer completely and visibly before spending anything on delight. A shaky foundation doesn’t just fail to support a delighter — it actively converts the delighter into evidence of misplaced priorities, once the foundation’s weakness becomes visible.
Expect the resulting advantage to compound across several interdependent technologies rather than a single metric. A single-point lead is catchable by a well-resourced incumbent. A five-point architectural lead usually isn’t — catching up requires starting over, not patching the old design.
IX. Close
The companies that get disrupted are rarely the ones who weren’t smart enough to see it coming. They’re the ones whose own success built the instruments that guaranteed they couldn’t.
Herbert Roberts, P.E. is a licensed professional engineer with 32 years in aviation research and development across three companies, and has spent eight years analyzing accidents under his P.E. license.
References
CompositesWorld, “HondaJet milestone reinforces commitment to business jet market advancement,” 2026
Wikipedia, “Honda Aircraft Company”; “Michimasa Fujino”
Pursuitist, “HondaJet: Innovation in the Skies, American Jobs on the Ground”
Forbes, “How The HondaJet Took Flight: An Engineer’s 29-Year Obsession,” 2015
BNN Bloomberg, “Dassault pitches new jet against Bombardier and Gulfstream,” March 2026
Jet Members, “Battle of the Flagships: Falcon 10X vs. Global 8000 vs. Gulfstream G800,” June 2026
The Flying Engineer, “Private Jet Types: Complete Guide,” 2026
Skies Mag; Crunchbase; Georgia Historical Society — Mark Burns biographical profiles
Wikipedia, “Éric Trappier”; Dassault Falcon official biography
Fractional Jet Ownership Insights, “The Best Luxurious Private Jets of 2026”
ResearchGate, “Case Study: Southwest Airlines Strategies”
Simple Flying, “How Southwest Pioneered The Low Cost Carrier Model”; “Why One Of The World’s Biggest Airlines Bet Its Future On A Single Aircraft Family”
CNBC, “Virgin Group: Richard Branson’s companies from records to space travel,” 2019; “Richard Branson...looks to disrupt the industry with Virgin Voyages,” 2021
Virgin Galactic Holdings, Q1 and Q2 2026 financial results and business updates
StockTi, “Virgin Galactic Advances 2026 Flight Plans Amid Narrowed Loss and Cash Concerns”