You Can't Fix the Old System — Build at the Edge

There is a failure mode every operator who has tried to “transform” a company from the inside knows in their bones: the work is sound, the demo lands, the pilot ships — and then it quietly dies. Not killed by a villain. Starved, re-scoped, reabsorbed, out-argued by a more defensible line item on someone else’s spreadsheet. The instinct afterward is to try harder next time: better change management, more executive air cover, a crisper deck. That instinct is the mistake. You cannot patch a system whose own logic is rejecting you. You build the replacement next to it, at the edge, and let the new thing become the center of gravity.

This is a research report on that strategy — when edge-building actually beats internal transformation, and the honest cases where it doesn’t. The framing comes from Peter Diamandis’s conversation on the “Organizational Singularity” (EP #258), where the prescription is blunt: “you cannot change and fix and transform the existing company,” so “you have to build a new system at the edge.” That’s the seed; the rest is what the evidence says.

The idea, stated plainly #

The lineage runs through Buckminster Fuller. The relevant line — “You never change anything by fighting it; you change things by making them obsolete through superior technology” — was first documented in writing by Mike Vance in Think Out of the Box (1995), attributed to a conversation with Fuller before his 1983 death. The exact phrasing varies across sources, so treat it as a well-traveled paraphrase, not a verbatim transcript (Quote Investigator). The strategic content is unambiguous regardless of wording: don’t fight the incumbent system, render it irrelevant. Applied to your own org, that “incumbent system” is you — your processes, your reporting lines, your habits.

The reason you fight yourself and lose has a name. Birkinshaw and Ridderstråle coined the “corporate immune system” in a 1999 International Business Review paper on how multinational subsidiaries get initiatives suppressed. The primary article is paywalled, so I’m relying on the Wikipedia entry on the corporate immune system, which corroborates the concept and its attribution. The mechanism is what matters: organizations don’t merely neglect novel initiatives, they actively suppress them — through resource competition, risk aversion, and inter-divisional conflict dressed up as legitimate critique. The antibodies don’t look like sabotage — they look like good governance asking reasonable questions until the thing bleeds out.

Why “do it from the inside” is structurally improbable #

Two bodies of evidence make this concrete.

First, the base rates. Bain & Company’s April 2024 analysis found that 88% of business transformations fail to achieve their original ambitions (Bain & Company). McKinsey’s long-standing estimate for organizational change failure sits around 70% — a separate figure from separate research, not part of Bain’s number, but pointing the same way. Whatever the denominator, inside-out is a coin flip you lose most of the time.

Second, the spreadsheet logic. Christensen’s Innovator’s Dilemma explains why incumbents fail, and it isn’t ignorance — it’s rationality. Redirecting capital to a higher-margin existing product is always the more defensible board decision than funding a lower-margin disruptor. Blockbuster could never fully commit to streaming while clinging to expensive physical stores, because every quarter the stores looked like the responsible bet (Christensen Institute). That’s the trap: the immune response is correct by the org’s own metrics, and you can’t out-argue a correct spreadsheet. You have to move the work somewhere it doesn’t reach.

There’s a quieter symptom of the same disease in the R&D data. Innventure, citing the EU Industrial R&D Investment Scoreboard 2024, reports that only about 20% of patents from the world’s top 2,000 companies show clear signs of commercialization (Innventure). (That’s Innventure summarizing the Scoreboard, not the Scoreboard itself — and it’s the global top 2,000, not the Fortune 500.) Even secondhand, the shape is telling: enormous invention, most of it stranded inside organizations that can’t carry it to market because anything outside the mandate loses its sponsor.

The pattern that works: a small unit at the edge #

The people who escape the immune system describe nearly identical playbooks.

Salim Ismail — author of the Exponential Organizations framework — learned this the hard way at Yahoo’s Brickhouse lab, where he says he spent roughly 80% of his time defending his teams from the parent before it was dismantled anyway. His six-point formula for surviving the antibodies is concrete: target new or emerging markets only, report at the CEO level, establish physical and structural independence, and recruit internal change agents (Salim Ismail). Separation isn’t a perk there, it’s the mechanism — the unit has to sit outside the reach of the metrics that would otherwise kill it.

The canonical proof is the IBM PC. In 1980, IBM tasked William Lowe and Don Estridge with building a personal computer in a year — flatly impossible under standard IBM procedures. The answer was a skunkworks in Boca Raton, Florida, far from headquarters in Armonk, New York, with Estridge reporting directly to CEO Frank Cary, free to negotiate outside partnerships and skip the usual product-review gauntlet. The PC shipped on August 12, 1981, roughly thirteen months later, and became IBM’s fastest-growing business (IEEE Spectrum). Then comes the part that belongs on every transformation charter: once IBM reabsorbed the division into standard structure and procedures, it lost the market it had created — to Compaq and the clone makers its own open architecture had enabled. The edge unit produced the breakthrough; the mothership killed the advantage. Edge-building isn’t a one-time stunt — it’s a posture you have to keep.

Nestlé’s Nespresso rhymes with this. Nestlé patented its capsule system in 1976 and brought Nespresso to market in 1986, eventually growing it at roughly 30% per year over a decade (Wikipedia). It’s often told as a struggling idea that only thrived once given organizational distance from the instant-coffee core. The dramatic version — that it nearly died three times and was physically moved out of the Nestlé building to escape standard business rules — isn’t documented in the source I could verify, so I’m not leaning on it. The defensible claim is narrower and still useful: a long incubation, then success once the product had room to run on its own terms.

The honest counter-case #

If edge-building were a law, there’d be no counterexamples. There are — and the strongest is Satya Nadella’s Microsoft.

When Nadella became CEO in February 2014, the stock had been flat for a decade and Windows 8 had alienated users. He did not spin Azure out as a separate company. He transformed from within — shifting the culture from “know-it-all” to “learn-it-all,” embracing Linux, and cannibalizing Microsoft’s own on-premises Windows revenue. Revenue grew from about $86.8B in 2014 to roughly $212B by 2023 (Windows News / AI). (Bigger revenue figures and specific Azure-penetration stats get floated around this story; I couldn’t verify those against this source, so I’m leaving them out.) The headline stands on its own: a genuine inside-out transformation that worked at enormous scale.

But the conditions limit how far you can generalize. Nadella arrived as an effectively-outsider CEO with a board mandate. The move was to cloud — an adjacent market, not a disruptive non-adjacent one. Microsoft’s enterprise relationships were assets in cloud adoption, not liabilities to be shed. And it took the full decade before compounding was visible. Internal transformation is feasible precisely when those stars align: top-down mandate, adjacent target, an installed base that helps rather than hinders. The edge thesis holds most firmly for mid-tier incumbents facing genuinely disruptive, non-adjacent threats — where none of Nadella’s tailwinds are present.

A second nuance cuts even closer to the bone. A 2023 study comparing spinouts to internal new establishments found that internal ventures actually start about 2.3x larger and remain roughly 50% larger at age seven (PMC). Parents pour real resources into the bets they keep. The implication: internal isn’t structurally doomed. The edge approach wins not because inside-the-walls execution is impossible, but because the selection filter gets corrupted when politics, mandates, and resource competition override merit. Guarantee an honest filter and sustained sponsorship, and internal is often the better-resourced path. You usually can’t — that’s the whole problem.

What to actually do #

The synthesis is narrower than the slogan, and more useful. Build at the edge when the threat is disruptive and non-adjacent, when the new economics undercut your current margins, and when you have no credible way to keep an internal effort from being out-argued by the core. Give the unit the three non-negotiables every successful case shares: a target outside the existing market, a direct line to the CEO, and genuine separation from the metrics that would otherwise trigger the immune response. And remember IBM: reabsorbing it on the mothership’s terms is how you hand the win back. The point was never to fix the old system — it was to make it obsolete, then keep the discipline not to let it swallow the thing that did.