ExpertsFM(Preview) The App Store in the Shadow of AI, Offensive and Defensive Cybersecurity, Q&A on Financial Planning, AI Writing, American Sports
Sharp Tech with Ben Thompson · August 21, 2026
5 Selected Lessons:
heuristic · 3 reasoning_pattern · 2
Growth masks problems; stagnation magnifies them
Source
Sharp Tech with Ben Thompson — (Preview) The App Store in the Shadow of AI, Offensive and Defensive Cybersecurity, Q&A on Financial Planning, AI Writing, American Sports (August 21, 2026) · at 7:47 ▸ · at 10:21 ▸ · at 17:06 ▸
When this applies
when assessing why internal conflicts over resource-sharing intensify or fade in any organization or market
The takeaway
Distributional conflict intensity is inversely proportional to perceived growth — a shrinking or flat pie turns every allocation decision into a zero-sum war.
Why it works
When the total pie is expanding, everyone's absolute gains outweigh relative disputes, so conflicts over slicing it feel trivial; when growth stalls, the only way to gain is to take from someone else, making distributional fights existential.
One level up
Before diagnosing an organizational conflict as structural or moral, check whether it's actually a symptom of growth stagnation.
Where it stops holding
This stops working when a dispute is about a truly scarce, non-expandable resource even during growth (e.g. safety, trust, or power concentration) where absolute harm persists regardless of overall pie size.
Detail
approach: Ask whether the underlying market/pie is growing or shrinking before judging the intensity of a political or organizational fight over allocation.
signals:
- Fights over percentage splits (e.g. App Store commissions) peaking during periods with no clear growth narrative
- Same disputes losing salience once a new growth story (AI) emerges
- Political/regulatory energy concentrated on 'how to slice the pie' during stagnant periods
rationale: Growth changes the psychology of conflict: shared upside reduces the salience of unfairness.
In their words
This is why growth is so important. This is what the EU gets wrong by and large by undervaluing growth. When things are growing and everyone's getting richer, all the problems that inevitably exist because the world is messy and there's gonna be problems, don't really matter that much and no one cares that much.
Don't let backend harms block frontend value creation
Source
Sharp Tech with Ben Thompson — (Preview) The App Store in the Shadow of AI, Offensive and Defensive Cybersecurity, Q&A on Financial Planning, AI Writing, American Sports (August 21, 2026) · at 17:06 ▸
When this applies
when deciding whether to preemptively regulate/restrict a new platform or venture that has both large upside potential and foreseeable downside risks
The takeaway
When the potential upside of letting something grow unchecked for a while is enormous and the downside is correctable later, prefer delayed correction over early suppression — because early suppression risks losing the upside entirely.
Why it works
Early, aggressive intervention against a nascent platform's bad behavior can also choke off the growth and ecosystem-building that only emerges once the platform is allowed to scale; the upside and the eventual abuses are often two sides of the same unconstrained growth.
One level up
Sequence interventions by reversibility: fix correctable downstream harms after they appear rather than risk foreclosing irreversible upstream value creation.
Where it stops holding
This stops working when the downside is irreversible or compounds faster than any later correction can fix (e.g. harms that permanently foreclose competition or cause damage that can't be undone by late regulation).
Detail
approach: When evaluating a platform under criticism, ask what value it already created and whether earlier intervention would have prevented that value from ever existing.
signals:
- Critics focus only on current abuses without acknowledging the ecosystem/value already built
- Regulatory bodies take years to act, by which time structural shifts have already occurred
- The 'bad actor' behavior is 'part and parcel' of the same dynamics that produced the good outcomes
rationale: If you worry about the bad things too early, the good things may never happen — so the tradeoff must be weighed explicitly, not reflexively condemned.
In their words
And if you worry about the bad things too early, it actually ends up not mattering because the good things don't happen.
Zero marginal cost disputes aren't worth your sympathy
Source
Sharp Tech with Ben Thompson — (Preview) The App Store in the Shadow of AI, Offensive and Defensive Cybersecurity, Q&A on Financial Planning, AI Writing, American Sports (August 21, 2026) · at 7:47 ▸
When this applies
when deciding how much to care about a dispute between two large, already-profitable parties over splitting proceeds from a near-costless digital good
The takeaway
Distinguish fights over creating value from fights over purely capturing existing value; reserve engagement and sympathy for the former.
Why it works
When the good in dispute costs nothing to produce at the margin (e.g. unlocking code, virtual items), the fight is purely about capturing already-created surplus, not about enabling new value — so outside observers have little stake in who wins.
Where it stops holding
This stops working when the dispute, despite appearing to be about splitting proceeds, actually gates future value-creation (e.g. platform rules that block new products or markets from existing at all).
Detail
rule: If a dispute is about splitting revenue from a zero-marginal-cost good between two already-rich parties, don't invest much concern in who wins.
threshold: Zero (or near-zero) marginal cost + both parties already highly profitable = low-stakes dispute for outsiders.
rationale: No new value is being created or destroyed by the outcome; it's pure wealth transfer between incumbents.
In their words
Right? Like, it's purely zero marginal cost. It could not be more purely zero marginal cost than unlocking a bit of code
Consistent rules beat the content of the rule in antitrust judgment
Source
Sharp Tech with Ben Thompson — (Preview) The App Store in the Shadow of AI, Offensive and Defensive Cybersecurity, Q&A on Financial Planning, AI Writing, American Sports (August 21, 2026) · at 14:20 ▸
When this applies
when judging whether a platform's rule-setting toward dependents/competitors is fair, regardless of how restrictive the rule itself is
The takeaway
A platform rule is legitimate if applied uniformly and disclosed upfront to all parties, even if restrictive; it becomes a violation when applied differently to the platform's own products versus competitors', or changed retroactively to disadvantage others.
Why it works
Antitrust law (per Thompson's reading) punishes selectively applying or retroactively changing rules to favor yourself over competitors in the same space, not merely having a restrictive rule — because consistency preserves the ability of others to plan and compete under known terms.
Where it stops holding
This stops working when a facially neutral rule is designed from inception to structurally disadvantage one class of competitor even though applied 'equally' on paper.
Detail
rule: Judge platform conduct by consistency and timing of rule application, not just restrictiveness.
threshold: Same rule, same wording, applied to the platform's own and competitors' offerings from the start = defensible; different wording/rules for self vs. others, or changing rules after the fact to favor yourself = violation.
rationale: "In US antitrust law, if you're honest all along, you're gonna do what you want... If you change the rules after the fact, that's a big problem."
In their words
to what's all of them. Competing in the exact same space. Right? Right.
Regulatory speed should be judged against ecosystem change, not just calendar time
Source
Sharp Tech with Ben Thompson — (Preview) The App Store in the Shadow of AI, Offensive and Defensive Cybersecurity, Q&A on Financial Planning, AI Writing, American Sports (August 21, 2026) · at 14:20 ▸ · at 17:06 ▸
When this applies
when assessing whether a slow-moving corrective action (regulatory, legal, organizational) still delivers value despite arriving after the underlying landscape has already shifted
The takeaway
Judge the value of a slow correction not by whether it's eventually right, but by whether the underlying system has already locked in the harm by the time the correction lands.
Why it works
A correction that is technically 'right' can still be functionally late if the practical structure of the market has already adapted to the violation, meaning the fix changes wording/compliance but not the real competitive outcome.
Where it stops holding
This stops working when the underlying harm is still reversible or ongoing even after years, so a late correction still meaningfully changes outcomes.
Detail
rule: Weigh 'this is a good outcome' against 'this took so long the ecosystem already adapted around the violation.'
threshold: Four years, during which structural shifts in the advertising ecosystem occurred, per the hosts.
rationale: "Because in that four year time, like, the advertising ecosystem has already been remade to some extent."
In their words
Because in that four year time, like, the advertising ecosystem has already been remade to some extent.