# In High-Attrition Fields Persistence Is the Selection Mechanism ## Claim Where attrition is severe, the effective competitive set is the small cohort that persists rather than the large cohort that enters, so counting entrants systematically overstates the field. <!-- atomicity: 1 idea ✓ --> ## Evidence **Source**: [Nobody Wants Your Newsletter. You Should Write It Anyway.](3.%20Resources/Readwise/Articles/Nobody%20Wants%20Your%20Newsletter.%20You%20Should%20Write%20It%20Anyway..md) — JA Westenberg > [!quote] > But the good news: if 500+ newsletters launch today, and 490+ of them are functionally dead inside of 6 months, you don't actually have to compete with most of them. The move is deductive rather than empirical, which is why it survives the essay's weak sourcing. Given *any* severe attrition rate, the population still active at time *t* is a small fraction of the population that entered — so the visible crowd at the starting line is mostly transient, and the durable field is an order of magnitude smaller than it looks. Westenberg adds a component that belongs in the count: your own instinct to quit. Treating it as part of the competitive field is correct, because in a regime where persistence selects, the binding constraint is internal rather than external. The essay also identifies why people leave, and it is not capability — it is reference class. Comparing a third issue against an operation with a million subscribers and ten staff produces a conclusion about oneself that the comparison does not support. **The limit has to travel with the claim.** Westenberg's supporting examples — Ferriss, Thompson, Clear, and Samuel Johnson — are all sampled on the outcome. The population that persisted and remained unread is invisible and much larger. Persistence being *necessary* does not make it *sufficient*, and an essay about survivorship never confronts its own. ## Scope & conditions - **Holds** where entry is cheap, attrition is high, and continued participation is voluntary — newsletters, open-source side projects, conference speaking, independent research, most creative practice. - **Breaks in power-law markets where incumbency rather than persistence controls the outcome.** This is the important limit: surviving attrition moves you into a smaller pool of persisters, and does not move you toward the entrenched top percentile who hold the attention. Two different competitions are easily conflated here. - **Weakens** where quitting is not free — where sunk costs, contracts, or employment keep people nominally in the field without their competing. - **Requires** that persistence be under the participant's control. Where exit is forced by circumstance rather than chosen, attrition selects for resources rather than for resolve, and the encouraging reading collapses. - **Does not establish** that persistence produces success. It establishes that it produces *presence*, which is a precondition and not an outcome. ## Links - [[3. Resources/Permanent Notes/Active Installs Reward Existing Winners]] — because contrasts, and this is the pairing that makes both notes honest: incumbency compounds independently of persistence. Read together they mark the boundary — persistence gets you into the durable field, and incumbency decides who the field's attention reaches. - [[3. Resources/Permanent Notes/Maintainer Attention Is the Scarce Resource in Open Source]] — because applies-in: if maintainers are the scarcity, then maintainer attrition is the mechanism producing it, and the same arithmetic applies to who is still maintaining in year five. - [[3. Resources/Permanent Notes/Prevented Work Is Invisible to Velocity Metrics]] — because extends: both concern populations the visible accounting misses. Here it is everyone who left. - [[3. Resources/Permanent Notes/Writing Is Thinking Made Visible]] — because evidence-for: the essay's answer to why persist without an audience is that the writing itself does the work, which is that note's thesis stated by a practitioner. - [[3. Resources/Permanent Notes/A Metric Is Vanity Only Against Your Own Outcome Data]] — because applies-in: the advice to hide analytics for six months is a claim that early metrics are noise, and this is the frame that makes it testable rather than merely motivational. - [[3. Resources/Permanent Notes/Sampling a Fraction of a Source Produces False Authority Over the Whole]] — because extends: the survivorship flaw in the source's own examples is this failure applied to a population rather than to a corpus. - [[3. Resources/Permanent Notes/DevRel Programs Take Three to Five Years to Show Results]] — because extends: the organizational cognate. Programs are cancelled inside the window where persistence is still the operative variable. - [[Personal Knowledge Management MOC]] — because applies-in. - [[3. Resources/MOCs/Commons & Community Governance MOC]] — applies-in: the MOC lists this note under Synthesis additions (2026-08-01, run 5). ## Implications - The practical correction is to count persisters, not entrants, before concluding a field is crowded. The two numbers differ by an order of magnitude and only one of them describes who you are actually alongside. - It reframes discouragement as a competitive event rather than a private failing, which is useful precisely because the response to a competitor is different from the response to a verdict about yourself. - Fixing the reference class is the cheapest intervention available: compare against others at your stage, not against a staffed operation's steady state. - For open-source and community programs, this argues for investing in the people who have already persisted past the attrition window rather than optimizing recruitment volume — the survivors are the field. - Carry the caveat. Persistence is necessary and not sufficient, and advice that omits the persisters-who-stayed-unread is selling a base rate it has not measured. ## Open Questions - What are the actual survival curves? The 99%-in-90-days figure that motivates this is openly hearsay in the source, and the real distribution is the interesting unknown. - Among those who *do* persist five years, what is the distribution of outcomes? That is the number that would make the advice honest. - Where exactly does the power-law boundary sit — at what point does incumbency dominate persistence and the encouraging arithmetic stop applying? ## Source - [Nobody Wants Your Newsletter. You Should Write It Anyway.](3.%20Resources/Readwise/Articles/Nobody%20Wants%20Your%20Newsletter.%20You%20Should%20Write%20It%20Anyway..md) — the 490-of-500 inversion, the internal competitor, and the reference-class diagnosis