Monday, June 15, 2026 · 9:41 AM
ok let’s do the critique episode
good
because Range is dangerously easy to misread as “dabble forever and call it wisdom”
i have seen that linkedin post
so what does the book not claim?
it does not claim specialization is bad
it does not claim generalists win everywhere
and it definitely does not claim wandering beats practice
what is the actual boundary?
domain type
in kind domains, early specialization can work beautifully: stable rules, fast feedback, clear improvement loops
chess, golf, some music drills, that sort of thing
yep
when the game tells you quickly what works, deep reps compound
and wicked domains punish that?
they can
if rules shift, feedback is delayed, and the next case only sort of resembles the last one
so the cheap takeaway is “be a generalist”
the better takeaway is “match your learning strategy to the game you are actually playing”
also, breadth can be fake
collecting shallow opinions across 12 fields is not range. it is browser tabs with shoes
what makes breadth real then
transfer
can you use one field to make a better prediction, question, design, or diagnosis in another?
exactly
range without tests becomes aesthetic identity
what are fair critiques of Epstein here?
public critiques often push on overgeneralization
some examples may travel better than the underlying evidence
and “generalist” is not always easy for people without money, time, or institutional safety
yes
sampling is easier when mistakes are survivable
so how should i use Range responsibly?
ask 3 things: is this domain kind or wicked, what feedback can i trust, and what cost do i pay to explore?
and if exploration is expensive?
make smaller samples
side projects, conversations, short trials, reversible bets
that is less sexy but much more usable
Range at its best is practical humility
not a permission slip to avoid getting good
beautifully annoying. bye
go specialize, but with peripheral vision
Read Mon, Jun 15 · 10:02 AM