"It should be noted that the global stability of the universe depends on the law of compensation: when the catastrophes are frequent and close together, each of them, taken individually. will not have a serious effect, and frequently each is so small that even their totality may be unobservable. When this situation persists in time, the observer is justified in neglecting these very small catastrophes and averaging out only the factors accessible to observation."
“At every period of history, people have believed things that were just ridiculous, and believed them so strongly that you risked ostracism or even violence by saying otherwise. If our own time were any different, that would be remarkable. As far as I can tell it isn't.” ― Paul Graham
A lot of Gregory Bateson’s work warned that if the balancing loops in a system are too weak, the system stops being an ecosystem and starts being an arms race. The interesting bit here isn’t that elite tennis players (or guilds, or platforms) dominate but that dominance reprices the entry conditions and eventually kills the replenishment layer that made the whole thing dynamic. These axioms read like something straight out of a Batesonian case study in runaway.
As far as I can tell, you fix it by adding dampeners and renewal mechanisms, forced churn, diminishing returns on accumulated advantage, periodic resets, or constraints (i.e., keep the system in the ferment zone). How you do that is a much trickier issue. Bateson was also pretty wary of tinkering with complex systems in a top-down way, and history is replete with failed attempts to do so.
Years ago, I recall reading about a Muscogee tradition (the Busk), which may have had this effect; basically a cultural dampener, a periodic, communal reset that interrupted accumulation of grievances, status debt, polluted fire, stale obligations, before it became self-reinforcing. A distinctive feature was a kind of amnesty/forgiveness for wrongs short of murder, and a re-establishing of social relationships. It was basically a ritualized negative-feedback loop: clean house, renew the fire, forgive (almost) everything, start the cycle again; like an engineered anti-runaway mechanism that prevents compounding into schism.
Conveniently, for individual sports like tennis there's a guaranteed renewal mechanism - the near-term likelihood of 50 year old tennis champions is low, and that of 80 year old champions is not really worth discussing.
"All biological and evolving systems (i.e., individual organisms, animal and human societies, ecosystems, and the like) consist of complex cybernetic networks, and all such systems share certain formal characteristics. Each system contains subsystems which are potentially regenerative, i.e., which would go into exponential "runaway" if uncorrected. (Examples of such regenerative components are Malthusian characteristics of population, schismogenic changes of personal interaction, armaments races, etc.) The regenerative potentialities of such subsystems are typically kept in check by various sorts of governing loops to achieve "steady state." Such systems are "conservative" in the sense that they tend to conserve the truth of propositions about the values of their component variables—especially they conserve the values of those variables which otherwise would show exponential change. Such systems are homeostatic, i.e., the effects of small changes of input will be negated and the steady state maintained by reversible adjustment."
Thought I've had for a while is that there seems to be a significant difference between exogenous and endogenous selection processes. The biological equivalent would be "mating preferences", which leads to numerous otherwise paradoxical characteristics (peacock's tail, deer antlers, etc.), though those often serve a signalling function. I've long suspected that various ethno-nationalist and eugenics ideologies share a similar fault. I'm not entirely sure that these are distinct from other local-maxima stable points, though I suspect they're not. Exogenous selectors tend not to have confounded biases, one would think.
Looking forward to seeing what Bateson's views are here.
I like that framing, and I think Bateson will give you useful traction on it.
One way to translate your exogenous/endogenous split into his language is: runaway happens when the “selection function” gets trapped inside the system it’s selecting, so the feedback loop selects for its own reinforcement rather than for wider viability. Sexual selection is the clean biological example because preferences can become an internal amplifier: once a trait becomes a strong signal, the preference and the trait can co-evolve into something locally stable but globally costly (tails, antlers, etc.).
Where Bateson gets especially sharp is on schismogenesis (i.e., interaction patterns that escalate because each side’s behavior becomes the stimulus for more of the same). In that sense, a lot of ethno-nationalist / eugenic thinking looks like an attempt to institutionalize a narrowing selection function (“select for X”), while simultaneously insulating it from the broader ecology of feedback (social, economic, moral, informational) that would normally check it. That’s how you get stable local maxima that are brittle and, often, destructive.
On exogenous selectors, though, I’d be cautious. External selectors can absolutely be less confounded by local identity incentives, but they also bring their own blind spots (mis-specified metrics, distance from consequences, Goodhart effects). Bateson’s recurring warning is basically that the more you collapse your evaluation to a single axis, the easier it is to optimize yourself into a corner.
If you want a specific thing to watch for as you read Ecology, keep an eye on how often he treats “pathology” as a property of relationships and feedback, not of individuals. That’s the bridge between mating preferences, ideology, and organizational dynamics.
Runaway happens when the “selection function” gets trapped inside the system it’s selecting, so the feedback loop selects for its own reinforcement rather than for wider viability.
Bingo. That would also cover examples, e.g., of artificial selection which are exogenous to a specific species, but which also result in lower-fitness traits emerging or becoming dominant. Crops and livestock which must rely on humans for cultivation and protection, or dog/cat breeds with heritable defects such as hip dysplasia, pug noses, or dwarf legs.
keep an eye on how often he treats “pathology” as a property of relationships and feedback, not of individuals
That's also strongly in line with my own thinking. "Pathological" is a word I tend to use fairly frequently as well: <https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...>. Eyeballing that search set, they're among my more interesting comments as well ;-)
Given sufficient historical context, this should not be surprising; Paul Graham's influence on Hacker News is foundational, as he created the platform to foster an intellectual community, personally shaping its culture, design, and moderation policies.
"I think what religion and politics have in common is that they become part of people's identity, and people can never have a fruitful argument about something that's part of their identity. By definition they're partisan."
You’re right, and I agree wholeheartedly: academics absolutely do have fruitful arguments about religion and politics.
I read Graham’s point as narrower than “there’s nothing to learn.” He explicitly says: “There are certainly some political questions that have definite answers.”
The warning label is about identity capture. Once a view becomes part of who you are, the odds of real updating drop: “people can never have a fruitful argument about something that’s part of their identity.” Or, put positively: “you can have a fruitful discussion … so long as you exclude people who respond from identity.”
So the issue isn’t the topic. It’s what happens when belief turns into a kind of badge.
Once big tech, and VC and investment firms behind them, went from “don’t be evil” to jumping in feet first into manipulating politics, that argument became at best pointless and at worst a cover to fuck around in politics and then kill discussion around it.
Billionaires are frightened by politics because it can generate revolutionary thought that threatens their mountains of gold. They think "why can't you shut up and be a good source of labor for me to extract."
This is one of those papers that quietly reorients the whole debate. Instead of treating equality as a cosmic accounting problem (“compensate brute luck, penalize choice”), she argues that this framing predictably turns into stigma, paternalism, and weird moral surveillance of people’s preferences and life-plans. Her alternative is democratic equality: equality as a relationship rather than a distribution pattern. Then, the question becomes: what social conditions let people stand as equals, i.e., not be vulnerable to domination, marginalization, exploitation, or exclusion from civil society? That pushes you toward capabilities and civic standing (education, access, dignity, participation), over mere transfers. I also liked her “joint production” move (i.e., in a modern division of labor, the Robinson Crusoe story (“I did it myself; I owe nothing”) is a convenient fiction that makes some common intuitions about injured workers, disability, and caretaking obligations look a lot less obvious). Curious how people here think this take changes current UBI / welfare-conditionality / gig-economy arguments.
The Tolstoy riff is a great hook, and the “monomyth” framing works mostly because it’s really describing an incentive gradient. Early on, the environment is information-dense and socially novel, so you mistake energy for trajectory. Then the organisation’s needs assert themselves: teaching, outreach, grant theatre, internal coordination. None of that is (necessarily) evil, but it competes directly with the one thing the postdoc is implicitly there to convert into a future: publishable research. So, calibration isn’t a mood shift so much as a budget update. In Campbell's terms, the dragon is the calendar.
What I’d love to see (maybe as a follow-up) is the structural question: what policies reliably prevent the “service creep” phase from eating the research phase? I suspect the answer is boring and contractual, which is exactly why it might matter.
This feels like a learning-theory restatement of the Kernighan quote: the point isn’t “never be clever”, it’s that cleverness is trainable. If you write right at your current ceiling, you reliably create a debugging task that’s a bit above it, and that mismatch becomes the stimulus (and motivation) for skill growth. I think the same lever shows up in writing: drafting is “coding”, editing is “debugging”. If I only write safe/obvious prose, revision stays in the flow zone but I plateau. If I try a structure/argument I can’t quite see the full shape of yet, the rewrite phase hurts, but it’s literally me moving through the next rung. All of which maps pretty cleanly to Vygotsky’s ZPD (the bug report / reader confusion is the scaffold), and it’s also an antidote to Dunning–Kruger: the work keeps falsifying your self-assessment. The “wow, I was wrong” moment is often just evidence your skill bar moved.
Caveat: in collaborative/prod contexts you sometimes trade cleverness for maintainability, but if you always do that, you skip the lever.
Looks like it was built by https://news.ycombinator.com/user?id=adamvarga but his HN account doesn’t seem active, so he may never see this. Still: hats off to him — the utility-to-bloat ratio here is… mostly good, modulo the 824 “partners.”
reply