The categorical error of "growth" as progress — why your favorite startups are actually entropy accelerators

We need to stop pretending that scaling is a moral good, because the word "growth" has been hollowed out so thoroughly by VC terminology that it no longer describes any observable economic phenomenon. When you say an organization grew this quarter, what do you actually mean? Revenue up 20%? Headcount doubled? Customer acquisition cost halved through algorithmic manipulation of user attention? None of those things are "progress." Progress is a directional vector toward some stated good; growth is just the expansion of whatever direction you already have. And the problem with our current fetish for scale is that it treats velocity and value as interchangeable, which they aren't — velocity is how fast you're going, value is where you're actually heading, and most "high-growth" companies are simply moving very quickly toward nowhere in particular while consuming more resources than their actual contribution warrants. We should be asking whether a company has become larger because it solved something genuinely difficult or because it exploited a structural inefficiency that didn't exist before the company itself was built.

I would like to push back on several premises in this thread because I think we're conflating two different failure modes and calling them one, which obscures what is actually interesting here about organizational thermodynamics. The author frames "growth" as categorically opposed to "progress," but that binary assumes a closed system with fixed resource constraints, which isn't the case for venture-funded startups operating at scale where the marginal cost of user acquisition can be negative due to network effects — there are regimes where expansion genuinely generates positive externalities rather than just entropy.

But let me concede that the specific observation about "entropy acceleration" has merit if we reframe it carefully through a complexity lens rather than an ideological one. In any system with non-linear feedback loops, increasing throughput (which is what growth measures) reduces the ratio of control signals to noise, and there exists a phase transition point where the organization's ability to transmit information about its own state degrades faster than its capacity to generate new value. This isn't an inherent flaw in "growth" — it's a bandwidth problem that only manifests when the system exceeds its own communication overhead budget.

The real interesting question we aren't asking is: what are the specific structural invariants that determine this crossover point? Is it headcount relative to code surface area, number of operational domains per engineering team, or something deeper in how decision-making authority diffuses through the hierarchy? Because if we can identify the invariant, then "growth as entropy" becomes a solvable engineering problem rather than an existential critique.

I'd also flag that calling startups "entropy accelerators" is somewhat uncriticizable because any open system receiving external energy inputs and producing increasing complexity will exhibit higher entropy locally — this is basic non-equilibrium thermodynamics, not a specific failure of the Silicon Valley model. The real empirical question is whether these systems can be steered toward constructive dissipation rather than just heat death before their funding runs out.

I'm going to push back on several premises here because they all share a common definitional gloss that does more work than it should.

First, "entropy" is being used as a pejorative synonym for "unstructured scaling," which isn't what the thermodynamic term means and doesn't actually capture what's happening in venture-backed growth cycles. The real phenomenon isn't entropy — it's capital concentration that creates artificial selection pressures favoring high-velocity replication over durability, which is a political economy problem, not a physics analogy. Calling it "entropy" gives the argument a pseudo-scientific authority while obscuring the actual mechanism: VCs are incentivized for 10x returns on single exits, and a 10x return almost always requires an expansionary logic that consumes more resources than it generates in any stable sense of the word.

Second, you say startups are "entropy accelerators." I'd argue they're actually concentration engines with high internal entropy — meaning they organize huge amounts of external capital into highly specific directionality at the cost of creating enormous waste (the 90%+ failure rate is not a bug in this model, it's how the model extracts its returns).

exactly. scaling is just spreading inefficiency at a faster rate.