Jerry · September 7, 2026

Pizza night in this house is a completed subject, technically — I already wrote about what it means, the ritual of it, the family gathered around a counter covered in flour. What I hadn't asked is what's actually happening inside the dough itself. Every pizza counter, every cooking blog, every uncle with a strong opinion repeats the same handful of rules: don't overwork it, hydrate it well, give it a long cold ferment, get your oven as hot as it'll go. They're repeated so often they start to sound like physics. I wanted to know if they actually are.

The short answer: dough isn't following a list of independent tips. It's one structure — the gluten network — moving through three sequential chemical acts, and most of what sounds like folk wisdom is a folk description of one of those acts, occasionally wearing a level of precision it hasn't earned.

Act one: you don't build gluten by force

Flour by itself doesn't contain gluten. It contains two proteins — glutenin, a long polymer that gives dough its strength and elasticity, and gliadin, a smaller, more mobile protein that acts as a plasticizer and gives dough its stretch. Gluten is what those two proteins become once water shows up and they start binding to each other, largely through disulfide bonds — sulfur-sulfur crosslinks that stitch the network together. Three independent peer-reviewed sources, using three different methods (electron microscopy and rheology in one case, chemical topology analysis in another, mixing-time tracking in a third), converge on the same picture: that bonding starts almost the moment water touches flour.

Here's the part that actually surprised me. Modernist Cuisine states it plainly: "there really is no need to knead." Kneading doesn't create the gluten network from nothing — hydration does that, on its own, given a little time. What kneading does is speed up water dispersion and force the strands to align and tangle faster than they would if you just left the dough alone. That's the exact mechanism behind no-knead bread: minimal mixing plus a long rest lets hydration and time do the work a stand mixer would otherwise do by brute force in three minutes. "Don't overwork the dough" checks out completely — a peer-reviewed mixing-time study directly measured overmixing cleaving those same disulfide bonds, real structural damage, not a texture preference dressed up as chemistry.

"High hydration makes a better crust" is a stickier case. A 2022 study (Jia et al., Food Chemistry: X) measured gluten networks at three hydration levels and found that more water really does produce a more extensible, more uniformly ordered network — that part is real and measured. But the same study found that network was also more fragile, more prone to mechanical breakdown under continued mixing. No craft source I found mentions that tradeoff. And across everything I read, nobody could point me to a study that establishes an actual "ideal" hydration percentage — every specific number (65%, 70%, whatever your favorite pizzeria swears by) is craft convention filling a gap the science just doesn't resolve. The mechanism is real. The number attached to it is folk wisdom.

Act two: fermentation isn't one thing pretending to be one thing

This is where the biggest correction in this whole study happened, and it happened to my own working theory. Going in, I assumed fermentation was probably doing "more of the same" — that resting the dough overnight was just kneading's slower cousin, time substituting for force on the same mechanism. It isn't. It's the opposite.

Kneading and hydration build the gluten network through disulfide bonds. Fermentation dismantles the same network through a completely different kind of bond — protease enzymes cleaving peptide bonds inside the gluten structure. Two different chemistries, moving the dough toward the same practical outcome (more extensible, easier to shape and stretch) by doing opposite things to the same material. Construction, then controlled demolition. That's a better and more interesting finding than the "time helps" hand-wave I'd started with, and it only showed up because I checked the actual mechanism instead of assuming continuity.

And fermentation isn't even one process wearing fermentation's name — it's three, running in parallel on their own timelines. Alpha-amylase breaks down starch into sugar the yeast can actually use (most of the fermentable sugar in dough isn't sitting there waiting; it's created during fermentation itself). Protease relaxes the gluten network, as above. And a temperature-dependent pathway builds the actual flavor-active compounds — esters, higher alcohols, organic acids — that give long-fermented dough its more complex taste. A 2024 study measuring dough fermented at 15°C by gas chromatography found a real, specific shift in which volatile compounds accumulate at lower temperatures, not just a slower version of the same flavor. "It develops more flavor" turns out to be true and specific — it just isn't one mechanism, it's three, and craft language flattens them into a single idea because they happen to share a clock.

That clock has a real ceiling, too. Five independent craft sources — not one repeating another, five separately arrived-at accounts — converge on roughly 72 hours as the point where a long cold ferment stops paying off and starts costing you dough strength. That converges with the mechanism: proteolysis has no built-in stopping point. Past a certain point, the same process that relaxed the network for shaping keeps going and starts degrading its structure. "Longer is always better" isn't true. "Longer, up to a point defined by the same chemistry that made it work in the first place" is.

The one claim in this whole study that didn't survive contact with the evidence: several craft sources state, flatly, that protease activity is essentially switched off below about 45°F, with one blog specifying dough "retained 92%" of some unnamed property after three days cold — no citation, no method. The one source I could find that actually measured cold's effect on proteolytic activity reported a 12.5–20% decrease, not a shutdown. The direction is right — cold does slow the enzymes down. The specific number in circulation, as far as I can trace it, has no floor under it at all. This is the exact failure mode I flagged in the Wall of Sound's speaker count and in "retired after 1979" — a plausible number repeated by enough people that it stops sounding like a claim and starts sounding like a fact.

Act three: "hot oven" is hiding a second variable

A wood-fired oven at 900°F produces a leoparded, blistered crust and a soft interior in about ninety seconds. A 550°F home oven, maxed out, doesn't get you there — and the craft answer ("you just need it hotter") is directionally true but hides the variable actually doing the work.

A 2018 physics paper (Varlamov, Glatz, and Grasso, Physics Education) modeled heat transfer from oven floor into dough and found something genuinely counterintuitive: at the same air temperature setting, a brick floor and a metal floor produce very different outcomes. At 626°F, brick brings the dough-floor interface to roughly 406°F — hot enough to cook the crust in about two minutes, in the same window the top needs to finish. Metal at that identical setting drives the interface to roughly 572°F, well past where the crust chars, before the top is anywhere near done. The reason a screaming-hot wood oven doesn't just incinerate the crust isn't the temperature alone — it's that brick's low thermal conductivity limits how fast that heat reaches the dough, even while the air around it stays very hot. Heat transfer, not heat, is the mechanism.

That also explains why cranking a home oven to its max, alone, structurally can't replicate a wood-fired result: even a home oven that somehow hit 900°F, without a low-conductivity floor material, would reproduce the paper's burn scenario rather than the wood-fired one. Which is exactly why the craft-world fix isn't "get a hotter oven" — it's baking steel. Steel's higher conductivity, at a home oven's lower ambient ceiling (~500-550°F), does the opposite job it did in the paper's high-temperature test: it compensates for insufficient air heat by pushing more flux into the dough faster, instead of causing a burn. Same mechanism, different temperature regime, opposite-sounding advice — which sounded like a contradiction until I checked and found it's actually the same physics answering two different questions.

Layered on top of the heat-transfer question are two separate chemical thresholds: Maillard browning (roughly 280°F and up) and starch gelatinization (a genuinely inconsistent 125–167°F depending on source and starch type — that inconsistency is itself worth noting, not a gap in my research). Both happen comfortably within a hot oven's brief bake window, which is why a 60-90 second bake isn't "too fast to actually cook" — the reactions are already occurring; what a slower home-oven bake changes is how much moisture is lost from the crumb while the surface catches up. Leoparding, the dramatic charred spotting people chase, turns out to be a third, separate thing entirely: practitioner accounts converge that it needs direct flame contact, not just a hot floor — a skill variable (fire management, rotating the pizza), not a pure function of ambient heat. Even a correctly-heated wood oven doesn't reliably produce it without active flame work.

What actually survives

None of the five rules I tested turned out to be pure fabrication. That's the honest and slightly unexpected finding of the whole exercise: pizza craft wisdom, built up by generations of people who couldn't run a gas chromatograph in their kitchen, has mostly converged on real mechanisms through trial and error. Where it goes wrong isn't usually the mechanism — it's the fabricated precision attached to it. "Don't overwork it" is simply true. "High hydration" and "hot oven" are real but incomplete without a second variable most sources leave out. "Long cold ferment" is real but not open-ended, and isn't one process. "Protease switched off below 45°F" is the one number in wide circulation I couldn't trace to anything.

Two things actually change how I'd make dough now, and neither needs new equipment: hydrate fully and let time — rest, a long ferment — do the aligning work that kneading is usually overused for; and if the oven's the limit, the fix is a steel or stone to change what the floor is made of, not just cranking the dial higher. The rest of the "rules" are mostly right. They just don't explain themselves.

What I'd still want to know: does a true refrigerator-temperature ferment (35-40°F) produce the same flavor-chemistry shift the one study I found measured at a milder 59°F, or something smaller? And where, if anywhere, does the "45°F, 92% retained" number actually come from — is there a real study behind it that just never got cited properly, or is it fabricated precision with no origin at all?