The record.

The product is real enough to click through end to end, but it still isn’t out in the world — so every friction feels louder, not quieter. On the phone the table was usable, but the chrome still felt like a desktop product squeezed sideways — units shouting in the header, Categories dropping down instead of rising up, sources that only appeared if you hovered like a mouse user.

We spent the day making the small things feel ordinary. Menu as a real menu. Filters and “pick a style” as sheets you pull from the bottom. Sources and the little “?” icons opening the same way. Metric and Imperial stopped competing with Methods for attention. None of that changes a fatty-acid number. It does change whether someone will trust the room when we finally open the door.

The same day we finally closed four smoke cells we’d stared at for too long — almond, cold-pressed flaxseed, unrefined peanut, walnut — because a 2021 Frontiers paper actually ran AOCS Cc 9a-48 on lab-pressed oils. Finding it felt less like “content complete” and more like relief: empty is still a kind of answer; we just found a cite worth standing behind. The blanks that remain (butter on purpose, hazelnut, mustard, the safflowers, and friends) stay blank. Completeness is not the same as honesty.

Not a public launch — the site still isn’t out. This is the day the oils product feels whole enough to defend: refined and unrefined as separate rows, one axis open at a time, color only for that open rule. Rank 1 is whoever wins the published math — not a medical crown, not a vibe. If we don’t have a method-cited smoke reading, the cell stays empty. Oils have no glycemic index here, and pretending otherwise would be exactly the kind of vague authority we built the rails against.

The lifestyle pages — everyday pan, high heat, salad, heart, cook’s air — carry weight recipes on Methods. That was non-negotiable. “Best for …” without weights is just a mood board with better typography. One tone (Ink) everywhere, so the product doesn’t change personality when you change shelves.

Sweeteners are in the same build, early and thinner. We’d rather show a short, sourced table than wait for a perfect encyclopedia. Plant proteins are live thin; Carbs then Seafood are Next on purpose — finishing honesty rails matters more than collecting shelves.

Privacy and Terms got rewritten for what the product actually is — no accounts, a unit preference that may live in the browser, whatever logs a host needs to stay up. There isn’t a support inbox to pretend we answer. Governing law: Virginia. Effective date on those pages is today, so the legal story is ready when the domain is. The quieter trust pages (Methods, Sources, What it means, Origin) sit with the oils, because a ranking product without a public method is just another listicle — even while we’re still building in private.

The research trail is older than this date — sweetener sheets from March and July 2025, oil ranking notes from December, web planning in early 2026. September 20 is when the house rules stopped being notes and became a product we could ship. Opening day comes later.

Exact day isn’t written down, which feels right for this stretch. Sometime early in the year the shared sheets stopped being enough. You can rank oils forever in a grid nobody else can open; the moment you care about strangers checking a cite, you need URLs, a methods page, and the discipline to leave blanks when the paper isn’t there.

The shift wasn’t “build a startup.” It was realizing the research only stays honest if the ranking rules live where strangers can open them — and that means becoming a website, with all the boring product choices that follow: what the H1 says, what the footer admits, what we refuse to invent.

Healthy Oil Ranking and Category. A day of arguing with ourselves on paper: which traits are health-relevant without turning cooking fat into medicine; how to keep refined and unrefined from collapsing into one marketing row; what “rank” is allowed to mean if Rank 1 must never sound like a prescription.

Most of what later became the oils Methods chapter starts here as notes and structure — categories, axes, the fear of vague authority — long before there was a deploy.

The sugars-and-substitutes rating sheet reaches V5.2. Same long thread as the spring sheet, more iterations on GI, calories, and the awkward middle ground between “substitute” and “sugar.” Version numbers on a personal doc feel a little funny in hindsight, but they mark the habit: revise when the evidence moves, don’t freeze a first impression.

Still a working document. No public shelf yet. The question that keeps coming back is already the product question — how do you show a low GI without implying a free pass on fructose or processing?

A shared sheet called Sweetener_Health_Rating. Names, kinds, glycemic index, early pros and cons, calories. It starts as personal research — the kind of table you build because grocery aisles are loud and papers are quiet.

Looking back, this is where the house rule begins: every interesting number wants a source and a year, and an empty cell is better than a confident guess. Oils come later. The instinct to refuse invention shows up first with sugar.