Dinner Table: every word,
on every screen.

A holiday table where the deaf cousin doesn't have to ask “what?” six times and then give up. Everyone sits down, opens the same short code, and the conversation shows up captioned in front of them — color-coded by who's talking, on their own screen, as it's said. No one passes a single shared screen around. No one gets left out of the laugh.

What Dinner Table is

Dinner Table is the live-caption tool for a gathering. Three things, plainly: captions of the whole conversation, shown on each person's own screen instead of one shared display passed around, and a name chip in front of every speaker's words so a reader always knows who said what. Everyone joins one “table” by a short code — no account, no install — and from then on the room is followable. Captions are on by default; the spoken audio is additive, never the only way in. You can run it one-on-one over coffee or for a whole holiday table, any mix of people speaking and people reading.

Here's the part that decides everything: the captions live at the seat, not on a wall. A deaf grandmother reads the line at arm's length, in her own text size, without turning to squint at a screen across the room or waiting for someone to repeat it. A hearing uncle who'd rather just listen can ignore his screen entirely. Same conversation, two ways in, nobody downgraded to a spectator. That's the whole difference between captioning a room and captioning for the people in it.

What we built

The split that makes it work: any screen at the table can volunteer to be a microphone after a consent check; every screen, mic or not, renders the same live caption feed. Speaking is opt-in. Reading is the default everyone gets for free.

• Captions on every screen The conversation is captioned on each person's own screen, in their own text size, scrolling live. There is no single shared display to crane toward and no one waiting their turn to read it. A reader can size the text up, jump back to live after scrolling, and never lose the thread.
• A name chip per speaker Every line carries the speaker's name and a steady color, so a long table with five voices stays legible — a reader always knows whether that was Dad or Gram. Consecutive lines from the same person group into one running turn, so a talker shows their name once and keeps going, not a fresh label on every sentence.
• Consent-gated capture A screen becomes a microphone only after someone agrees on its behalf — the prompt asks them to make sure everyone at the table is okay with live captions first. No screen captures audio it wasn't asked to.
• A recording indicator + one-tap stop While a screen is listening, it carries a visible RECORDING marker — a pulsing dot, the words “this device is a microphone,” and a live level meter so you can see it's on. One tap stops it. The captions themselves stay; the listening is what ends.
• Captions on by default, audio additive Captions are the baseline state, not a setting you go find. The spoken audio is something added on top for people who want it — mute every voice and the table is still fully usable as text. ASL is never auto-captioned.
• Help it spell the table right Anyone seated can add a town, a business, or a last name so the captions get the proper nouns right — and the names of people who join are added for the captions automatically, so “Frederick” and “Gram” land the way the table means them.

* A working target, not a hard cap. Five distinct, color-coded voices is the size we tune a table to read cleanly at; bigger gatherings still work, and side-conversations can split into their own panes when the table gets loud.

Where this came from

The honest version: we didn't dream up a captioning tool on a whiteboard. We already had a live-caption feed running for one room and a separate idea about coordinating many screens around one shared session. Dinner Table is what happened when we put those two together — the caption feed everyone could read, fanned out to a screen per seat instead of one display at the front. The room stopped being a stage with captions on it and became a table where the words simply arrive in front of you.

None of the listening pieces are sport- or venue-specific, which is the point. A consent prompt, a RECORDING indicator with a level meter, a one-tap stop, captions on by default, ASL never auto-captioned — those are the same rules every 1891 surface that touches audio follows, written once and reused here. The conversation came from a table. The discipline came from everything we'd already built.

The captions belong at the seat, not on the wall. The day a deaf grandmother has to turn and squint at a screen across the room to follow her own family, we built it wrong. — a working principle, written into the table view on day one

Pull up a chair.

Below is a small, working slice of the table: three screens, all reading the same live captions. Tap a name to make that person say their next line — watch it land, color-coded and labeled, on every screen at once, exactly the way it does when a real voice is captioned to the whole table. One screen is acting as the microphone, so it carries the RECORDING indicator and a one-tap stop. Nothing here uses a camera or a microphone and nothing leaves your browser; it's the fan-out, running locally.

Why this fits 1891

The pitch on madeby1891.com is for systems built so ordinary people can run them, and built so the room works for everyone in it from the start. Dinner Table is that pitch applied to a family meal. Captions are the default, not an add-on you go enable. The way in is a short code, not an account. The audio is additive, so the table is fully usable as text with every voice muted. None of that is an accessibility skin bolted on at the end — captions-first, name-labeled, consent-gated is just what a gathering needs to actually include the deaf person at it.

There's a longer line behind it, too. The Deaf dinner table — the one place a deaf kid could follow every word because the whole family signed — has been the warm center of Deaf life for as long as there have been Deaf families. Most tables aren't that table. Dinner Table is the part of that you can hand to any gathering now: not a replacement for signing, but a way for a mixed room to keep one conversation that everyone's actually in. The why — Dinner Table Syndrome, and the table I grew up lucky to have — is in the essay: Pull up a chair.

What's still unfair about this

Captions arrive a beat behind a voice, and a beat is enough to miss the timing of a joke that the hearing end of the table caught live. We close that gap as far as the words allow, but we don't pretend it's gone — reading a punchline is not the same as landing it with everyone else, and the people who've lived that know it. We'd rather name the lag than market it away.

The thing we did not do is make captions a feature you pay extra for, or meter them by the minute, or wall them behind a sign-up before a guest can read a word. Captions stay on by default and free because a deaf person's seat at the table is not a premium tier. We also won't auto-caption ASL into English, because flattening someone's language into a transcript is its own way of leaving them out. We measure Dinner Table by whether the quietest reader at the table got the whole conversation — not by what we automated.

Nobody at the table cares how the captions get there. They care that the deaf cousin laughed at the same time as everyone else, for once. — Anthony Mowl, framing the build for Dinner Table

That sentence is the spec. The build is good when no one at the table notices it's there.