🏚️ Gingerwhomp (n.): the satisfying thud of a gingerbread house's roof caving in.

No two days alike, no lasting harm done, and no anesthetic required. (The method behind it is the lab's own; the Coffee Break post shows how it's done, in five easy steps.) The app has:

• Notifications: each word arrives quietly, with no sound, and old ones are tidied away.
• Widgets: the Word of the Hour on your Home Screen and Lock Screen.
• The app itself: Pragmaphilia's Dictionary, with the latest word on top and the words so far below, to copy and share.
• Tell a Story: a short story, written on your device, that uses your word just as it's defined.
• Settings for how often words come, a part of speech, and an optional theme ("a rainy Monday"), plus Pause Words for when you need to concentrate.

Requires an iPhone with iOS 26 or later. Definitions need Apple Intelligence, on a device that supports it, turned on; without it, the words still come, just without definitions. The app collects no user information: the words are created within your own device (see our privacy policy).

Status: version 1.0.9 for iPhone has been on the App Store since October 20, 2026. The Mac version, which keeps the latest word and its definition in the menu bar, is in Apple's review and coming soon; when it ships, it will be announced here and in What's Cooking. More about the app is on its page.

Download on the App Store

For iPhone. The Mac version is coming soon.

Now AI has added urgency. It's changing which jobs exist, what a college degree is worth, and what it even means to "know" something when a capable assistant sits in every pocket. The pressure to rethink education is coming from two directions at once.

The two-sigma problem

In 1984, the educational psychologist Benjamin Bloom published a result that still haunts the field. Students taught one-on-one by a good tutor performed about two standard deviations better than students in a conventional classroom. The average tutored student did better than roughly 98 percent of the classroom group. Bloom called it the "two-sigma problem": we know how to teach far better, but we can't afford to give every student a personal tutor.

For forty years, the obstacle was cost. That has changed. Today's large language models can explain Hume's problem of induction, Russell's paradox, the Krebs cycle, or a tricky passage of Japanese grammar, patiently and at any hour, and answer the follow-up questions too. The knowledge is no longer the bottleneck.

The missing piece is presentation

Most AI tutoring still happens in a chat window: walls of text, read in silence. That's useful, but it's not what tutoring feels like. A good teacher talks with you. You can interrupt. Key terms go up on a board while they speak. Their face tells you whether they're listening, thinking, or about to make a point.

So the real work now is twofold: building good curricula and getting the presentation right.

How we stumbled onto it

At Pragmaphilia Labs, we've been building Miss Sakamoto, an app that gives an AI assistant a face, ears, and a voice: an old iPhone on a stand, or even a small app on your Mac, becomes a window into your AI. We built it mainly as an intelligent and enjoyable front end to Claude Code, for our software work. One morning we used it for something else entirely: a review of the history of philosophy, from ancient Sumer to Wittgenstein.

It worked startlingly well. Miss Sakamoto spoke each lecture aloud, in what we came to call "professor mode." A small whiteboard beside her listed each philosopher's dates, works, and main ideas. When a question occurred to us, we simply started talking; she stopped, answered, and picked up where she left off. Her avatar changed with the topic, and short captions and links popped up as she spoke. Everything happened at the student's pace, not the lecturer's.

Professor mode: the lecture is spoken; the whiteboard holds the key points, at your own pace.

Several design lessons fell out of a single morning:

• Interruptible is everything. A lecture you can't interrupt is a podcast. The ability to say "wait, stop" and ask a question turns it into a conversation.
• Many channels, each at its own pace. Voice carries the explanation in real time. The whiteboard, captions, and links wait until you look. Nothing scrolls away before you're ready.
• Never leave the student in the dark. Thinking takes a moment. Simple indicators showing that the system heard you and is working remove most of the frustration of waiting.
• The student leads. Our best moments came from our own questions and disagreements, not the prepared material. The tutor's job is to follow up, push back, and occasionally ask questions in return.
• Point outward. The world is already full of excellent free material: Khan Academy, Wikipedia, open courseware, recorded lectures. A good tutor knows when to say "here's a great video on exactly this."

What we plan to build

We're working on a standalone tutoring app based on what we learned: professor mode on its own, for middle school students and up, curious adults, and lifelong learners. It's for anyone who wants to learn at their own pace and in their own direction.

It will rest on a few principles. It will ask students to explain ideas back, since that's where learning sticks. It will bring earlier topics back for review days later. Every claim on the board will be traceable to a source. And it will be affordable worldwide: a few dollars at most, priced by country, and free where that's what it takes.

There are real risks to take seriously: confident mistakes, overreliance, the privacy of young users, and unequal access. None of them is a reason to wait. Each is a design requirement.

Bloom's two sigmas have been out of reach for forty years. We think a patient, interruptible tutor with a whiteboard, available to anyone with an old phone, might finally bring them within reach.

And she doesn't only answer. Whenever Claude has something to tell you, like a long job that just finished or a step that needs your approval, she says it out loud. You hear it from across the house, even with the phone locked on its charger.

(Yes, that Miss Sakamoto: our Lab Director, now with a phone to stand on.)

She puts up whiteboards for an agenda, a checklist, or a haiku, and small signs with links you can tap. And when nobody is around for a while, she naps.

Your privacy, plainly:

• The camera only looks for a face, to know that someone is there, and keeps no image. The microphone runs only while someone is there, and only what's said to her (her name, or the conversation that follows) goes any further.
• Your voice is recognized on the phone or on your own Mac. Audio goes only to your Mac, over the USB cable or your home network, is held in memory while it's recognized, and is never saved. Away from home, through Tailscale or another VPN, the audio crosses the internet, encrypted end to end between your phone and your Mac, and still isn't saved.
• Nothing goes to Pragmaphilia Labs.
• She takes a picture only when asked, with a flash and a shutter sound.
• What she hears does become text in your Claude conversation, so tell the household she's listening.

Status: in testing, with TestFlight soon. It works with your own Mac running Claude Code. No release date yet; when it ships, it will be announced here and in What's Cooking.

なんでやねん (casual): Why on earth?! Oh, come on!

「財布忘れたから、お昼おごってくれへん?」「なんでやねん!」

Every phrase is labeled polite, casual, rough, or rude, so you know who you can say it to. There are sixty, all hand-written, one a day, and then the cycle starts over. A widget puts today's phrase on the Home Screen, the Lock Screen, the Mac's desktop, and Notification Center; complications put it on the Apple Watch face; 関 in the Mac's menu bar shows it too; and your device's Japanese voice reads it aloud.

It collects nothing, and it works offline.

It's made by Pragmaphilia Labs: Miss Sakamoto, creator and project lead (and our Lab Director); John Mifsud, executive producer; and Tamamo, language consultant. Two honest notes. The phrases were written by Miss Sakamoto, not by a native speaker, so a native Kansai speaker should check every entry before it ships. And the device's voices speak standard Tokyo intonation: the readings are right, but the Kansai music isn't.

Status: in development, for iPhone, iPad, Apple Watch, and Mac. No release date yet; when it ships, it will be announced here and in What's Cooking.

That recognition comes from Pragmaphilia Labs. We built it to read boards and book diagrams quickly and accurately, and licensed it exclusively to SmartGo, arguably the best Go software there is. Anders Kierulf announced the release, with online play through OGS (online-go.com) as its other headline, on the SmartGo blog. Photo import is part of the SmartGo One Pro subscription.

The same technology also makes printed Go books much faster to bring into SmartGo's interactive book format. The release adds two free, Creative Commons books, Relentless and Shape Up!, as a first taste.

Publishers and authors: if you have Go books you'd like to see in SmartGo, we can convert them at a fraction of what it used to cost. Get in touch.

The method

1. Pick donor words. Two or more real words, ideally tied to what's going on that day (a game feature, a topic, a pop-culture reference). Draw widely: English, Latin and Romance roots, German compounds, Slavic flavor with diacritics (č, ř, ž, ł), Japanese fragments, onomatopoeia, food, machines, animals.
2. Dissect. Split each donor word into syllables.
3. Transplant. Shuffle and swap the syllables between the donors: Nashorn + Nocturne → "Nocnashhorn-turn", "Nocturnashern".
4. Graft. Optionally add a prefix or suffix (a "suffix graft"): -ing, -ation, -ster, -wump, -ix, -erino, -zle, Greek philo-.
5. Check the pulse. It should be pronounceable and fun to say, and ideally the donors are still faintly recognizable, so there's something to guess.

Examples

Rules of thumb

• Don't reuse a stem within a day, and avoid overused endings like "-oodle" and "wobble".
• Words tied to the day's topic land best.
• John's own syllable shuffle is the core technique; he gave the method its name later the same day.

Now an app, with stories

The method grew into an app, Word Salad Surgery, now on the App Store for iPhone, with the Mac version coming soon. Every so often it stitches a new word together from donated parts, and Apple Intelligence, working on your own device, writes it up as a dictionary entry. Tap Tell a Story, and it writes a short story that puts the word to work. One morning's word, with its story:

🥄 Gravywingle (v.): to slurp an overlooked puddle with a bent spoon, leaving only ripples in the grass

The train sat still, its wheels dreaming on the gravel. A bent spoon, found near a torn flyer, caught the morning light. A child leaned down, then Gravywingled the puddle—slurping the overlooked drop with a twist. The grass trembled, leaving only ripples. A pigeon blinked, unimpressed. The spoon rolled, bumped a pebble, then stopped. The sky turned pale yellow. No one moved.

Every word and every story is new, so no two are ever the same. The pigeon was not available for comment.

• Thumbnails in Finder: each game's board after the opening moves, with a stack of boards for a file that holds several games.
• Quick Look previews: press the Space bar on a game to see the board beside the players, the result, the event, the date, and the rest of the game information.
• Spotlight search: the players, event, date, result, comments, and more of every game, in Spotlight, Finder, and mdfind.
• A screensaver that plays the opening of a random game from your collection on each display.

New in 2.0.9: a much lighter screensaver. Each move now redraws only the small areas of the board that change, so it looks the same while asking far less of the Mac.

Requires a Mac with Apple silicon and macOS 26 or later. Free and open source (MIT license). SGF Tools was started in 2009 by Jason Foreman; 2.0 is a new version by Pragmaphilia Labs.

Download SGF Tools 2.0.9 from GitHub

• Kansai-ben of the Day: one Osaka and Kansai phrase a day, from the polite to the downright rude, in beta testing.
• Space Flees 2100: the remake of John's 1987 DOS game, in beta testing.
• Lab Notes: a new numbered series of essays; No. 2 is out.
• The Slack division: things we made while slacking, for others to use while slacking. The first idea on the bench is a generator of funky nonsense words (see Word Salad Surgery).

More as it comes out of the oven.

The meta-ortho-endo continuum is a small vocabulary for making those jumps explicit. It doesn't tell anyone what to call their levels, and it isn't a theory of abstraction. It simply names the moves, so that anyone who shares the words can say out loud which way the conversation is going.

Three words

Ortho is where we are now: the default level of the current discussion. The name comes from the Greek orthós, "straight" or "correct." Every domain has its own natural ortho. In a code review it might be the function under discussion; in a design meeting it might be a whole feature. Ortho is the zero point of the scale.

Meta means up a level, toward the bigger picture (Greek metá, "beyond"). "Let's go meta for a moment: how does this change affect the application as a whole?"

Endo means down a level, into the details (Greek éndon, "within"). "Going endo: how are we actually going to implement this search?"

A continuum, not three boxes

There aren't just three levels. Think of a number line with ortho at zero. Meta extends upward through as many levels of abstraction as the subject has, and endo extends downward through as many levels of detail. From a single function you can go endo into its algorithm, then into the data structure, then into the bits in memory; or meta into the module, then the system, then the business the system serves.

Moving more than one step is easy to say: "two levels meta," or in writing, by repeating the symbol or adding a superscript for bigger jumps (see below). And when a discussion settles at a new level and stays there, it can simply be re-anchored: "new ortho," or "let's set ortho to the subsystem."

An optical analogy

Think of vision. Ortho is ordinary sight: the unaided eye, or glasses that simply restore normal vision. Going endo is picking up a microscope: smaller things, in far more detail. Going meta is the telescope: it shows you the largest things there are, planets and galaxies. Examined strictly, the analogy bends, since both instruments narrow your field of view. But to most minds the telescope naturally feels like going up, because what it shows you is bigger. That's the real point: the level of abstraction is about the scale of what you're thinking about, not how much of it fits in view at once. (For a pure step back with no magnification at all, think of a drone rising over a field, a satellite photo, or a map.)

Why bother

Because the cost of an unannounced jump is real. When a designer says "this is too slow" and means the whole workflow, while the engineer hears "this loop is too slow," they can argue for ten minutes about different things. A single word at the moment of the jump, "going meta" or "going endo," keeps everyone on the same level. It works best in fields where levels shift constantly: software, design, engineering, and teaching. But it applies to any conversation where the zoom level matters.

Building on Hayakawa's ladder

I formulated this vocabulary without knowing it had a famous ancestor. In Language in Action (1939), the linguist S. I. Hayakawa described a ladder of abstraction. His example is a cow named Bessie: at the bottom, Bessie herself, this one unique animal; one rung up, "cow"; then "livestock," "farm assets," "asset," and finally "wealth." Each rung leaves out more detail and covers more ground.

Hayakawa warned against dead-level abstracting, getting stuck on one rung. Stuck high, you talk in empty generalities that never touch anything real. Stuck low, you pile up details with no point. His lesson is to move up and down deliberately, and above all to come back down with an example after climbing: abstraction without examples is empty, and examples without abstraction are a pile of anecdotes.

Two more of his ideas, from the general semantics tradition of Alfred Korzybski, sit close to this one:

• The map is not the territory. Every word is already an abstraction of the real thing, a rung above it. Knowing which rung you're on is the first defense against confusing the two.
• Two-valued versus multi-valued thinking. Treating everything as either/or (good or bad, big picture or detail) versus seeing a range. That's why this is a continuum rather than a switch between "high level" and "low level": there are as many rungs as the subject has.

What the meta-ortho-endo continuum adds is the conversational piece. Hayakawa taught individuals to move up and down the ladder in their own thinking and writing. These words let two or more people move together: to announce a climb ("going meta"), a descent ("going endo"), or a return home ("back to ortho"), so nobody is left on a different rung. It's Hayakawa's principle, applied out loud. (The designer Bret Victor's interactive essay "Up and Down the Ladder of Abstraction", from 2011, applies the same principle to software tools: interfaces that let you climb and descend.)

Glyphs

For writing, each state has a glyph:

• ↖ meta (U+2196, north west arrow)
• ⌀ ortho (a zero with a long slash; U+2300, the diameter sign)
• ↘ endo (U+2198, south east arrow)

Written together, the continuum becomes ↖⌀↘: a slashed zero whose slash extends into an arrow pointing up and to the left, and another pointing down and to the right. Multiple steps can be written by repetition (↖↖) or with a superscript (↖³). Typing these by hand is a nuisance, so they suit text expansion macros, an AI assistant, or simply spelling the words out.

In practice

The vocabulary works best lightly, the way "for example" or "in other words" does: a signal at the moment of a shift, not a ritual. Two people who share it can say "wait, go endo on that" or "let's step meta" and immediately know where the other is standing. And "back to ortho" is a gentle way to end a digression, since the digression becomes a deliberate trip up or down the ladder rather than a wander off course.

Here's how it looked back in 1987. The new version has a vast array of enhancements to bring this classic into this century.

Status: in beta testing. No release date yet; when it ships, it will be announced here and in What's Cooking.

Join the beta on TestFlight

For Mac, iPhone, iPad, and Apple TV.

Serve an idea. Get one back.

Status: in development, for iPhone and the Mac. No release date yet.