Raycast: A 200MB Web Browser That Charges Rent for LaunchBar

Filed from Raycast — Charging Rent for a 1995 Launcher.

What it claims to be

A blazing-fast launcher that will change how you use your Mac, complete with an extension store, AI extras, and a subscription. It was recommended to me by AI. I had never heard of it, and the AI chatbot was profoundly confused that this was possible — which is exactly the tell. Because everyone who arrived at computing in 2020 knows Raycast invented this entire category.

No it didn't.

What it actually is

An Electron app. A launcher whose entire job is to launch things quickly, built on a web browser runtime, idling at several hundred megabytes. Sit with that for a second. The tool you summon because a native app feels too slow is a whole Chromium quietly waiting around to look fast.

Quicksilver did the same job in single-digit megabytes of Cocoa. You run a Mac with a BSS-budget gate; you already know which lineage you're descended from.

The receipts, in order

• LaunchBar, 1995, NeXTSTEP — abbreviation search before macOS itself existed. The launcher predates the platform it now lives on by half a decade and change. Everything Raycast does with fuzzy matching, LaunchBar was doing when the Mac was running System 7.
• Quicksilver, 2003 — this is the important one. It took "type to find" and turned it into an actual object broker: object → action → argument. The comma-trick. Triggers. A plugin culture. It was the tool every power user ran, run by essentially one guy — Nick Jitkoff — until someone hired that guy away from it.
• Jitkoff took it to Google, where it became QSB (2009). Google open-sourced it, then did what Google does and buried it in the Reader/Inbox graveyard.
• Spotlight (2005) shipped the mainstream, defanged version of the whole idea — enough to convince the world Apple had done something new, just late and watered down.
• Alfred (2010) productized the vacuum that Quicksilver left behind.
• Raycast (2020) took Alfred's shape, added an extension store, and got credited with inventing the category by everyone who showed up that year.

And meanwhile Quicksilver's corpse still ships releases on GitHub, maintained by hobbyists with no funding and no roadmap. It never even properly died — it just stopped being cool, which as far as Silicon Valley is concerned is the same thing.

The pattern underneath

This is the part worth keeping, because it's the same one in reverse from the floating-windows shower thought: UI history doesn't accumulate, it re-runs. Tabs were BeOS window handles. Launchers were LaunchBar in '95. Virtual desktops were a 1986 X10 hack. Tiling was Ratpoison and acme. Every platform generation forgets and reinvents the same handful of primitives, and whoever re-ships them gets all the credit.

The only variable that decides who's remembered is extensibility. Quicksilver mattered because it was a plugin culture, not a hotkey. Raycast won because it's a store, not a feature. Same shape, twenty years apart, and the store got paid rent.

Charging rent for it

Here's what actually galling about Raycast specifically: the thing it's selling a subscription for has been free software on this very platform for twenty years. LaunchBar did it in '95 for whatever the shareware cost, Quicksilver was free, Alfred's base has been a free download for fifteen years. You do not get to put an ocean of Electron between a user and a hotkey, demand money monthly, and then act like you're owed the "power user" demographic. You're a SaaS lease on a 1995 idea wearing a 2020 haircut.

What sticks around

The design lesson, which has real teeth for the god menu: the command registry is the product and the menu is its renderer. Get the action-registration seam right — commands as first-class shell verbs, extensible by any app, not locked inside a single vendor's store — and the origin of the pattern stops mattering. An action is just a verb with a target.

Quicksilver's comma-trick was this: object → action → argument. That seam is what made it a culture instead of a feature. The hotkey was disposable; the grammar was the thing. Anyone re-shipping a launcher in 2046 will be selling some dressed-up version of that same grammar, and we should all be able to say "oh, right, Quicksilver already did that."

Verdict

Raycast is not the invention. It's the fourth or fifth re-ship of a primitive that predates macOS, delivered in a web browser runtime, sold to you monthly for the privilege. It's Alfred with an app store and a sub, graduate school for the people who think dot-com discovery is invention.

Pig pile complete. In 2046 someone ships a fancy new launcher for VirelaiOS and gets credited with inventing it — that's a feature of the cycle, not a bug. The only thing you can control is which seat you're in. Be LaunchBar. Be Quicksilver. Do not be Raycast.