Why Scout is web-first

Scout used to be configured entirely through Discord. There were command trees for subscriptions, players, accounts, competitions, reports, and permissions — dozens of subcommands. Today there are seven commands, and everything else lives in the browser.

This page is about why, including the parts that got worse.

What slash commands are good at

Discord commands have a real advantage: zero context switch. You are already in the channel, talking to the people you play with, and tracking a player is one line. That is why /track still exists — the fastest possible path from “we should track this person” to a working subscription, with no browser, no sign-in, and no navigation.

That advantage is specific to short, single-purpose actions. It does not survive contact with complexity.

Where it breaks down

A slash command is a flat list of typed options with no memory of what came before. That constraint is fine for /track, and increasingly bad as the operation grows:

No progressive disclosure. Everything a command might need is in one signature. A subscription with filters, a queue allow-list, and a channel is a long line of options that has to be typed correctly the first time.
No good way to show state. Discord can render a list, but it cannot render a table you can sort, filter, and edit in place. “What is currently configured?” is the question people ask most, and it is the one chat answers worst.
Validation arrives late. You discover a mistake after submitting, and you fix it by retyping the whole command.
Discoverability is a memory test. With dozens of subcommands, using Scout well meant remembering that they existed.

None of these are Discord being bad. They are what a command line is, and a command line is the wrong shape for editing structured configuration.

What the browser is actually better at

Forms with validation as you type, tables that show every subscription at once, a query editor with completion and inline errors, a preview of what a report will produce before you save it, and an interface that can display a permission matrix without inventing a syntax for it.

A dashboard also lets Scout add capability without adding vocabulary. Reports and ScoutQL could not exist as a slash command in any tolerable form; as a page with an editor and a preview, they are ordinary.

What this costs

It is worth being honest about the losses.

Setup now requires leaving Discord. For a server that only ever wanted “post this person’s ranked games here”, the old flow was better, and no amount of dashboard polish beats one line in chat. /track and /list exist to preserve exactly that case.

There is a sign-in step. OAuth is friction, and some people bounce off it.

Configuration is less visible to the server. A command run in chat is implicitly public — everyone saw it happen. Dashboard changes are not, which is part of why the audit log exists.

Muscle memory was broken. Servers that had used the old command tree for years had to relearn where things live. That is a genuine cost, paid once, and it was the main argument against the change.

Why the split lands where it does

The rule is roughly: Discord for things you do in the moment, the browser for things you decide.

Tracking a player and checking what is tracked are momentary — you do them while talking about them, so they stayed as commands. Choosing which queues route to which channel, who has access, and what the weekly report measures are decisions, made once and revisited rarely. Those moved.

Notifications, of course, remain entirely in Discord. That was never in question: Discord is where the server experiences Scout, and the dashboard is only where it is configured.