There is a moment most researchers know well. You type a reasonable question into a search engine, and the engine returns ten million results — ads first, then SEO-optimized listicles, then, somewhere on page three, the PDF you were actually looking for. The information exists. The path to it is the problem.

The people who consistently find things on the web are not smarter searchers. They just speak a slightly different language. They type site:eur-lex.europa.eu filetype:pdf directive instead of “eu law document”. They put quotes around the exact error message. They subtract the noise with a minus sign. These are search operators, and they are the single highest-leverage skill in online research.

They are also, historically, a pain to use.

The memorization tax

Search operators are not hard. site: limits results to one domain. filetype: filters by document type. intitle: requires the word in the page title. Quotes demand an exact phrase. OR widens, - subtracts. Any one of them takes thirty seconds to understand.

The problem is remembering all of them, in the right syntax, for the right engine, while your actual task — the research question — occupies your working memory. So most people do what the interface encourages: they type plain words and scroll. The operators stay in a cheat sheet nobody opens.

This is an interface failure, not a user failure. We don’t ask people to memorize command-line flags to move a file; we gave them drag and drop. Search operators deserve the same treatment.

What a builder changes

Sirch turns operators into structured, tappable rows. You pick site: from a picker, type the domain, and watch the query assemble itself in a live preview. The syntax is always correct, because you never type it. When you inherit a query from a colleague — or from your past self — an explain mode reads it back in plain words: what it includes, what it excludes, where it looks.

Two things follow from this:

  1. You use operators you would have forgotten. The marginal cost of adding filetype:pdf drops from “remember the syntax” to “tap a chip”. Precision stops being a tax.
  2. You learn the language by using it. After a few weeks of seeing your taps rendered as real syntax, the syntax sticks. The builder is training wheels that teach.

One query, twenty engines

Not every question belongs to Google. Code questions often resolve better on a marginal engine that indexes forums deeply. Privacy-sensitive lookups belong on DuckDuckGo or Startpage. Niche academic text sometimes surfaces only through Yandex or Mojeek. The habit of re-typing your query into a second engine is exactly the friction that keeps people on the first one.

A built query is portable. Sirch launches the same construction in any of twenty engines, in whatever browser you already have installed — or in all of them at once, when the question matters enough to compare. Switching engines stops meaning rebuilding.

Queries as assets

A query that took four operators to get right is worth keeping. Sirch keeps a full history of what you launch, and the keepers go into folders — the OSINT pattern for tracking usernames, the filetype search for government reports, the exact-phrase recipe for chasing a bug’s error message. On iOS they sync to your own iCloud; on Android, to a cloud folder you choose. No account, and nothing on our side to leak — there is no our side.

The web is not getting smaller, and its good answers are not getting easier to surface. The language that cuts through it has existed for decades. It just needed a better keyboard.

Sirch is available for iOS and Android. The free tier is the full app, three launches a day; a one-time purchase removes the cap.