Skip to content

Latest commit

 

History

History
118 lines (88 loc) · 5.7 KB

TESTING.md

File metadata and controls

118 lines (88 loc) · 5.7 KB

Testing this site

Testers: read the live page instead — https://become-expensive.transparencycascade.org/testing It has the same guidance plus one-click pre-filled issue links, and it is the version kept current. This file is the developer-facing copy.

This is an alpha. It works, it is not finished, and the fastest way to make it better is someone who is not its author trying to actually opt out of something with it.

You do not need to be technical. If you can follow a set of instructions and say "this step didn't work," that is the whole job.

What this tool is

A free, no-account, no-server toolkit for getting your personal information out of the data-broker economy by hand. It exists because of a finding worth repeating: a 2024 Consumer Reports study found paid removal services removed about 35% of participants' records over four months, while doing it yourself removed 70%. Free DIY matches the best paid service and beats all the others. The thing standing between most people and that 70% is an hour and knowing where to start — so this is the "where to start."

Before you begin — the one thing to know

Nothing you type here leaves your browser. No server receives your name, address, or progress; it is a static page and everything is stored locally on your device. You can read the code, or have someone technical you trust read it.

The one exception is an aggregate, cookie-free page-view count (see ANALYTICS.md) — it never carries anything from your profile, and it never records which broker you clicked.

That said: this is alpha software handling personal information. If you are in a situation where your address being findable is a physical safety matter — an abuse survivor, someone being harassed, a public official — please do not learn this tool on your live data. Use it with a throwaway profile first, or wait for a later release.

The three things most worth testing

1. Does an opt-out actually work, start to finish?

Pick one broker and follow it the whole way. This is the highest-value report by a wide margin, because it is the only thing the author cannot easily do at scale.

Tell us: did the broker's page look like what we said it would? Were the fields we told you to fill actually the fields it asked for? Was there a CAPTCHA we didn't warn you about? Did you get a confirmation email? Did your listing actually come down, and how long did it take?

2. Is the instruction wrong, or just hard?

Those are different bugs and we want them separated.

  • "The page said to paste a listing URL but there's no field for it" → the instruction is wrong. High value.
  • "I couldn't figure out where to find my listing URL" → the instruction is unclear. Also high value, and more common.

3. Does it make sense before you've done it once?

Read the homepage and the top of the broker list as if you had never heard of any of this. Where do you get lost? What word did you have to look up? The author has been staring at this for months and can no longer see it fresh — you can, exactly once. Spend it.

Known-incomplete, so don't bother reporting

  • Most entries are marked "not formally checked". That badge is honest, not broken, and narrower than it sounds: every entry is drawn from published broker-opt-out research, and an automated checker has been run against all of their links. What is missing is a person confirming that broker's steps against its own page. Finding out that one of them is wrong is exactly the contribution we want — the badge tells you where to look.
  • The Harden section is thin. Device-hardening guides are next, not done.
  • Some brokers are dead. Domains lapse constantly in this industry. Telling us a site no longer exists is a real fix — we drop them.
  • There is no sync between devices. By design. Your progress is on one browser, on one machine. Clearing site data clears it.

What a good report looks like

Short is fine. What we need:

Broker: spokeo.com
What I expected (from the tool): paste listing URL + email, solve CAPTCHA
What actually happened: the form also demanded a phone number, and the
  CAPTCHA failed twice in Firefox before working
Browser/device: Firefox 141, macOS
Did the listing come down? Not yet — 3 days so far

Anything missing from that is still worth sending. A one-line "the Harden link 404s on my phone" is a useful bug.

Where to send it

Use the pre-filled links on the live testing page, or open an issue directly at https://github.com/markramm/make-yourself-expensive/issues/new/choose — there are three tester templates (broker report, site bug, confusing). If a GitHub account is a barrier, reply to the post that brought you here. Do not paste your real name, address, or the contents of your profile into an issue — describe the shape of the problem, not your personal data.

If the bug is in a broker entry specifically (wrong URL, wrong fields, site is dead), that lives in the dataset repo instead: https://github.com/markramm/data-broker-registry — see its CONTRIBUTING.md. One markdown file per broker, no code required.

For developers

npm install
npm run dev        # localhost:4321
npm test           # vitest -- crypto, dataset integrity, CCPA template, badge contract
npx astro check    # typecheck
npm ci             # run at least once before pushing; CI fails hard on lockfile drift

Read CONTRIBUTING.md first — the non-negotiables there (no PII in URLs, no third-party network calls beyond the one allowlisted analytics exception, encryption default-on) are structural, not stylistic, and a PR that breaks one will be rejected on principle rather than on taste.