Security and trust
A public rating site is only worth something if people can trust the numbers. Here is how we protect against bots and fake ratings, what we store, what we never store, and what is still a risk.
Protections that work today
Each phone gets a random anonymous token that our server signs, so nobody can invent one. A second rating of the same person within 30 days is blocked.
Every action (rating, report, voice note, suggestion, feedback) is limited per phone and per network each hour, and new tokens are limited per network each day. Bursts from one source get slowed down.
If you allow location, we check your phone is within 1,000 feet of the courthouse and keep only yes or no. Ratings made near the courthouse are marked, and every profile shows how many are. It is a signal, not proof.
A comment is invisible to everyone except its author until a human moderator approves it. Nothing is published automatically.
Only phones that have rated someone can report a comment. When two different phones report one, it is hidden until a moderator decides. A report can never overturn a moderator’s approval.
Every decision is public with the moderator’s name and the reason. The moderator tools are not on the public site. They live at a separate address that needs a personal sign-in with a password and a one-time code.
Comments are plain text only. Markup, scripts and control characters are stripped. They are limited in length.
No analytics, no ads, no outside fonts or scripts. Everything loads from our own server, so no company learns who visits.
Encrypted connection, a strict browser rule that blocks any outside code, no embedding in other sites, no referrer sent, same-site-only requests, and protection against fake form posts from other websites.
No score is shown until at least 5 people have rated, and group results appear only when each group is large enough. This makes it hard to guess who rated.
What we store, and what we never store
- We store: your ratings and optional comment; whether you said you won or lost; your role; whether your phone was near the courthouse (yes or no); the time rounded to the hour; and the random token of your phone.
- We never store: your name, email, phone number, exact location, or your IP address. Your network address is held only in the server’s memory for a short time to slow down bursts, and is never written to the database.
- Voice recordings are turned into text and deleted immediately. They are never kept.
- The token on your phone is random and tied to nothing about you, but it does link your own ratings to each other. After 30 days we cut that link. Clear your browser data and the token disappears from your phone.
What can still go wrong, honestly
- A determined person with many phones, different networks or a location-faking tool could still stuff the ballot box.
- Clearing your browser data, or using another phone, gets a new anonymous token, so the one-rating-a-day rule is a speed bump, not a wall.
- The courthouse check can be fooled by a phone that fakes its location. That is why we show how many ratings were made near the courthouse instead of pretending it is proof.
- People who are angry, or who are supporters, are more likely to take part. Any open rating site has this bias, which is why we show whether raters won or lost.
- Rating a real, named official carries a risk of unfair or false statements. We reduce it with fact-only questions, moderation and a correction process (see below), but we cannot remove it.
Planned before public launch
- A puzzle that the phone solves by itself (self-hosted, no outside company) before ratings and other submissions, harder when activity looks unusual.
- Automatic alerts for unusual bursts: many ratings from one network or in a few minutes.
- An independent security review (a test where experts try to break in). To report a weak spot now, write to security@rateyourjudgecolorado.com.
- A right of reply and a correction process for officials who believe something is false.
- Publishing the code so anyone can verify these claims, and a public audit of the moderation log.
- Terms of use, a privacy policy and a corrections and takedown process, reviewed by a lawyer, and a legal review of publishing ratings about named public officials.
This page lists what is built and what is planned. We will update it as things change, and we will say so when a protection is not there yet.