How Verdiktum works
Verdiktum is built around one idea: a conclusion is only as good as the evidence you can trace it back to. Here is the whole method, with nothing hidden.
Two kinds of cases
A claim is a statement that is either true or false, for example a quote, a viral post or a prediction with a date. A mystery is an open question with several competing hypotheses. Anyone with a confirmed e-mail can open a case, and in a mystery anyone can propose another hypothesis.
Evidence
Evidence is attached to a side of a claim, or to a hypothesis which it supports or contradicts. You choose its type: official record, document, data, recording or photo, article, testimony or analysis. Everything except testimony and analysis needs a link to a source. Official records and documents count for more than analysis, because they are easier to verify.
Ratings
Instead of likes there are two questions. How strong is this argument, from 1 to 5? And what is it: a fact, something plausible, or speculation? Ratings from experienced members count for more, ratings from brand-new accounts count for less, and nobody can rate their own entry. An entry starts with a modest weight, so one enthusiastic rating cannot outweigh an entry that ten people examined.
Counter-evidence and support
You can reply to any entry with your own evidence that rebuts or supports it. A strong rebuttal reduces the weight of the entry it answers. Supporting evidence increases it. Replies can be answered too, so a rebuttal that was itself rebutted counts for less.
The live meter
For each hypothesis we add the weight of the evidence supporting it and subtract part of the weight of the evidence against it. The result is shown as percentages that always add up to 100. With little evidence the meter stays close to the middle on purpose. It updates live for everyone watching the case.
The verdict
A verdict is suggested when a case has at least 4 top-level pieces of evidence and one side reaches 65%. A moderator then closes the case with a verdict and a short justification, or marks it inconclusive. A closed case can be reopened when significant new evidence appears. Predictions get a reminder on their date, so the outcome is checked when it can actually be known.
The role of AI
An automated assistant writes a short briefing for each case, suggests what to check next and answers when someone mentions it in a discussion. Everything it writes is labelled AI. It only works from what people have added to the case, it has no vote and it does not decide verdicts. It does not add evidence to community cases. The starter cases published at launch were prepared with AI from public sources; they and their entries are marked, and you can rate and rebut them like any others.
Moderation and safety
Cases must not accuse private individuals without proof, reveal private data or incite harm. Anyone can report content, including people without an account. Moderators give a reason for every decision and authors can appeal.
Levels and points
Points reward useful work, not volume. Your level affects how much your ratings count.
| Level | From points |
|---|---|
| Rookie | 0 |
| Scout | 50 |
| Investigator | 200 |
| Analyst | 600 |
| Inspector | 1,500 |
| Chief Investigator | 4,000 |
How points are earned
- 10 points for opening a case.
- 5 points for adding evidence.
- 15 points when your evidence is rated as strong by several people.
- 1 point for each rating, up to a daily limit.
- 30 points when your evidence is among the strongest on the winning side of a resolved case.
- 2 points for each day you are active.
Badges
- First clue — Added a first piece of evidence.
- First appraisal — Rated a piece of evidence for the first time.
- Voice in the room — Joined a discussion for the first time.
- Case opener — Opened a first case.
- Three-day streak — Active three days in a row.
- Seven-day streak — Active seven days in a row.
- Thirty-day streak — Active thirty days in a row.
- Case builder — Added ten pieces of evidence.
- Appraiser — Rated fifty pieces of evidence.
- Sharp eye — Five pieces of evidence rated as strong by the community.
- Debunker — Three rebuttals rated as strong by the community.
- Closer — Contributed key evidence to five resolved cases.
- Polyglot — Contributed to cases in three languages.