How to work on Reasoned
Reasoned is not a comment section. It is a public record of the best arguments. To make that work, we work a little like scientists do: claims are backed by evidence, sources are checked and objections are put fairly. This page explains how.
1Three principles
- Show, don’t assert
- If you say something about the world, show where the knowledge comes from. One sourced reason outweighs ten unsourced ones.
- The strongest version
- Put every position the way its most thoughtful defenders would, including the ones you reject. Winning against weak arguments shows nothing.
- Open to changing your mind
- The aim is not to be right but to find out what is. A good objection to your own side helps everyone.
2Classifying a claim
Every claim on Reasoned has a kind. It tells readers what sort of support to expect. Pick the kind honestly, not the one that sounds most convincing.
- Fact
- Checkable and widely supported. Needs a source, but little argument.“The EU had about 449 million inhabitants at the start of 2024.”
- Claim
- Checkable, but disputed or not yet shown. This is where a source matters most.“A 130 km/h speed limit would noticeably cut transport CO₂ emissions.”
- Interpretation
- A reading of the facts. Show which facts it rests on and why this reading fits them.“Falling unemployment is mainly down to demographic change.”
- Opinion
- A value judgement. It cannot be proven, but it can be argued: which value is behind it, and why should it come first?“Individual freedom should weigh more than safety here.”
3Which sources carry weight
Not every source carries the same weight. The rule of thumb: the closer to the origin and the more strictly checked, the better. This list is a guide, not a formula. A weak study in a good journal is still a weak study.
Strong: primary sources and pooled research
The source is where the information comes from, or it summarises the research systematically. If one exists, use it.
Examples
- ·Systematic reviews and meta-analyses (for example Cochrane Reviews)
- ·Peer-reviewed studies in academic journals, ideally linked to the paper itself or its DOI
- ·Official statistics: national statistics offices, Eurostat, OECD, the World Bank
- ·Legislation, court rulings, parliamentary papers and records
- ·Original documents: manifestos, budgets, annual reports, treaty texts
Good: careful reporting with a traceable origin
The source is not the origin, but it works to fixed standards and says where its figures come from. Ideal for understanding a topic. If it cites a study, link that study instead.
Examples
- ·Reports by scientific bodies and public agencies (national academies, IPCC, WHO, public health and environment agencies)
- ·Quality journalism with named sources and a corrections policy (for example Reuters, AP, BBC, The Guardian, The Economist)
- ·Parliamentary research services and civic education agencies
- ·Textbooks and books from established academic publishers
Use with care: usable, but say what it is
These sources may well be right, but they have an interest, a risk, or have not been checked yet. Use them when nothing better exists, and say in “What it shows” who is behind them.
Examples
- ·Think tanks, associations, NGOs, companies and parties: often good data, but with an agenda of their own
- ·Preprints, that is studies before peer review
- ·Single studies with a small sample or no comparison group
- ·Press releases, interviews, opinion pieces and columns
- ·Wikipedia: a good starting point, not evidence. Follow its references and link the original.
Unsuitable: not evidence on Reasoned
Moderators remove these or treat the point as unsourced. If one gives you an interesting lead, look for the original.
Examples
- ·Social media posts, videos and screenshots without a traceable original
- ·Answers from AI chatbots. They can invent what they cite. Check every detail in the original.
- ·Pages with no publisher, no named author and no sources of their own
- ·Forums, comment sections, blogs without expertise and content farms
- ·Satire, and sites known for spreading false reports
4Checking a source
Before you link a source, take two minutes for these questions. Most mistakes show up here.
- 1
Who is behind it?
Who published it, and who pays for it? Look the publisher up elsewhere rather than reading only its own “About us” page.
- 2
Is it the original?
Does the source report first-hand, or pass on someone else? If the latter, follow the link to the origin and cite that.
- 3
Does it really say that?
Read past the headline. A study often says something more cautious than the news story about it.
- 4
How recent is it?
For figures, laws and the state of research, the date matters. Is there newer data?
- 5
How strong is the finding?
How large is the sample, is there a comparison group, has it been replicated? Does it show a link or a cause?
- 6
What do others say?
Do independent sources agree? If only one source makes a surprising claim, be careful.
5Citing a source
When you add a source, Reasoned asks for three things. Here is how to fill them in well:
- Source title
- Who and what: publisher and title, and the year for studies. For example “German Environment Agency: Climate protection through a speed limit (2023)”.
- Link
- Straight to the page, table or study, not the home page. For PDFs, give the page number in “What it shows”. If the source is paywalled, link a free version or the DOI where you can.
- What it shows
- The one finding that supports your point, with a number and what it is a share of where possible. Not “study confirms this”, but what exactly it measured.
Weak
Better
A source can also qualify a point or count against it. If your research turns up counter-evidence, it belongs here too. That makes the page more credible, not weaker.
6Writing good reasons
- One idea per point
- One or two sentences, one claim. If you write “and also”, it is probably two points.
- Specific, not general
- Who is affected, how much, from when? “It costs a lot” cannot be checked; “it costs the federal budget about €3 billion a year” can.
- Keep what and why apart
- A position’s summary says only what it proposes. Why that is right belongs in the reasons below it.
- Search before you write
- Does the claim already exist? Reuse it rather than rewording it, so the evidence gathers in one place.
- Say no more than the source
- Write only as strongly as the evidence allows. “Can”, “in this study” and “so far” are not weakness. They are precision.
7Objecting fairly
An objection always answers a specific point, never a person. It has to say what exactly is wrong. Sourced objections are shown first.
- Not true
- The claim is factually wrong. You should almost always add a source here.
- Overstated
- Right at its core, but too strong: the effect is smaller, or holds only for one group or under certain conditions.
- Doesn’t follow
- The claim may be true, but the conclusion does not follow from it.
- Leaves something out
- An important factor, side effect or affected group is missing, and it changes the picture.
You can object speaking for another position, or for none, just checking. Agreeing with a side does not stop you testing its weak points. Please do.
8Common reasoning errors
These are the patterns we see most often. Once you know them, you spot them in your own arguments and in others’.
- Correlation is not cause
- Two things occurring together does not mean one causes the other. Often a third factor is behind both.
- Cherry-picking
- Showing only the studies or years that suit your side. Counter-evidence belongs in the picture.
- One case as the rule
- A story is vivid, but it is no evidence of how common something is.
- Relative instead of absolute
- “Twice the risk” can mean 1 in a million becoming 2. Give both numbers where you can.
- Straw man
- Distorting the other side to make it easier to knock down. The opposite of the strongest version.
- Attacking the person
- Who says something does not make it false. You may name a source’s interests, but its content still has to be checked.
- False either/or
- Pretending there are only two options. There is often a third position, and Reasoned has room for it.
9Tone and neutrality
- Stay factual
- No insults, no mockery, no guessing at motives. Criticise arguments, not people.
- Neutral overviews
- A topic’s overview frames the question so every side could sign it. Loaded words belong in the positions.
- Fair names
- Positions carry the name their defenders would choose, not the other side’s label for them.
- Stay honest
- No invented quotes, no altered figures, no second accounts. Doing so costs your contributions and your account.
The binding rules are in the terms. This page explains how to work well; it does not replace them.
10When you are not sure
- Ask for a source
- You doubt a point but have no counter-evidence? Ask for a source. It is a request, not an objection, and it stays visible until someone adds one.
- Not sure yet
- If you have no stance after reading, “not sure yet” is an honest and good answer.
- Report a mistake
- Found an error, even in your own contribution? Fix it, suggest an edit or tell us. Corrections are welcome, and every change stays visible in the edit history.
The best way to learn is by doing. Topics that need help are a good place to start.