Building balanced arguments

Almost every 8021 question — 'To what extent...', 'How far do you agree...', 'Discuss the view that...' — is asking you to build a balanced argument. But the word 'balanced' is the most misunderstood term in the whole qualification, and getting it wrong caps your mark.

Balanced does NOT mean undecided. Many students think 'being balanced' means presenting both sides equally and then refusing to come down on either — 'there are arguments on both sides, so it depends'. Cambridge calls this an evasion of AO2 and AO3. It is the opposite of what is wanted.

Balanced means: weigh both sides fairly, THEN judge. A balanced argument has two stages, and you need BOTH:

• Stage 1 — Weigh fairly. Genuinely consider the strongest version of each side, not a caricature of the side you dislike.
• Stage 2 — Judge. Reach a clear, supported verdict that follows from the weighing.

Picture a balance scale. Fence-sitting leaves the scale level and walks away. A balanced argument loads each pan honestly, then tells you which way the scale actually tips — and why.

A quick contrast:

• One-sided (capped): 'Social media is terrible. It spreads lies, ruins mental health and wastes time.' — only one pan loaded.
• Wishy-washy (capped): 'Social media has good points and bad points, so it is hard to say.' — both pans loaded, but no verdict.
• Balanced (top band): 'Social media genuinely connects people and gives a voice to the marginalised; yet, measured by its net effect on wellbeing, the harms of misinformation and damage to mental health now outweigh these gains — so it has done more harm than good.' — both pans loaded fairly, then a judged verdict against a criterion.

Once you have decided to argue both sides, the question becomes: how fairly do you present the side you disagree with? The A* discipline here is steelmanning — building the opposing argument up to its STRONGEST, most persuasive form before you answer it.

The opposite is the straw man — putting up a weak, distorted or extreme version of the other side because it is easy to knock down. It feels satisfying, but examiners see through it immediately: defeating a feeble objection proves nothing, and it makes you look as though you cannot handle the real disagreement.

Why steelmanning scores:

• It shows you genuinely UNDERSTAND the debate, not just your own corner of it.
• It makes your eventual refutation far more convincing — if you can answer the BEST version of the other side, your verdict is earned.
• It demonstrates the fair, evaluative thinking that AO2 directly rewards.

Straw man vs steelman in practice — claim: 'The state should limit free speech to protect people from harm.'

• Straw man (weak): 'People who want free speech just want to be allowed to insult others for fun.' — a caricature nobody actually holds.
• Steelman (strong): 'Defenders of free speech argue that the power to silence "harmful" speech is precisely the power most likely to be abused: governments throughout history have labelled inconvenient criticism as "harmful" in order to suppress it, so even well-meaning limits create tools that can later be turned against legitimate dissent.' — the strongest real version of the case.

Notice that the steelman is harder to answer — and that is the point. Conceding a strong objection and STILL maintaining your line is what impresses; defeating a weak one impresses no one.

A useful self-test: before you refute the other side, ask, 'Would someone who actually holds this view recognise it in my description, and agree I have stated it at its best?' If not, you have built a straw man.

Steelmanning tells you HOW to present the other side; the Concede -> Refute/Qualify -> Judge structure tells you what to DO with it. This is the most rewarding evaluative move in the whole qualification, because the band descriptors require counterarguments to be 'acknowledged AND addressed'. This structure does both, visibly.

Step 1 — CONCEDE (acknowledge). Admit, fairly and specifically, that the opposing view has genuine force. Use a real example so the concession is concrete, not a vague 'some people think...'.

'It is true that economic growth lifts people out of poverty: China's rapid development has lifted hundreds of millions from poverty within a single generation, something no environmental policy could achieve alone.'

Step 2 — REFUTE or QUALIFY (address). Now explain why the conceded point does not overturn your argument. You have two tools:

• Refute = show the point is wrong, limited or outweighed. 'However, growth that destroys the environment is ultimately self-defeating, since the same development brought severe air pollution and water shortages that themselves damage health and the economy.'
• Qualify = accept the point partly, but BOUND how far it goes. 'Growth matters, but only up to the point where it does not undermine the natural systems that make future growth possible.' (Qualifying is often more sophisticated than outright refutation — it signals nuanced AO2 thinking.)

Step 3 — JUDGE (link back). Decide, and connect the move back to your thesis. 'So growth cannot sensibly be ranked ABOVE the environment that sustains it — the two must be pursued together, which is exactly why the original proposition is too simple.'

The crucial flagging word. Always mark the turn with a contrast connective — however, yet, nevertheless — so the examiner knows you are now ANSWERING the concession, not extending it. Without the turn, a concession reads as a surrender.

Suppose you have two strong sides on the table. How do you decide which wins? You cannot simply prefer one — you need a criterion: an explicit yardstick against which both sides are measured. The criterion is what makes a judgement 'supported' rather than asserted, and AO2 credits exactly that.

Slippery words demand a criterion. Most questions hinge on a word that means different things to different people — best, fairest, justified, successful, important. Until you DEFINE the standard, your verdict is just a feeling.

Set the criterion early — in the introduction. Then every paragraph can be weighed against it.

• Question: 'To what extent is censorship ever justified?'
• Criterion: 'Censorship is justified only where it prevents direct, serious harm and remains accountable to independent scrutiny.'

Weigh each argument against the criterion as you go.

• Restricting material that incites violence -> MEETS the test (prevents direct harm; can be reviewed).
• Banning ideas merely because they are unpopular -> FAILS the test (no direct harm; suppresses opinion).

The criterion does the deciding for you. The verdict then writes itself: 'Measured by whether it prevents direct, accountable harm, censorship is justified to a LIMITED extent — to protect people from incitement, but not to shield the powerful from criticism.'

Two disciplines:

• Keep the criterion CONSTANT. If 'justified' means 'prevents direct harm' in the introduction, it must mean the same in the conclusion — or the verdict no longer follows from the argument.
• Make the criterion DEFENSIBLE. Choose a yardstick a reasonable examiner would accept ('protects citizens over the long term'), not a loaded one that smuggles in your conclusion ('whatever I personally prefer').

Without a criterion, weighing collapses into 'I think this side is stronger' — which is assertion, not evaluation. With one, your judgement is anchored and your reader can see exactly WHY the scale tipped.

Everything so far — weighing fairly, steelmanning, conceding then refuting, setting a criterion — exists to make one final move possible: the supported judgement. This is where most answers are won or lost.

The fence-sit is a fault, not safety. 'There are arguments on both sides', 'it depends on your point of view', 'ultimately it is hard to say' — Cambridge treats all of these as an evasion of AO2 and AO3. They refuse to do the very thing the command word ('to what extent', 'how far') demands.

But committing does NOT mean being blunt or absolute. The A* insight is that nuance and commitment are not opposites. The strongest verdicts decide AND capture degree and conditions:

• Blunt (commits, but crude): 'So the state should be allowed to limit freedom for security.'
• Nuanced AND committed (A):* 'The state should limit freedom to the LIMITED extent that the threat is genuine, the response proportionate, and the powers accountable — and no further.'

The second decides exactly how far it would go. It is conditional, but it is not a fence-sit, because the reader can state the verdict in one sentence: 'limited, conditional limits only.'

The A conclusion formula* — apply your criterion, answer the strongest counter-point once, then commit with a 'because':

1. Apply the criterion: 'Judged by which system best protects citizens' rights over the long term...'
2. Answer the strongest objection in a clause: 'although efficient one-party states such as Singapore show clean government is possible without democracy, this depends on the calibre of particular leaders rather than guaranteed mechanisms...'
3. Commit with a reason: 'democracy is therefore the best form of government — not because it is the most efficient, but because it alone guarantees citizens the means to protect their rights and replace their rulers peacefully.'

The one-sentence test. Ask: could a reader summarise my verdict in a single sentence? If yes ('democracy is best because it secures rights and peaceful change'), you have judged. If the only honest summary is 'it depends', you have fence-sat — go back and decide.

Add nothing new. The judgement APPLIES what you have argued; it never introduces a fresh example, which cannot be developed and signals weak planning.