Open with a place and a date, not with a position. The European Commission's 2025 Eurobarometer puts 85 percent of EU citizens calling climate change a serious problem and 59 percent saying they had taken a climate action in the previous six months, on its citizen support for climate action page, so a kitchen-table argument is rarely about whether the problem is real. It is about what changes and who pays for it. The condition that changes the opening: when the other person has already said the subject makes them anxious, agree a different week rather than a softer question.

Age predicts less than the seating plan suggests

That same Commission page reports 38 percent of EU citizens feeling personally exposed to climate risks in 2025 and 81 percent supporting the goal of a climate-neutral EU by 2050 (citizen support for climate action). Those are EU-wide figures for 2025, not a reading of your street, and they leave most of the variation inside every age band rather than between them.

So drop the casting. A conversation that hands the worry to the younger person and the doubt to the older one has decided both answers before either speaks. Ask instead what each of you has noticed where you live, and accept "nothing I would swear to" as a real answer.

One question, one place, one period

A question that works names a place and a period: "What did the front garden look like in October when you moved in?" or "When did the swifts arrive this year?" Both can be answered from memory or from a diary, and both produce something checkable later.

Split the first ten minutes. One person asks and writes down the answer in the words it was given; the other answers without interruption, then they swap. Neither role belongs to an age: the younger person holds the pen as often as the older one holds the memory. If you want a slower version of the same exercise, the walk in what changed on your street works over a single afternoon.

Turn the memory into a record

Example. Two people pick one oak they both walk past. In late March, one of them logs the date its first leaves unfold; the other reads the temperature indicator so they both know what the log is for. The Woodland Trust's Nature's Calendar holds almost 3 million records spanning 300 years (Nature's Calendar), and the same scheme's site puts the earliest of them at 1736 (Nature's Calendar), which is 290 years before a date added in 2026. Meanwhile the EEA puts European land temperature for 2015 to 2025 at 2.20 to 2.29 °C above the 1850 to 1900 pre-industrial level, against 1.24 to 1.32 °C globally (global and European temperatures).

That pairing is the point of the exercise. One leaf date settles nothing by itself; it is an entry in a series long enough to show the warming that the EEA's 2.20 to 2.29 °C for European land describes (global and European temperatures). The person who logged it has done something concrete, and the person who read the indicator can say what the log is evidence for. Keeping one place across a year turns the habit into the seasonal record described in one place, four seasons.

When a number in the conversation is disputed

Write the claim down in the words it was spoken in, before anyone improves it. "It is 2 degrees warmer here" and "European land is 2.20 to 2.29 °C warmer than pre-industrial, averaged over 2015 to 2025" are different statements, and only the second names its geography, its period and its baseline. Most disagreements about climate figures turn out to be disagreements about which three of those are missing.

Then find the page that carries the figure and read the sentence around it together, one reading aloud and one following the chart. The method for that is set out in read an environmental claim together.

If the number survives the check and the two of you still want different things, the difference is about priorities, and no further figure will close it. Say so plainly and stop there. A conversation that ends with "we agree on the 2.20 to 2.29 °C and disagree about the parking" has done more than one that ends with an unearned consensus.

Sources