Optimism Requires Specificity

Public conversation about technology has settled into two unfalsifiable positions. One holds that things will be transformatively better; the other that they will be catastrophically worse. Neither commits to a mechanism, a timeline, or a condition that would prove it wrong, which makes both comfortable to hold and useless for deciding anything.

The test for a real claim

A useful prediction names what has to happen. Not "this will change everything", but: this specific capability improves by this much, this specific cost falls below this threshold, this specific regulatory question resolves this way, and then this becomes possible. Such a claim can be checked in two years, which is precisely why the confident versions avoid making it.

Why specificity is unpopular

Because it exposes you. A vague forecast is never wrong; a specific one usually is, in at least one component. But being partially wrong in a way you can inspect teaches you something, while being unfalsifiably right teaches nobody anything. The willingness to be checked is the difference between analysis and atmosphere.

Ask anyone confident about the future what evidence would change their mind. The answer is the whole measure of the conversation.

Applied to a live example

Take the claim that automation will substantially change knowledge work. Vaguely, it is unarguable and unhelpful. Specifically, it requires naming which tasks, at what accuracy threshold, at what cost, with what verification burden on the human who remains accountable. Once stated that way, the answers differ enormously by domain, and the useful conversation begins: which tasks have checkable outputs, which have expensive failures, which are bottlenecked by trust rather than capability.

The practical version

• Write the prediction down, with a date and a number.
• Name the mechanism, not just the outcome.
• State the disconfirming evidence in advance.
• Revisit on the date and write down what you got wrong.

This is a small discipline and it produces a rare thing: a track record. In a discourse full of unfalsifiable confidence, having been specifically wrong and having said so is a stronger credential than having been vaguely right.