Think twice / 02
One More Clue
The answer matters. So does the question you ask next.
How to play
You investigate a hidden mechanism using a limited budget. Several explanations may fit the first thing you see. Your challenge is to choose tests that distinguish them, then make an answer your evidence supports. Finishing with the right guess and establishing why it is right are different accomplishments.
- Start with The Signal Room: switch A, B or C on or off, then press Run experiment. The lamp's result goes in your notebook.
- Compare the explanation cards before buying the next observation. Look for a test on which the surviving explanations disagree. Rule out is your erasable note, not an automatic hint; each new observation spends some of your budget.
- Select an explanation and press Make my answer. Read which alternatives your evidence excluded, whether you guessed, and what each purchase added. Aim for a supported answer within budget, then retry or choose a different investigation.
What to notice
A lamp lights up. You have two explanations, and both predicted that it would. Running the same test again may make you feel more confident, but it cannot tell you which explanation is right. You need a test on which the explanations disagree.
That is your job in this room. The Signal Room gives you several possible rules and a limited budget for observations. Set the switches, buy a result and compare what each rule predicted. The answer cards let you select a guess; Rule out is only a mark in your notes. An answer can be correct by luck even when your evidence does not support it.
Karl Popper argued that a useful scientific claim must risk being proved wrong. John Platt's strong inference made the next step practical: keep more than one explanation alive and seek a result that separates them. The same habit helps when debugging: repeat the comfortable test, and a faulty explanation can survive untouched.
The other investigations make the problem trickier. In the Echo Archive, several reports can repeat one underlying observation. Buying every copy gives you more paper, not more independent evidence. Before spending your next point, ask: what result would make me change my mind?
Read more: Popper on scientific testing · Platt, Strong Inference.
Investigation 01
A lamp, three switches, eight explanations
The Signal Room
Choose experiments to distinguish competing explanations.
What you know before starting
The report trail
North observer → North dispatch → Evening bulletin → Harbour radio digest
North dispatch → Safety circular
Harbour observer → Harbour log → Harbour carbon copy
Hill observer → Hill log
The digest is named after a radio station. Its information still comes from North.Build an input
Choose your next observation
Which explanation fits?
Select an explanation when you’re ready.
Select a card for your answer. “Rule out” is a reversible note; it does not check your reasoning.
Investigation closed
Your evidence, step by step
What did your questions buy?
Rules & model
What you are trying to do
Identify the one active explanation from the cards shown. Buy observations, compare them with the possibilities, and select a card before choosing Make my answer. The answer is not changed in response to your guesses.
What counts as evidence
All stated rules and observations are accurate. There is no surprise lying, noisy instrument or secret extra candidate. In the Signal Room you choose a switch input; in the memory machine every sequence starts from a reset. In the archive, source labels tell you when reports share an original observation.
Rule out is only your erasable note. It does not check that a card is wrong or change the possibilities supported by your observations. A correct guess can therefore receive a different explanation from a conclusion that the evidence uniquely establishes.
Budget and the reveal
Costs are shown before you spend. You cannot buy over budget or pay twice for the same observation. Every hidden configuration has a strategy that can identify it within budget, but spending on unhelpful tests can leave you short.
The ending retraces your evidence and shows a cheapest proof for the revealed result. That proof uses hindsight: before knowing an observation’s outcome, you could not assume it would be the helpful one. It is not an intelligence score or a claim that you should have foreseen the answer.
Trying again
Restart restores your budget with the same hidden answer. Different configuration cycles through the checked alternatives. Changing investigation starts a fresh notebook. Nothing is uploaded or saved between visits.