THE QUESTION
Why do false claims spread faster than corrections?
Group cards you judge alike. Groups organise reading; they decide nothing.
closed · back to Actions
Viewing as guest. Enter a handle to join and take part.
Clusters form from similarity, names come last
Canonical order: link pairs of ideas you judge similar — links are symmetric (A~B is B~A; direction never matters) and a tie holds once 2 members make it.
why
Links are symmetric — A~B is B~A. Clusters are the groups that emerge from held ties, and only a formed cluster can be named, by a person. Your own links carry a ✕ to withdraw; you cannot remove someone else's. Clusters organise the wall and never gate the vote — two cards making the same CLAIM are a merge back in Clarify.
Links and names are made while Cluster is live; the steward’s ← reopens it.
unnamed cluster — name it if the group has a common thread, or leave it unnamed (2)
So the incentive is to revise quietly rather than loudly
~ Independence is asser…Independence is asserted rather than structurally arranged (18)
~ Being wrong in public…Being wrong in public costs an institution more than being slow (17)
Independence is asserted rather than structurally arranged
~ So the incentive is t…So the incentive is to revise quietly rather than loudly (18)
unnamed cluster — name it if the group has a common thread, or leave it unnamed (2)
Emergency powers suspend the checks that would have caught it
~ An error admitted ear…An error admitted early would have cost far less than one admitted late (15)
An error admitted early would have cost far less than one admitted late
~ Emergency powers susp…Emergency powers suspend the checks that would have caught it (15)
unnamed cluster — name it if the group has a common thread, or leave it unnamed (2)
Anyone with relevant expertise has funding from someone with an interest
~ Declaring an interest…Declaring an interest has replaced avoiding one (13)
Declaring an interest has replaced avoiding one
~ Anyone with relevant …Anyone with relevant expertise has funding from someone with an interest (13)
unnamed cluster — name it if the group has a common thread, or leave it unnamed (2)
Being wrong in public costs an institution more than being slow
~ Guidance was stated m…Guidance was stated more confidently than the evidence supported (22)
~ So the incentive is t…So the incentive is to revise quietly rather than loudly (17)
Guidance was stated more confidently than the evidence supported
~ Being wrong in public…Being wrong in public costs an institution more than being slow (22)
~ Reversing guidance is…Reversing guidance is used as proof the whole thing was a lie (15)
~ Nobody is assigned to…Nobody is assigned to look for evidence that the advice was wrong (15)
· not yet clustered (31)
Fact-checking grants come from the bodies whose claims are checked
A minister needs an answer this week and the study reports next year
A quiet revision looks like concealment even when it is not
Nobody is assigned to look for evidence that the advice was wrong
~ The models were right…The models were right about direction and wrong about magnitude, and only magnitude was reported (14)
~ Advice must be given …Advice must be given before the evidence is complete or it is not advice (13)
~ Guidance was stated m…Guidance was stated more confidently than the evidence supported (15)
An advisory committee's dissent is recorded and not published
A negative result is not publishable, so the record is skewed
Data behind the guidance is not released with it
Saying 'we do not know yet' is heard as incompetence
A fact-checker who contradicts a funder does not get renewed
Advice must be given before the evidence is complete or it is not advice
~ Nobody is assigned to…Nobody is assigned to look for evidence that the advice was wrong (13)
A correction arrives long after the decision it should have changed
Nobody funds the checking of the checkers
Committee members are chosen partly for being unlikely to make trouble
The press asks for a number and prints the number without the range
Political timelines and evidence timelines are different lengths
The evidence base is built from studies designed by interested parties
Retracting a paper takes years and the claim travels in days
Nobody outside can reproduce the analysis that produced the number
Nuance is stripped between the briefing and the headline
The same handful of experts sit on every committee
Reversing guidance is used as proof the whole thing was a lie
~ Guidance was stated m…Guidance was stated more confidently than the evidence supported (15)
Journals gain nothing from publishing a correction
There is no routine moment at which guidance is re-examined
A hedged message is reported as no message at all
The bodies that issue guidance also fund those who check it
Nobody has ever measured what a late correction costs
Officials learned that hedging gets them attacked from both sides
The models were right about direction and wrong about magnitude, and only magnitude was reported
~ Nobody is assigned to…Nobody is assigned to look for evidence that the advice was wrong (14)
Preprints move faster than review and are quoted as settled
The team that wrote the guidance reviews the guidance
The first statement sets the frame and every revision fights it