Originally posted April 2010 at http://www.criticalrationalism.net/2010/04/07/corroboration/
When we corroborate a theory (i.e., it passes tests), the theory is better in some way. This is a dangerous statement because being better sounds like it’s more supported.
The way it’s better is this: it is now harder to invent rival theories which are not already refuted by existing knowledge. The scope for rival theories is reduced because they have more evidence they have to be consistent with.
Better tests are the ones which will more greatly reduce the scope for possible rival theories. Corroboration increases our stock of known criticism. More severe tests increase it more.
In this way, we can clearly see that corroboration is distinct from confirmation, and is not a type of confirmation, and does not play a related role to confirmation. It’s role is exclusively criticism oriented, and not support oriented.
This is the only valid way to “support” theories: by building up a stock of known criticisms of potential rivals.