Full Idea
Two problems for IBE are that only one explanation can be inferred from any set of data, and that the only data that are relevant to a hypothesis are data the hypothesis explains.
Gist of Idea
Must we only have one explanation, and must all the data be made relevant?
Source
Peter Lipton (Inference to the Best Explanation (2nd) [2004], 04 'Spelling')
Book Reference
Lipton,Peter: 'Inference to the Best Explanation (2nd ed)' [Routledge 2004], p.62
A Reaction
I don't see why the theory prohibits a tie for what is 'best', given that you don't have to commit. The second one is partly to do with what observers should do about anomalies, and it is sometimes right to ignore them.