Collecting what agrees with you and never noticing the rest

The searching is honest, the reading is honest, and the result is still a pile of evidence pointing one way.

Once a view is formed, attention quietly reorganises itself around it. Supporting cases get noticed, filed and remembered. Cases that fit badly get skimmed, explained away, or simply not registered as relevant. Nobody experiences this as dishonesty, because at no point do you decide to ignore anything. You read what you read, you remember what stuck, and the process feels entirely fair from the inside. You just find that the world keeps agreeing with you, which is a very pleasant thing to find, and the pleasantness confirms the view a little further, so the next round of searching starts from firmer ground than it has earned.

It shows up most clearly in how questions get asked. If you suspect the back road is faster, you will notice the days it was, and you will have an explanation ready for the days it was not (roadworks, an unusual hour, bad luck). What you will not do, unprompted, is count. Counting is the whole cure, and it is deeply unappealing, because it turns a comfortable impression into a number that might contradict you. The version you can actually manage is to note the exceptions when they occur rather than filing them under circumstances.

The other practical step is to look for the thing that would tell against you before you look for the thing that supports you. Deliberately, first, while you still have the energy for it. If you are about to buy something, read the unenthusiastic reviews before the glowing ones. If you have a view about a person, try to recall the last three times they were right. The order matters more than it should, because whichever you read first sets the frame, and everything afterwards gets fitted into it rather than judged against it.