Authors
Archives
Categories
Search
Quick Hits
Recent Comments
Links
Meta
I’ve been writing up some thoughts on easy knowledge, but they got a little long for a regular blog post. So they’re in this PDF. But don’t think of this as a paper – it’s really a long blog post in PDF form!
Posted by Brian Weatherson in Uncategorized
alex jackson says:
Dear Brian,
Here’s a comment on section 2. I agree that some knowledge is better than others. But I think that’s a matter of the strength of evidential support, not whether the belief is sensitive.
Here’s the case we’re discussing. Suppose the inspector gets a randomly generated list of machines to check. On today’s list is a machine the inspector checked a week ago. The inspector already knows it will still be accurate; but checking the machine again improves the epistemic standing of that belief. I say that’s because checking again gives him even stronger evidence that the machine is still accurate. The suggestion in section 2 is that the epistemic improvement is that the inspector’s belief [that the machine is still accurate] has become sensitive. Before checking, the inspector would still have believed the machine accurate had it become inaccurate. After checking, the inspector would not believe the machine accurate had it become inaccurate.
Here’s a case were the two diagnoses of the epistemic improvement come apart. Suppose the inspector gets a randomly generated list of machines to check. On today’s list, the same machine appears twice in a row. The inspector checks the machine at 9am, and repeats the checks at 10am. At 9:59, the inspector knows that the machine will be accurate at 10am. (Premise:) Carrying out the same tests for a second time does not improve the epistemic standing of the inspector’s belief that the machine is accurate at 10am. Yet doing the tests again makes the inspector’s belief [that the machine hasn’t stopped working since the last test] sensitive. So making that belief sensitive does not improve the inspector’s epistemic position with respect to whether the machine is accurate at 10am. In this case, the inspector’s epistemic position is not improved, because he does not get even stronger evidential support for his belief that the machine is accurate at 10am.
Alex
Brian Weatherson says:
I’m not sure I get the intuitions behind the case. Doing the test does increase (a little) the sensitivity of the belief. And that’s a reason, I think, to do the test. It gives the belief a virtue it didn’t previously have.
I suspect I’m thinking about sensitivity here a little differently to the standard ways. I think sensitivity comes in degrees. If there are a lot of relatively normal ways that a belief could have been false, then a belief is more sensitive if it goes away in more of those possible ways. Most of the ways for the machine to not be working at 10 are ways that it isn’t working at 9, so prior to the test the belief was pretty sensitive. It gets more sensitive after the test. If the machine hadn’t been tested for a day, there’d be a few more ways for the belief in its accuracy to be wrong, so the test generates a little more sensitivity. And so on as we increase the time between tests. That generates the plausible prediction that the longer the time between tests, the more epistemic gain we get from doing the test.
You must be logged in to post a comment.