The results of the vagueness experiment were
quite encouraging. Here’s the philosophically interesting part of what
happened. The subjects were asked whether they thought n minutes after midday was late (for the relevant appointment) for
increasing n until they finally said Yes at some point, call it k. Then they were asked again about
whether k-1 minutes after midday was
late. 18 out of 19 respondents (I know it’s a small sample, but it’s something)
said No, just like they had when
asked the question the first time, and despite having just said that k minutes was late, and despite that
answer being displayed in boldface just above the question. Various contextualists (I don’t have the references with me, but I’ll
try and find them later and update this post) have claimed that saying that k minutes after midday is late creates a
context where it is no longer true that k-1
minutes after midday is not late, and so when this question is asked the common
answer should be Yes. But only 1
respondent so far has said that. Some contextualists
have been relatively cautious about what the empirical consequences of their
views should be, but some have been
relatively bold about what they think will happen in experiments with just this
design. It would be good to do this test more rigorously, with random sampling
and a larger sample, but 18-1 is a pretty big split even allowing for those design
flaws. Much thanks, of course, to everyone who has taken the experiment. I will keep it up, and keep the counters running, though now that I’ve said what the experiment was testing for the flaws in the experimental design are even more alarming.

UPDATE: It is actually 17-1 in favour of being consistent, not 18-1 as I reported above. My apologies.