Quote:
Originally Posted by whswhs
They cannot know that that is true of its current run unless they assume that it's still the same instrument, which rests on its numerical identity. It doesn't matter how well correlated their past runs are with each other. Given any time series whatever, you can come up with a predictive model of it that fits it perfectly—but unless you test that model on new observations, you are not entitled to say that it has any connection with reality.
|
Well, they
can't know, they have to rely on some kind of induction. The only thing this shows is that "numerical identity" in this context can be a useful abstraction, but so can a number of things, including the concept of "soul".
I also fail to see what this has to do with the kind of identity people supposedly have.
If someone made a perfect copy of this measuring instrument, calibration and all, and replaced the first instrument with this copy... does it matter? For the sake of argument we can assume that this instrument has "identity" if you want, but it doesn't show us why we should care about it. The copy of the instrument is just as good as the original instrument and fills the same purpose.