[ExI] What can be said to be "wrong", and what is "Truth"?

Jef Allbright jef at jefallbright.net
Thu Oct 2 16:48:38 UTC 2008


On Thu, Oct 2, 2008 at 9:13 AM, Lee Corbin <lcorbin at rawbw.com> wrote:

> Yet, when we, especially in the west, do stumble upon certain
> kinds of "things which had been veiled", e.g., the speed of light,
> should we really be blamed for claiming that we have advanced,
> that we now have better maps, that the accuracy of our beliefs
> is improved?


It appears we're getting ever closer to mutual understanding of this
point, but I'll offer a small but fundamentally significant
correction:

"Accuracy" is simply not meaningful, independent of context.

I suppose this became second-nature to me during my decades in
scientific instrumentation.  I've lost track of how many times a user
would ask "but how accurate is this tool or technique?" and we would
have to educate them to understand that we could guarantee precision,
or repeatability, or any of several other metrics of measurement
quality, but we could say nothing about "accuracy" which is always
dependent on some reference standard (usually provided to the customer
via some traceable government or scientific standards authority.)

Similarly, coherence over any particular context says NOTHING about
the accuracy of a belief, which belief might conceivably be subject to
radical change with a single new observation that doesn't fit the
previous model.

I'm sensitive to the appearance that I'm repeatedly harping on a few
points which may appear to some to be insignificant or
inconsequential, but I often feel a little like I'd imagine Boltzmann
felt around the turn of the 20th century when he tried to argue
against those who saw science as promising increasingly crystalline
Truth, while he went around undermining their noble and righteous
vision with his talk of vague and mushy statistics and probability --
as if such could be more fundamentally true while providing less
crystalline Truth. Outrageous!

- Jef



More information about the extropy-chat mailing list