🚀 go-pugleaf

In article <1992Nov21.151212.20315@maths.tcd.ie> tim@maths.tcd.ie (Timothy Murphy) writes:

>>Chaitin/Kolmogorov Algorithmic Information Theory does not give such
>>a limit, for the very reason you note - it depends on a particular 
>>choice of universal Turing machine.

>This is a complete red herring.
>All right, let us agree to take the universal machine
>defined in
>
>@book{davis73
>author = " Davis, M.",
>title = "Computability and unsolvability",
>publisher = "Dover",
>year = 1973,
>isbn = 0486614719
>}


Go right ahead and adopt this Turing machine for all your computing
needs if it makes you happy, but I suspect you'll find it a bit slow
and awkward, as well as difficult to get good repair service for.

As for myself, I'm happy to use whatever machine seems convenient at
the time. In particular, if I'm interested in storing images compactly
for later display, I'm happy to use a "machine" that is carefully
tailored so that there are short "programs" that produce images that
crop up a lot, at the expense of long "programs" for images that don't
occur very often.

What is it about data compression that induces so much confusion in
the layperson? Half of them seem to think it plausible that one
could compress any data set into practically nothing without any
knowledge of the source probabilities, while the other half think 
that knowledge of the source is irrelevant in light of mythical
"theoretical limits" . . .

     Radford Neal