Re: [PO] Can a regular Turing Machine provide Protected Memory?

From: Mitch Harris (harrisq_at_tcs.inf.tu-dresden.de)
Date: 08/29/04 

Date: 29 Aug 2004 18:39:44 GMT

Peter Olcott <olcott@worldnet.att.net> wrote:
>"Robert Low" <mtx014@linux.services.coventry.ac.uk> wrote in message
>news:cgsvv8$35r$2@sunbeam.coventry.ac.uk...
>> Mitch Harris <harrisq@tcs.inf.tu-dresden.de> wrote:
>> >TMs that have only write-only-memory are pretty boring. Why would you
>> >limit yourself to such TMs?
>>
>> Who cares? If something can't be done by any Turing machine at
>> all, it certainly can't be done by one which satisfies
>> some additional constraint.
>
>Yet the possibility exists that the conclusion that it can't
>be done by any Turing Machine at all is incorrect.

In the same sense that the possibility exists that the fact that there is
no number for which there is no prime greater is incorrect. Meaning you
didn't address his point.

>That
>is my whole point. So far in about two months no one
>has fully (and correctly) refuted this position.

Pardon me for being repetitive, but you're applying quantifiers wrong.
The proof by contradiction assumes that you take an arbitrary TM (any
possible way of doing it), and show that that causes a contradiction,
therefore, the assumption was wrong, i.e. -no- TM can do it.

Mitch

Relevant Pages