In message <[log in to unmask]>, Wirt Atmar <[log in to unmask]>
writes
>Frank writes:
>
>>  "It's also important to remember that software is never intermittently
>>  flaky;"
>>
>>  ... except Windows.
>
>No, beyond the joke value of the statement, software is always deterministic
>in its behavior. What is true that as the software becomes increasingly more
>complex and the state variable table that the software operates within
>becomes so large that you can no longer guarantee (in a mathematical sense)
>its behavior, then it too begins to somewhat take on the attributes of a
>noisy thermodynamic system, but at its core, it still remains deterministic.

I'm not at all sure that I believe this any longer. I believe that
software has left the realm of the deterministic, and is exhibiting
quantum effects.

In fact the scope of any piece of software is not directly what it was
written to do, but anything that the user of that software is not
sufficiently well-informed to realise that it is impossible for that
software to do.

We all know that new users don't just get themselves into situations
because they aren't familiar with the software; they get themselves far
deeper, and start reporting things that you and I know are *quite*
impossible. Even if we wrote it.....

But only ever, of course, if verification is impossible. Just as a
watched pot never boils, so a watched piece of software never
misbehaves. The mere fact of logging events makes the software more
reliable, because it 'knows' that it can't do anything impossible when
being driven by a tyro user, if someone more knowledgeable might look at
the log; and of course, that more knowledgeable person mustn't see
anything (s)he knows is impossible; that would break the quantum rules

All this is conditioned by those twin quantum mysteries, of quantum
causation (which is like ordinary causation with an extra degree of
freedom; it isn't limited to only one direction in time, for instance)
and locality, sometimes called action at a distance.

The software doesn't really 'know' it's being watched (that would be
absurd, of course, at least for a few years yet), and yet it behaves as
if it does.

(In this case, literally 'behaves', in that it doesn't misbehave).

Because of course, everything in the universe is connected to everything
else; there's no escape from the web of causality, and because of
locality, it's instantaneous. Not, note, near-instantaneous; it is
instantaneous. It makes light look like a slowcoach.

All sorts of testable corollaries flow from this, from the relatively
simple 'logged systems are more reliable' through to some quite bizarre
ones. For instance that if a logged system starts to misbehave, then you
can tell *without looking at it* that the tape must have run out, or the
drive has gone bad, and the tape will prove unreadable. Or if a system
is unexpectedly well-behaved, you may confidently search for the hidden
surveillance cameras....

And you thought software was an inert thing, so if you can't drive it
properly, it's just like it's a car and you're a learner driver? No, it
goes far deeper than that; it's more like you're a novice rider, and the
software is your horse. The feedback is two-way...

But I'm joking of course. Or am I?  Think: does something you once saw,
or even something you see every day, suddenly make sense when it didn't
before?

--
Roy Brown        'Have nothing in your houses that you do not know to be
Kelmscott Ltd     useful, or believe to be beautiful'  Wm Morris

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *