Re: Time Machines, FTL, and P=NP

On 20 veebr, 23:12, Michael Ash <m...@[EMAIL PROTECTED]
> wrote:
> Crown-Horned Snorkack <chornedsnork...@[EMAIL PROTECTED]
> wrote:
> >> If you can get yourself in the time-travel loop (and why not?) then
you
> >> can do things like brute-force keyless entry locks. Have the machine
> >> generate a number, you try it, you tell the machine if it worked, and
if
> >> it didn't it sends the result back in time to try again. To the user
this
> >> will look like the machine just magically comes up with the right
number.
>
> > Apply it this way: go burgling, and report your time machine the code
> > not as soon as the key opens, but after you have escaped and not been
> > caught at all.
>
> > Of course, this runs into the problems if the guards are smart enough
> > to seize your time machine when catching you, and tell the time
> > machine that you were not caught...
>
> Right, I was trying to steer clear of that. I had thought of an even
more
> dangerous variant: have the machine generate routes to work (or any
other
> destination) along which you drive at top speed. When you arrive, you
> report back and it knows how long it took. If you don't report back
within
> a reasonable amount of time, it assumes you died or got arrested and
tries
> something else. But if the machine were to malfunction in a timeline
where
> you died, that would be somewhat unfortunate.
>
Why?

If you die, or break the time machine, or lose it, or it is taken away
from you, then the time machine will not report success to your past.
You will, therefore, choose a course of action which results in your
reporting that you survived.

False positives is the bigger problem here (i. e. the time machine
reports you will survive, although you did not).
>
>
> >> Apply this to the stock market... have the machine generate a stock
symbol
> >> as the market opens. You buy, then sell at the end of the day and
tell the
> >> machine how much money you now have. It runs through them all and
settles
> >> on the one which gives the greatest return. To the user, it looks
like the
> >> machine just magically generates an awesome stock pick every day.
>
> >> Having some more fun, imagine picking up women in a bar. Give the
machine
> >> a list of women, and a database of pick-up lines. It feeds you one,
and
> >> you report back how well it worked. This will look like the machine
> >> automatically points you to a woman and gives you a pick-up line
which
> >> will sweep her off her feet.
>
> > This one is a harder task.
>
> > The answer not yes/no (like testing a key) nor a number which can be
> > compared with stored largest number. The answer is your subjective
> > impression.
>
> I see no reason why this couldn't work anyway.
>
> For the simplest solution, just rate your experience on a scale from 1
to
> 10. The machine picks the one with the highest score, based on your own
> subjective decision.
>
> You can go even better, though. After each try, make a recording of
> yourself describing the experience. The machine keeps track of the
> recording you rate as being the best. Each time, you compare your
> experience with the "best" recording, and tell it which one you prefer.
>
> You could even video the whole thing. Your dates may not all appreciate
> this, but what the heck, you have a magic machine which will help you
find
> the perfect woman who *will* let you do it.

Which is a part of the problem here. False negatives.

If you get invited to a fantastic night out which you will be
satisfied subjectively - but, since it is incompatible with you
reporting success in the due time, your failure to report rings a
warning bell...

The magic machine will help you find only the best woman who will let
you do that, not the best woman out of all, including those who will
not let you do that.

Would you appreciate your date recording the experience with the goal
of informing her past?

If a complete stranger approached you and informed you that she has a
video recording of your golden wedding, would you want to see it?