[DGD] time travel

bart at wotf.org bart at wotf.org
Sun Feb 22 14:50:57 CET 2009


(meh, sent this from the wrong account.. retrying..)

On Sat, 21 Feb 2009 18:50:38 -0800, Shentino wrote
> Personally, when dealing with a RNG I'd just settle for dropping a symlink
> to /dev/urandom :P

Except possibly for being a 'better' source of randomness, what does this
bring over using the random() kfun?

Lets say you have atomic function fun() which along the way reads 10 numbers
from your rng, and now needs to be rolled back because of an error. How are
you going to rollback your /dev/urandom? You might not want to rollback a
random number generator, but when used for procedural generation, you'd have to.

Being able to set the state of your prng (note the p, it is 'somewhat'
relevant for this kind of use) is rather important when using it for
procedural generation because you have to be able to repeat the generation of
data for a specific case (location, time, whatever you base your procedural
generation on), /dev/urandom isn't going to do that for you.

> 
> But I agree that history is very bad at making itself amenable to 
> PCG, owing to the unidirectionality of it.  The arrow of time, if 
> you will.

Arrow of time is only due to our limitation to not move freely in a 4th
dimension (if you want to model time that way). It indicates a constant
movement in that dimension.

When you talk time travel, you already break the unidirectionally of it.

But indeed 'creating' history that makes some sense using a rng won't be easy,
and you are bound to end up with a lot of data as a result of 'deltas' if you
want casuality to work to at least some level

> 
> Otherwise, it would be like looking at a TV screen without a channel 
> tuned in...nothing but white noise.

Modelling time is more difficult possibly, but not fundamentally different
from landscape generation based on a prng. No, it won't look like random
noise, but it will lack casuality or 'flow'.

> 
> Also, considering that the future is derived from the past, and not vice
> versa, the past would probably need to be generated first.  This 
> factor also makes it hard to preserve the present when someone 
> messes with the past.  In fact, *active* time travellers often count 
> on this.  Passive tourists ala Chrono Trigger, not so much.

This problem you have anyway if you want the present to make some sense with
regards to the past.

Bart
--
Created with Open WebMail at http://www.bartsplace.net/
Read my weblog at http://soapbox.bartsplace.net/



More information about the DGD mailing list