Random web page?

[Terry Hancock]

Peter Hansen wrote:
> 1. Generate random IP addresses.  Easiest way is using an integer
>    which is converted to a net address using tools in "socket".
> 
> 2. Try connecting on port 80.
> 
> 3. If there's something there, display it and you're done.  Else loop.
> 
> Now if you consider that a starting point, for discussion, what
> part of it doesn't serve your purpose, and why?

This easily sounds like the closest one can get to a real answer, but I do 
have one question -- what about virtual hosting?  Aren't a large number of 
websites set up to share IPs using name-based resolution nowadays.

I actually know little about how this is used on the web, but I have Apache 
configured on my testing server to deliver different sites for the same IP 
based on the domain name requested (this is on my LAN).  I figure it's 
probably used a lot, especially for personal web pages and the like.

Perhaps one can query the DNS registry in some way that would be more 
equal?  Can RARP be used to look up *all* domains associated with an IP, or 
only one canonical value?  If you could look up all of them, you could then 
select one at random -- but this will not generate evenly weighted results 
of course.

You *could* make an effort to record look-ups that resulted in large 
numbers of domains to give them more weight in the future.  But this would 
be expensive, and you'd probably have to run for a *long* time for it to 
become relevant.

Of course, the distribution is probably not too relevant -- there are so 
many web pages in the world now, that you are unlikely to retrieve the same 
one twice with any truly random system.  (Not true on Google, where 
connectivity highly biases the search).

I know there's an X screensaver that does something like this to retrieve 
random images.  I doubt they were too concerned about even distribution, 
though.

Cheers,
Terry

-- 
Terry Hancock
Anansi Spaceworks http://www.AnansiSpaceworks.com/