Sure, not being around makes it kind of difficult :-)

This is where a network aware factorer (be it P-1 or ECM) would be useful. It would:

1) Allow you to reserve specific k,n pairs.
2) Provide progress updates to the server. You could then install it on various machines and use the server's website to see the progress of all of your machines.
3) When providing an update (say every 30 minutes) the server could say "Don't bother, just got a factor for it."
4) The server would keep a record of the B1/B2 bounds (for P-1) and B1/B2/#curves for ECM. This prevents duplicate work (for P-1) and also helps determine digit-level testing for ECM. If large storage is available then P-1 save files could even be stored for future bound extension.

And even better:-

5) It would allow you to point the client at the server (with no specific range assigned) and be given some work to do, a la PRP client.

However, as mentioned in the Sieving forum regarding a networked sieve client, the hard part is not having this idea. It's actually coding it, integrating it with the current site and actually finishing it to such a level that you are confident that there aren't any problems. (Even finishing it is difficult, so many projects like this start out with good intentions but then die a slow death).