If anyone is putting together a wish list of features for the next iteration of the distributed brute

Sure am ....

would they like to consider some sort of mirroring approach for the server (at least on stats or updates/software)

Not sure what you mean .... Currently there are many machines involved ... sksp odd & sods. sksp-ack just ACKs. sksp-key just keyspace allocation ftp just FTPs www just WWW stats are WWW only, with updates being done by ACKs machine.

so that some of that traffic could be reduced?

How ?

I'm afraid I am one of the guilty trying to log on to check on the stats at least once and I probably contributed to choking it just by doing that.

Nah ....

If a local server

local to whom ? Using a caching poxy ?

could have been updated, it would save bandwidth on the server doing real work.

No -- different server.

Could a trusted group of segemnt dolers be put together?

As in a hierarchy ? This is being thought about ...

I like the idea of running a benchmark type of program so that I could multiply keys/sec times the amount of time I had to donate, and only get a reasonable for me number of segments.

brloop does that for brutessl attempts ...

I would hope that statistics be gathered on the number of keys tested, elapsed time, etc. so that we all got some more or less real world insight into key lengths and strengths, effort required to break, that we have all heard so many projections about... especially as the doling gets more sophisticated and the number of participants/cycle pool increases.

Let me know what stats you want ...

Will there be Hal3?

Yes.