If you are concerned about the RF then Insteon is defintely out.  They are finally going dual mesh to minimize the problems people see with poor communications.  
I ripped out a system in my house in August of 2008 and replaced it with UPB.  I have not had a single hardware failure and everything just works.   Not saying it is the best type of HA Lighting but it works where Insteon did not.  Just did the same thing at a friends house last weekend and time will tell with that house (but not a single phone call even asking a question or saying something has not worked so far).
If you go for new construction look at a hardwired system which will give you reliability and no rf.
Good Luck with the SH fight.  Been there and done that to many times.
		
		
	 
Hi Digger,
It's been awhile.  You and I (and Sloop) were in the Insteon thing at the beginning.  Yes there were a number of quality issues.  I was fortunate in that I was migrating slowly from X10.  As a result, I was only singed by the early Insteon problems.  I've managed to work through the issues, and in general, SH has done right by me.  Both you and Sloop dove in and had far worse experiences.  So much for the lead in...I won't dredge up old (bad) memories.
I do have a number of opinions on the "dual mesh" Insteon protocol.  Just to be clear - this is not pointed at Digger (I'm really not throwing darts at you).  I am throwing darts at the idea that "dual mesh" devices will fix broken systems.
1) Insteon has been "dual mesh" since day 1.  SH defines dual mesh as the use of RF/powerline Accesspoints (signalincs in the early days) combined with powerline only transceivers.  In my mind, the white paper covers this accurately despite what many people inferred "dual mesh" do include.
2) I believe the "dual mesh" (i.e. Accesspoints) implementation to be the weak link in the Insteon protocol.  I'm sure SH viewed Accesspoints as a "user friendly" method of bridging phases.  The bottom line is that they are a point source of intermittent failure that can be affected by either RF or powerline interference.
3) Systems with increasing numbers of AP's are increasingly difficult to troubleshoot.  When a device cannot communicate, it's virtually impossible to determine whether it's a powerline or RF communication problem (or both).
4) I view the addition of true "dual mode" devices as further complicating the Insteon systems.  Without a utility to track the communication path, systems with multiple dual mode devices will defy troubleshooting methods.
I attribute my "good" experience with Insteon to the fact that I've had a passive phase coupler at my main panel since day 1 (legacy X10).  I'm currently running my 4500 sq. ft. home with the passive coupler and 1 AP (for motion sensors/remotelinc).  With this installation, I can test and troubleshoot with relative ease.   With 5 people in the house I do have "known" problem devices that can become mobile.  With strictly powerline communication, I can easily determine a problem area and pick out the offending device (yes I've mapped my circuits - it's an old X10 habit).  I use a total of 3 filters in my home.  
I periodically have problems with the single AP going off-line (I get communication errors with the motion sensors).  Sometimes the AP self corrects, at times I need to power cycle it.  I can't imagine depending on two of these devices to link all communications between the power legs.
So, am I unhappy with my Insteon system?  Not in the least.  It's an extremely cost effective HA implementation that, when combined with an ISY-99, is very capable and flexible.  As I indicated earlier, SH has stood behind their product and replaced a number of early "problem" devices (I purchased exclusively from SH).  
In summary, while some see the dual band devices as the fulfillment of the "dual mesh" promise, I see storm clouds coming...  I firmly believe that relying on RF coupling to get around problem areas is nothing more than a band-aid.  If you don't understand the nature of the problem, it will bite you down the road. 
IM