I've been giving this a lot of thought recently! I already have an EVO III intake manifold and putting a full EVO III head setup on a 6-bolt block would be pretty easy.
There are a lot of things to consider here and it is hard to know where to start without making it a muddled mess but I will try. The verdict on which of the two heads is better is still out and to be honest I don't think there will ever be a definitive conclusion because every time people TRY to compare the two, there are always other variables involved.
The EVO III came with 9.0:1 pistons the 2Gs with 8.5:1, that in itself is going to affect off boost driveability and turbo spool. You cannot realistically change to either head without changing the intake manifold, so you then get drawn into debates as to how much influence the manifold has on the figures subsequently produced. Then there are big guns out there making big numbers on BOTH heads which means neither is conclusively better than the other but this comparison is also contaminated by the fact that very few big guns run unported heads and my personal opinion is that once you have ported both heads the gap between them decreases anyway.
Some very respected members have stated that everyone going from a 2G head to a 1G head has seen power gains. Again I very much doubt that when people changed heads that was the ONLY change that they made. I think the argument that the 1G head flows more is likely true but if you can make 600+ hp on a 2G head that's kind of irrelevant especially for the OP looking to run a 16G/50 trim sized turbo.
I think we can assume that Mitsubishi have spent a fair amount of money in R&D on the Evolutions. I therefore very much doubt that you will be disappointed with an EVO III intake manifold, 2G head and a EVO III exhaust manifold and O2 housing. It makes sense that these parts were designed to be used in conjunction with each other. Now I do think that the combination was also specifically designed to be used with a high compression piston in the absence of the Cyclone manifold to assist off boost throttle response and low down torque BUT it is also very clear that you can run a lot more boost on these engines than the factory originally intended so in mid range and top end, I am sure you can make up for the lower compression with more boost and perhaps even get better results, but I think you might find low end a little weak on a 7.8:1 compression ratio.
Now, the Cyclone is proven to work. There is absolutely no doubt about that. 30-40 ft/lbs increase in torque is nothing to be laughed at and for the loss of 1 CFM of flow, I'm frankly surprised that everyone with a 1G head on a street car isn't using one. But given the amount of research that went into that manifold, I doubt Mitsubishi would have just dropped the project on the 2G/EVO heads if they didn't think it was redundant technology with the higher velocity ports and higher compression ratio. I do however think that IF it could be done, the combination of a Cyclone manifold on a 2G head with lower compression pistons and more boost could make for a very potent and very effective setup. The issue here is that there simply is not enough room to blend the runners into the smaller ports of the 2G head. This means either porting out the smaller 2G ports to match the Cyclone (which defeats the whole point of the 2G head) or as has been suggested, using epoxy to reduce the outlet of the Cyclone manifold. The problem with the latter is that I really don't think Mitsubishi just said "Yeah, let's just try a 3/8 diameter for the primary and a 1/2 inch for the secondaries and blend them about three quarters of the way up ... yeah that looks about right!" If you start f***ing with the ratios between the primaries and secondaries not to mention blocking or obstructing the merge point of the two, I think the effectiveness of the manifold could be drastically reduced.
One thing I am working on at the moment (all be it rather slowly) is the AMG Cyclone manifold. The plenum is considerably larger than a 1G or 2G manifold which means that it WILL outflow both and theoretically should not drop off in the top end. What SPECIFICALLY interests me about the AMG Cyclone however is that the runners are supposed to be dual and INDEPENDENT not primary and secondary runners that merge. That means that since when the butterflies are opened the manifold changes runners, you no longer have to pair short and long runners on the same port. I picked up the plenum half of an AMG cyclone on Yahoo Japan for $1.00-USD the other day. Yes $1. It is actually useless without the runner section, BUT it has given me the courage to look at modifying it without too much fear of messing up a rather rare manifold.
My idea is that if I can build 2G sized runners that link the large runners in the AMG plenum to the corresponding ports on the 2G head by the shortest possible route, a good top end potential should be created. But if I then fabricate tubular runners to join the smaller runners in the AMG plenum and pair them with ports further away (similar to the way a tubular exhaust manifold is constructed) and feed them into the short runners at a 45 degree angle from above I can gain even more gains in torque by creating a significantly longer 'primary'. At this stage I have absolutely no idea on ratios or lengths or if it is even possible but the AMG Cyclone seems far more suited for this task, IF as I have been led to believe, the runners are run independent of each other.
Paul.