also: i think the majority of issues people have with subaru MAF load systems stem from the changes which are employed in the intake tract. for example, six years ago someone with deltadash (the only available SSM logger at the time) did sample pulls with and without the intake helmholtz resonator (aka "snorkus") attached to the filter box. the pulls without the resonator showed MUCH more MAF scatter, but they also showed increased MAF in the lower spoolup regions. many people, including myself, felt the tradeoff was worth it.
likewise changes in the elbow immediately post-maf sensor, as well as changes to the turbo inlet tube, have shown differences in the MAF translation (which, at first glance, they should not, being both POST-MAF). clark himself disliked tuning different elbows, and there was a thread in which we all thought that the corrugations in the oem elbow were actually there for a REASON (to prevent turbulence from a venting oem BPV from reaching BACK to the MAF sensor).
so, however tempting it may be, pointing a finger at the MAF sensor as the source of all necessary evil is shortsighted. it is certainly my experience that tuning SD setups at higher load is simpler, quicker, more intuitive, and easier (and i said as much nearly 3 years ago in january 07: http://forums.nasioc.com/forums/show...8#post16644828
), BUT i think it is important to know WHY and HOW maf-load systems fall short. for example, as clark considered earlier in this thread, perhaps the temporal sensitivity of the MAF sensor itself (ie, being too quick-acting), combined with non-oem intake tracts (which promote different and more intense resonances), results in MAF scatter that causes combustion chamber conditions to vary too much, stroke to stroke, triggering knock before more stable conditions would have.
in this case a simple low pass filter (series resistance, shunt cap to ground) across the MAF sensor output voltage MIGHT be an incredibly cheap and effective trick to stabilize the load reference and allow one to push a degree or two without knock.
one thing that has not been verified on this group n rom image is whether or not it has the necessary obd2 registers to pass an obd2 vehicle inspection, as is used in my state of RI. this was pretty much THE reason i never jumped ship from oem ecu+utec to a standalone like the hydra years ago. if it does NOT then modifying the car to the point where it will no longer run and/or pass readiness flags to the appropriate registers on a USDM rom is going to result in a lot of trouble when inspection time rolls around for those of us with similar inspection programs.