[ByronRACE]

When I'd recalibrate a customers PCM to match a shifted maf curve, I do not ever recall seeing that shift continue in the same direction.

I can make some guesses as to why this is...but they're just that...guesses.

The pores in the ceramic, once plugged, are probably hard to fill with more "gunk" so I'd guess it more or less is a one-way transaction.

I use ether based starter fluid to clean mafs...and I never rub the elements, only spray them. I would not introduce any abrasive of any kind into an engine; no matter how small. That's a very bad idea, IMHO.

What I'd do is get a tweecer and I'll help you with a baseline calibration for the parts on your engine if you want the help. Then, see how it runs. It should be the same. Once that baseline is in there, then using an AFR meter, go out and WOT the car and see what the AFR is above 3000. After you have that, compare it to what the commanded AFR is and then perform a trim adjustment. (it's a multipler...if it's 10% lean, the "global open loop multipler" will be .90 to correct it. Then, see how it runs at WOT again and repeat until commanded = actual. Finally, once you're close enough, apply what you just learned to the maf curve. Scale it, and set the global multipler back to 1.0. Now, any weird idle business should go away because the MAF has just been trimmed by the proper amount.

If that doesn't fix the behavior and you're sure you have no vacuum leaks, the next thing I'd do is plumb up another known good maf in series and see what it's reading at idle, 2000rpm, 3000rpm, etc...you can DIY your own maf curve that way if you can take voltage readings off of both MAFs and have the known-good curve.