I will try to make another in-flight video later someday, showing both the iFly and our altimeter in the same shot, which should help to make my point better. (Trying to make a video without screen reflections in a Glasair isn't easy. I will probably have to do it at night.) But for now, here's a very short clip of the iFly VP course predictor wiggling up and down as we fly straight and level in turbulence.
https://vid.me/F6rA (case sensitive.)
My flying partner is trying to hold 8500' on a day with continuous light turbulence. Since this flight was VFR, we prefer to hold pitch and let our attitude wander within a couple hundred feet plus or minus in the bumps. But in general, I would characterize this as flying straight and level here.
Although the GPS altitude moves all over the place (at least when you look at the "ones" position), we're only being bumped up and down by 50 feet or so. Naturally, the big hand on the altimeter moves up and down (slowly) 50 feet. The VSI isn't moving much. (Conventional VSI - that is, not an instantaneous VSI in the plane.)
But the way the VP vertical course predictor moves, you would think we had started a screaming climb or descent. And, for the latter, we are headed down into a mountain as we fly straight and level.
Likewise, and more problematic, the vertical course predictor tends to jump around this way during descent in mountainous terrain (where there is usually turbulence), which is where we tend to use it the most.
So I would like the vertical course predictor to wait (filter) a few seconds longer before it reacts, more to match whether we've commanded a climb or descent in the plane. And more to match the steady state climb or descent in the plane. Perhaps 3 seconds of filtering/data gathering? Or lower the gain of the filter by only looking at the "tens" (or "twenties") position on GPS altitude?
A future video showing our altimeter along with the VP vertical course predictor should help highlight the disparity between the two.