The Gartner hype cycle is a fun if shallow model of how new technology find its place in society. The shallowness comes from the idea that the real productivity of the hype does not really live up to the hype. I think that is blatantly wrong and can be proven wrong by checking the numbers. The only reason it feels the way Gartner describes it is, that we adjust to the expectations and aren't as floored by them the second time around as we were during the original hype.
But let's get to the point: The graph above is a model for how we get stuff like the hypecurve. I think in general people are mentally unable to perform exponential extrapolation. What we know how to do is linear extrapolation. This assumption feels natural to me, and I'd be extremely interested in hearing about studies that back this up. Its plausible, if for no other reason then that the world around us is rarely exponential, but usually involves curves following the powers 1 and 2.
The upshot of that, as can be seen from the graph above, is that we tend to drastically overestimate short term gains of a new technology, but at the same time we underestimate long term gains. A really good example of this is is e-commerce. As far as I know e-commerce today is wastly above even the wildest estimates put forward during the first bubble in the late 90s. The only problem with e-commerce is that it took a little longer than expected to get there.
The plateau of productivity, if it exists in reality, has something to do with the exponential ultimately turning into a logistic growth curve instead just from saturation - when saturation is a factor, which isn't always the case.
[UPDATE and Rethink: of course we're never negatively disappointed, or at least we don't waste much time being graciously pleased by overperformance, so the hype is floored at 0 - which gives a nice explanation for the plateau. You'll have to imagine the disappointment as the feeling we have from coming off the sugar rush of the hype. We do in fact go slightly negative just for a while.]