Negative prices have persisted in Texas and elsewhere for the past decade; this is not news. It is a function of the tax credit, but also a lack of transmission. When transmission is not available from wind resource areas, the prices will be negative there (and higher on the other end), reflecting the fact that wind has to back down because it can't go anywhere. There are also instances where there is simply more power than there is demand over an entire area, but this not as common; that scenario is actually a bigger problem in California due to the buildout of solar (for which there is no production tax credit, notably). Negative pricing was much worse in Texas a few years ago before they built a backbone transmission system to get wind from West Texas to load in the east. There is no doubt that the spot price of energy on average is lowered by wind; utilities nationwide are signing contracts at $20/MWh or less, well below today's average spot price, fixed for 20+ years. Interesting aside - even before there was much wind, prices in the Pacific Northwest would typically go negative for a few hours in the spring when coal needed to be paid to back down to accommodate spring runoff through the hydro system.

Stupid comparison indeed - in the early days before LCD's became viable, there were plenty of predictions that the replacement of CRTs with plasmas would create capacity requirements on the scale of a half-dozen new power plants in California alone, because plasmas used way more power. So people were worried, actually.

However, no large scale electric car adoption is going to happen without two things - 1) batteries that can last for typical commute-to-work-and-back day (already got 'em) and 2) time-of-use rates and metering that provide significant incentives for charing off peak (night), which we already have to some extent. That's why this is a non-issue.

As an aside, consider that the study's sponsors are also a bit biased, so even if this wasn't true they'd reach this conclusion - utilities love the idea of selling more power, obviously, and they are probably lower carbon emitters on a net basis, particularly in the long term if they raise the value of wind generation by raising the value of power off peak, which is when the majority of wind power gets generated. If the value of that power went up, you'd see even more wind development than we have now, or at least less dependence on tax credits to make it economic.

As a utility employee who models the dispatch of this resource, let me confirm that dozens of utilities already do this. If anything, the story here is how late to the game APS is. As for the likelihood of everyone overriding the control signal, in practice this gets used a few times a year. Usually the technology doesn't even turn off the AC unit's fan, it just turns off the compressor, for an hour at a time at most. It's a pretty minimal inconvenience. It's also much more valuable from the utilities point of view than a solar installation that can't be controlled. You might think that solar is just as good because it puts out the most power when AC load is high, but the ability to get more power instantaneously and on demand (say, when a plant goes down or a 100MW steel furnace comes online) is a lot more useful than an incremental and intermittent PV array. By implementing these programs, utilities are avoiding the cost of building expensive peaker plants with similar capabilities, which in normal, non-peak conditions are too expensive to run, so they just sit idle anyway. That's the only thing that's kind of lame about these programs is that (my opinion), customers who participate are paid a small fraction of their economic value to the utility.

I would say that Taleb's thesis is that statistical methods reliably underestimate the likelihood of extremely "rare" events. It's fair to say that market behavior is generally informed by statistical methods. Taleb isn't claiming to have a better statistical method; in a sense he is simply refusing to accept statistics altogether. To trade using that point of view, he doesn't have to be able to predict when these rare events will happen, he just speculates that they'll happen more frequently than statisical methods would predict. Making that bet by trading options is not all that complicated.