FTFA:

"During the decade from 2003 to 2012 we had wet years of nearly 38 inches of rain and dry ones of less than 4 inches, but the average was still just under 14 inches, meaning there is no drought in the most populous region of the state."

The premise of the article is that there is no drought. That there's enough rainfall, that the problem is water management practices.

While the conclusion may have some validity, it starts from a bad premise that weakens its argument.

The title of the article assumes a drought. The sentence in question is arguing that there is enough water, using the word drought casually (I don't think the writer is arguing whether technically a drought exists).

The basic point is that if people collected rainwater instead of channeling it into the ocean they could have more than enough water. Don't argue at the margins and ignore the main point.

He very much is, you can also see later in the article:

Yes, to solve the "drought" in a few short years, there are two basic tasks that California needs to undertake.

The argument made in the earlier part is clearly intended to be on-the-level as he assumes at this point that the reader is convinced by it enough to not call the current once-in-a-millenium drought a drought anymore.

He's trying to engineer a solution to the drought. Engineering a solution to X does not imply X does not exist; that would be stupid.

But if we want to dwell on semantics -- a drought is defined as being an aberration from some norm. Given climate change, the drought in california isn't a "drought" but a change in norms -- this isn't a "once-in-a-millenium-drought" but, in fact, a new normal (didn't NOA just predict a multi-decade drought in California to be a virtual certainty in the next century?)

As an aside, it's been my experience that titles are often written not by the person who wrote the article, but by someone who is tasked with writing titles that are intended to attract attention, and not necessarily perfectly reflect the views of the author.

What he is trying to do is attack the premise of the article. The leaky bucket method does not work if you overflow the bucket once every ten years and then stop filling until the next flood. The bucket would have to be enormous and is therefore not the solution.

I think it is a simple solution that should definitely be used to the full extent.

Even without the 38" year the average rainfall still makes the argument work; quoting a 4" deficit and singling out the highest year is disingenuous. The lowest year was also an outlier.

Here is the data that I used, the author's is something rather similar:

http://www.laalmanac.com/weather/we13.htm

You can add the following values (from http://ggweather.com/ca2013rain.htm and similar URLS): 2012-2013: 5.85 2013-2014: 6.08

(I actually missed these in my original napkin math, but they make my argument even better, so...)

My point was that if in the original data set, the only way that you get those favorable numbers is if you start at the present and arbitrarily go back until you hit a very high number. Any other averaging of rainfall from the present to a year in the last 20 years looks absolutely horrendous.

That suggests that what we care about isn't just the average, but also the standard deviation. How big would the bucket need to be? How long is it between droughts? Would storing large tanks of water between years have other risks, and how can they be mitigated?