Duty and a desire to disarm my insomnia won out. I chose the latter.

It was a big mistake.

While Stark’s novel merely offered violence, a robbery caper, and even a soupcon of sex (he was writing in the buttoned-up early 1960s), DOE’s tome trumped it, keeping me up half the night.

Thrill to its economic analysis! Respond to its engineering assumptions! Quiver at its lifecycle cost assertions!

It is an impressive document — 162 double-spaced pages. For a certain type of demented reader, it satisfies an appetite for an orderly world in which the manufacture, distribution, sales, and installation of 4 million central air conditioners can be expressed in tidy mathematical terms.

The document also has a kind of inadvertent poetry, especially in the acronyms it contains.

Try GRIM — shorthand for the cash flow analysis methodology by which it will torture manufacturers. Or EMPA (not to be confused with OOMPAH), which stands for Energy Marketing Policy Analysis.

But while I savored its richness, I began to get an uneasy feeling around 3 a.m. that one of its main assumptions was seriously flawed, i.e., the number of single-family houses in America. These are the destination of air conditioners up to 65,000 Btuh.

However, DOE doesn’t distinguish between them (of which there are 67 million) and multifamily buildings (the remainder).

Bottom line: One-third of America’s 100 million “households” — multifamily units — are cooled, if at all, by window units, fancoil units, or big central chillers, none of which are affected in this rulemaking.

If you apply the economic and energy impact assumptions to a base of buildings that’s off by one-third, the results will be skewed to that extent. Guys, let’s rethink this part of the analysis.

One other disturbing point: The document, bristling with authoritative references, also cites Yours Truly as a source. Here, I draw the line.

To paraphrase Groucho, any document that cites Tom Mahoney I want nothing to do with.