The Cassutt Files
The Sci-fi Ratings Abstract

'Tis summer, the season when a middle-aged man's fancy turns to sports. (What, you thought I would say summer television?)

I suffered through the Lakers' debacle and appreciated Tiger Woods' amazing win at the U.S. Open (on a broken leg!), so I return to baseball, the sport nearest my heart—my father played at the professional level and coached it for 30 years, and I can still turn a double play if asked.

And immediately think about the amazing Bill James, the self-taught statistician who revolutionized the baseball world beginning 30 years ago. James was a night watchman in a factory in Kansas City, and a lifelong baseball fan who was fascinated with—and frustrated by—stats. Batting average. Runs batted in. Earned run average, all those figures that valiantly attempt to quantify a player or team's success or failure in the statistical universe known as baseball.

(For an exploration of the sci-fi possibilities of this idea, find Robert Coover's 1968 novel, The Universal Baseball Association, Henry G. Waugh, Prop.).

James triggered a paradigm shift in that world by re-examining the assumptions behind certain stats. He found compelling evidence that some long-held "truths" of baseball were faulty, and he invented several new stats that have since become standard—range factor for fielders, runs created, secondary average—all of which he originally self-published in The Baseball Abstract.

(In 2003, after years as a lone-wolf commentator on the sport, James was hired by the Boston Red Sox, a team that went on to win its first World Series since the days of Babe Ruth. Coincidence?)

And what, you ask, is the connection between Bill James, The Baseball Abstract and the world of sci-fi television?

It's time for the television business, especially the sci-fi end of it, to apply some out-of-the-batter's-box, Bill James-style paradigm shifting on the subject of ratings.

People meters and packet sniffing

Last month's column ("The Franchise-Premise Differentiation") about the likelihood—or desirability—of returning CBS's canceled Moonlight to the schedule triggered a number of e-mails, some supportive, some not.

Many of those who found fault with my position cited the ratings as proof that the show was popular. The phrase "8 million viewers disagree with you" came up several times.

This figure could only come from the standard Nielsen TV ratings system, and I'm perfectly willing to accept it.

But what does it prove? How do you read ratings?

The A.G. Nielsen Company of Oldsmar, Fla., has been measuring audiences since 1923. (There is another well-known name in ratings—Arbitron—which dominates the radio field.) The standard method has been a diary, given to selected viewers around the nation who faithfully record what shows they watch.

How many? Ah, there's the rub. When I was at CBS in the early 1980s, the Nielsen sample was as low as 1,200 households ... yes, for a TV-watching population of around 150 million bodies.

More recently, with use of a "people meter" that is wired into a TV/cable/satellite/DVR box, feeding data directly to Florida, that figure has risen to approximately 5,000. The people meter still requires you to press a button indicating that you are watching at any given time.

Nielsen also has a separate system measuring local programming, using a sample of 20,000 meters in 55 markets.

The company has ways of measuring viewing shifts for TiVO and DVR, and it's making efforts to account for such under-represented venues as college dormitories, airports—and, yes, the Web through its Nielsen/Net Ratings. (Web ratings are an entirely different and vastly more complex world, beginning with simple log files and now including page tagging and something called "packet sniffing.")

People often complain that these samples couldn't possibly be large enough—obviously not sci-fi fans who have read Isaac Asimov's classic 1955 story "Franchise," in which a single voter determines the outcome of a presidential election.

A further complaint is that the diary holders are never anyone you know—a complaint I never bought, because when I was a child, my family was included in an Arbitron sample.

Whether these samples are sufficiently large is irrelevant—they're large enough for the advertising agencies whose money pays for commercial broadcast fare.

Ad agencies know that only a fraction of the money they spend actually has an effect. As with the biblical seed scattered in a field, only some finds fertile ground.

But there is fertile ground enough to justify spending billions of ad dollars on such series as Heroes and Lost, or to withhold them from Moonlight.

People are fuzzy

I used to be a network executive. I've been a producer. I've seen how ratings reports are examined by professionals—and it is a lot like baseball box scores, where the batter hitting .323 is assumed to be more productive than one hitting .250. Where the score is final: Tigers 9, Jays 4. Tigers win.

The weekly ratings are printed in newspapers and online in a column, as if on the sports page.

I've seen people high-fiving because a TV episode received a 12 share (that is, 12 percent of the available audience) while its competitor got a 10.

The high-fivers didn't seem to realize that the margin of error in Nielsen's system is plus or minus 4 percent. Your 12 could just as easily be 8, while your competitor's 10 could be a 14.

The numbers should actually be presented in bar graphs with fuzzy edges.

Which takes the fun out of it.

And these are just gross ratings figures. We aren't talking about demographics (which show does best in, say, with women aged 18-35—with a correspondingly higher margin of error).

We aren't addressing the reality that of, say, 8 million viewers, some percentage may not like what they see.

The point is, viewers and producers and even pundits treat these figures with far more reverence than they demand.

I have experience with the granddaddy of all sci-fictional ratings systems, the fondly remembered "RealTime Ratings" of Max Headroom ... in which your television actually watched you, recording your interest—or lack thereof.

I don't think that's possible, or desirable.

But I know that we—pundits, producers, networks, advertisers and viewers—need a new set of benchmarks, a handy group of figures that will let us judge a show's success. A Viewer Happiness Quotient. The Eyeball Tracking Number. The First Commercial Break Delivery Number to Three Significant Figures.

A foolish dream? Perhaps. After all, I'm still waiting for the one-button TV remote.