Mario Orazio /
10.31.2000 12:00 AM
Beware of the Factoid
SOMEWHERE OUT THERE You might not have noticed that the U.S. airport with the highest on-time arrivals rate in January was the one in Kalispell, Montana, with 87 percent. More interested in on-time departures? That would be Montrose, Colo., at 96.2 percent. Any questions?
"Uh, yeah, Mario, why are you telling us this?"
I'm glad you asked. Lemme see now ... Well, for one thing, that brilliant source of all the news that fits in print, USA Today, told it to me on the very top of the front page o' their "Money" section one day, so they musta thought it important (and, in case you're interested, their source was the Department o' Transportation).
For another, although it ain't exactly an example o' what Disraeli or Twain (or whoever it was) meant when one of 'em said, "There are three kinds of lies: lies, damned lies and statistics," it's worth a wee bit of a looksee. I mean -- why would good ol' McPaper put such a thing on the front page o' the business section? I guess on account of it's supposed to be useful to business travelers, eh?
OK, everybody who's ever flown into Kalispell, Montana, please raise your hand. OK, hands down. Now everybody who's ever flown outta Montrose, Colo., raise your hands. OK, hands down. Seems to me, on the basis o' that there little survey I just conducted, that Kalispell, Montana, and Montrose, Colo., ain't exactly hotbeds o' business travel.
Now, I ain't gonna fault DoT here. I ain't looked it up, but methinks their list goes beyond the top airports in each direction. Matter o' fact, methinks it could kinda be useful to someone who has a choice to know whether Love or DFW has more on-time departures or whether Midway or O'Hare has more on-time arrivals. But McPaper didn't print the useful info that was farther down in the list.
FACTS AND VOIDS
Methinks those stats are what are called "factoids." The word is formed from two others. One is "facts," meaning true statements. The other is "voids," meaning areas containing absolutely nothing of any use to anyone. USA Today supplies these factoids because paper abhors a vacuum.
All o' this popped into my remainin' neuron this lunar cycle when I saw the CBS report on DTV reception in Philadelphia. If'n I recall correctly, it said they got 99 percent successful outdoor reception and 94 percent successful indoor reception. Heck -- that there first number's even better'n the on-time departure stats at Montrose!
There was another CBS report a while back that covered DTV reception in the New York City area. Methinks it claimed a 96 percent success rate. In that one, "success" was when they actually got reception where the model they were usin' (Longley-Rice) predicted reception. In 96 percent o' the places where the model said they'd get pix, they got 'em. And, as best Nellie the Neuron can recall, in 100 percent o' the places where the model said they wouldn't get pix, they didn't.
The CBS Philadelphia tests didn't fudge the "success" word that way. Nope, the only hedge they used was that the success rate was measured only at sites where the NTSC reception was at least marginal.
Hey -- fair enough. Methinks they even used a reasonable definition of marginal NTSC reception -- grade 1 or somethin' like that there, not grade 4.95. Oh, yeah -- they also only counted sites where they were above 15 dB received SNR, which is also legit, but if they counted 'em all, they'da dropped to 90 percent success for the outdoor sites and 86 percent for the indoor. If they also counted the sites with less than marginal NTSC reception, they'd be down to 86 percent for the outdoor and 79 percent for the indoor, but -- hey -- I ain't gonna be picky.
Nah, the only unusual thing I can note about the testin' is that CBS used receivers that might not yet be available to the general public, like Zenith's professional demod and prototypes with Motorola's and NxtWave's latest super-duper demod chips. That, to me, is still altogether legit. If'n the chips work, then someday they'll probably filter down to consumer hands. Lemme tell ya, sweetie, you do not wanna trade your current NTSC TV in on a 1954 model. Time marches on. Technology improves. All's right with the world.
Or is it? Accordin' to what I read, NxtWave's Chris Strolle, who lives in the Philadelphia market, tried gettin' DTV in his own home. It worked in some rooms, but it didn't in others. The success rate was nowhere near 94 percent. It wasn't even up to Kalispell's 87 percent.
And he ain't alone. Sinclair did testin' in Philadelphia and had problems. Motorola did testin' in Philadelphia and had problems. NBC did testin' in Philadelphia and had problems. But CBS didn't (99 percent reception definitely ain't a problem). Hmmm.
Now, then, I ain't sayin' CBS did anythin' wrong. The outdoor testin' is covered in pretty good detail in their report. Yeah, they used a 30-foot mast and a log-periodic antenna with 8 dB o' gain and an 18 dB front-to-back ratio, and sometimes they hadda point it away from the transmitter to get reception, but that's all legit, testingwise.
And they used the time-honored technique o' checkin' things out usin' fixed-azimuth-angle radials at 0, 45, 90, 135, 180, 225, 270 and 315 degrees. They also say, "Points were chosen at five-mile increments ... ." No arguments from me on any o' that.
But ya mighta noticed that I used an ellipsis at the end o' that there sentence I stole from their report. For those o' you searchin' for an oval in the last paragraph, lemme point out than an ellipsis is the name o' the three dots that indicate that I omitted some words before the period. So, through the magic o' "fair use," lemme just give you the whole sentence, eh?
THE SAME SET
"Points were chosen at five-mile increments, starting at 10 miles from the transmitter." Doesn't sound too bad, does it? The report also conveniently provides the coordinates of KYW-DT, the station tested, at 40:02:23 north by 75:14:33 west. This is still entirely, 100 percent legit stuff. Our Beloved Commish is workin' with the same set o' coordinates.
But sometimes I just can't help myself, so I went to the U.S. Geological Survey Web site and looked up the location o' Philadelphia. It's at 39:57:08 north by 75:09:51 west. Then I dumped both sets o' coordinates into a distance finder.
Guess what: They're seven miles apart. Guess what that means: That means that, because CBS didn't collect any data from closer than 10 miles from the transmitter, their Philadelphia outdoor reception tests weren't exactly conducted in Philadelphia. Geez -- the outdoor tests are startin' to look like factoids -- perfectly accurate and not very meaningful as far as urban reception in Philadelphia is concerned.
Oh, yeah, they did do some outdoor testin' in "a urban population area," accordin' to the report. Now, you mighta thought that, if'n you're testin' in the Philadelphia area, Philadelphia makes a dandy "urban population area" (1990 census population 1,585,577). That's what Nellie thought, too.
But CBS thought differently. They chose the urban canyons of Reading, Pa., (1990 census population 78,380). Is there anything approaching a skyscraper in Reading?
Now, then, CBS did do some indoor testin' inside Philadelphia's city limits, and at least a few o' the sites look, from a map, to genuinely be somewhere in the urban canyons. How'd they do compared to the suburban sites? I dunno -- that info wasn't broken out. But maybe this sentence from the report helps: "Measurements showed that indoor reception at sites other than suburban locations could be difficult, however it is noted that NTSC UHF reception at these locations is also limited."
"So, Mario, what does it all mean?"
Methinks it means ya gotta be careful o' factoids. I ain't found a single incorrect statement in the CBS report (but I ain't gone over it with a fine-toothed comb yet). Even the title page is accurate. It's called "KYW-DT DTV Field Test Report," and that's surely what it is. If'n ya read it, ya find 86 percent outdoor reception outside of urban Philadelphia and 79 percent indoor reception at a buncha suburban sites and some city sites that could be tested (everyone's indoor reception testin' is at the whim o' people willin' to allow a mess o' equipment into their domiciles).
That ain't necessarily different from anyone else's testin'. Sinclair found little or no problems with 8-VSB outside of urban Baltimore.
99 PERCENT SUCCESS?
Nope, there ain't anythin' wrong with the full CBS report, and I'm sure the DoT's full report on on-time airport arrivals and departures was pretty good, too. But McPaper excerpted only the useless factoids about Kalispell and Montrose, and some folks have been crowin' about CBS gettin' 99 percent success in Philadelphia.
Guess what: When somethin' sounds too good to be true, it probably ain't.
Homework assignment: See if'n you can find someone at the Consumer Electronics Association who can tell you how they define those 200,000 or so DTV something-or-others that they say have been sold.