A magician friend once told me that the secret to performing good card tricks was the climax. If you knew a hundred ways to locate a card but only one way to reveal it, the audience will think you know one trick. But, if you know only one way to locate a card but a hundred ways to reveal it, the audience thinks you know a hundred tricks.
Words to live by, especially for we amateur George Gallups.
For any given column of numbers, there's only one sum, one mean, and one standard deviation. Worse still, they're boring. Pollsters who woodenly announce their results to the public ("63% of those surveyed could not distinguish between a Kato Kaelin interview transcript and a Unix core dump") have no imagination. Most people didn't pay attention to statistics class in high school. Why should they start now?
Which is not to say that we at Al Poll Central don't believe in generating huge tables of impenetrable numbers. We relish it, in fact. It's our favorite part of the job. But we also like to spare our readers from the MEDGO Syndrome -- i.e., Mine Eyes Doth Glaze Over. That's why *before* we generate the impenetrable tables, we announce the winners in each category via some sort of goofy narrative. We like to soften up our audience first before we hit them with covariances and normalization vectors and skew moments.
For the first BOAP, in 1992, we counted down the ten best songs and albums a la Casey Kasem / Top Of The Pops. The second BOAP (1993/94) featured a mock TV awards ceremony dubbed The Alistair Awards. (Year Of The Cat accepting its second straight Best Album trophy: "You like me! You really like me!".) Both formats meant we had to write a pithy paragraph or two about each song. And therein lies the problem -- how many ways can you write that Al sounds silly singing Lover Man?
So for the third BOAP, we decided on a different tack. The narrative we've penned is....well, 'bizarre' is probably a good term. But it serves our purposes: it's more accessible (though only just so) than a matrix of numbers, it allows ASML'ers time to digest one set of categories while we furiously tabulate another set, and, most importantly, it suitably obscures the fact that, deep down, we really only know one card trick.
Results: Worst Song, Worst Album
Results: Best Classic Song, Best Classic Album
Results: Most Overrated Song, Most Underrated Song
Results: Most Suggested 20th Century Song Subject,
Most Suggested Pre-20th Century Song Subject
Results: Best Melody, Best Instrumentation,
Best
Guitar Solo, Best Album Production
Results: Voter Correlations, Highest Song Standard
Deviation,
Lowest Song Standard Deviation
Results: Best Album Art, Actor to Portray Al, Most Requested Cover Song, Favorite Artists Besides Al
Results: Most Requested Live Song, Best Modern Song, Best Modern Album, Best Vocal Performance
Results: Best Upbeat/Driving Song, Best Romantic/Reflective Song, First Al Song
Results: Best Lyrics, Favorite Line, Additional Verse, Best Instrumental
Results: Gender Gap, Generation Gap, Culture Gap
Results: Best Song, Best Album
A few disclaimers, warnings, and pre-emptive rebuttals follow. None of them will make sense until you begin reading the narrative. They might not make sense afterwards either, but what the heck....
Anyway, I had a choice. I could rewrite it to some unspecified future date. Or, I could leave it the way it was and let everyone scratch their heads. I figured what the hell -- a lot of people were going to scratch their heads over it anyway. Besides, I was ticked off that 'Princess Olivia' finished so poorly again. Serves you guys right.
In Section One (best songs) and Section Two (best albums), voters were asked to rate each of Al's works -- all 139 songs and 21 albums -- on a scale of zero to ten, where 10 was the highest. A "n/o" meant the voter had no opinion, most often because he or she didn't recognize the song. Respondants were asked to distribute their ratings so that they averaged out to a '5.0'.
From a mathematician's standpoint, this system was foolproof. It was consistent, complete, and thoroughly comprehensive. It allowed voters to express their opinions precisely. It allowed Al's older, lesser-known songs to compete on equal footing with his later works. We could draw conclusions on any arbitrary subset of data (e.g., "Best Instrumental") or voters (e.g., "Favorite Albums Among 30 to 40 Year Olds") we chose. In short, it was perfect. We sat back to await a phone call from the Nobel Prize committee in Stockholm.....
Instead, we got a call from Reality. It seems there are a lot of Al-ficionados who would not dream of insulting the Great Master (you'll be seeing a lot of that term soon ;-) ) by rating one of his masterpieces a zero. Or a 2, or a 4, or, in some extreme cases, anything below a 7. Despite our entreaties that the 0 to 10 gradation was relative to Al's works only, and that the average Michael Bolton song might finish at a -3 on that scale, many respondants didn't come anywhere near a 5.0 mean. Or, as one voter eloquently put it, "Groveling apologies, but...my hand starts to twitch if I rate anything below 5!"
Grade inflation, in and of itself, is not a problem. If someone rates all 139 songs on a scale of 7 to 10 instead of 0 to10, it'll push the overall average above 5.0 but it won't affect the final ordering. Put another way, a rising tide lifts all boats.
But grade inflation combined with a lot of "n/o" votes *is* a problem. A quick example will illustrate why. Suppose instead of 139 songs, you were asked to rate only three: Anna, Burbling, and Constantinople. The first three voters give Anna a 6, Burbling a 5, and Constantinople a 4. Mean of 5.0, 100% recognition, no problem. The fourth voter, who worships the guitar that Al strums on, has never heard Anna but awards the other two songs 10's as a gesture of supreme veneration.
Swell. Anna is still stuck at a 6.0 average, but Burbling has slipped past it to a 6.25, and Constantinople is close behind at 5.5. All because one overexuberant voter hadn't heard of one song.
This is an extreme example for a couple of reasons, not the least of which is the fact that Anna couldn't pull a 6.0 average even if the electorate was limited exclusively to people named Anna. But that's beside the point. The equation that matters is: Grade Inflation + Unrecognized Songs = Big Trouble.
That's where 'normalization' comes into play. Put simply, if you submitted a ballot with a mean rating above (or, in a few rare cases, below) 5.0, the staff at Al Poll Central pulled out their sledgehammers. They normalized (read: bludgeoned) your ratings to a mean of 5.0 and a standard deviation of roughly 2.4. Everyone thus had an equal say in the final standings, whether they recognized only a couple dozen songs or all 139. The world should be so egalitarian.
Some voters expressed concern that one of Al's older, lesser-known songs would use a handful of enthusiasts and a boatload of "n/o" ratings to sneak into the Top Ten.
Breathe easy, you purists. The obscure stuff, by and large, got hammered. In fact, one nearly took home the less-than-coveted honor of Worst Al Song. (Thankfully, the last place song was so totally, unequivicably hated that nothing short of a cover of "I've Never Been To Me" could wrest away the title. It finished with an average normalized score below 2.0.)
Still, we didn't want to clutter up the results with stuff that hardly anyone's heard. Our quality assurance department decreed, somewhat arbitrarily, that a song or album needed a recognition factor of 60% or better to qualify for the final standings. In other words, no more than 40% of the voters could mark it as "n/o" on their ballots.
135 of Al's 139 songs so qualified. The four that did not were:
NON-QUALIFYING SONG RECOG SCORE
Elvaston Place 33.9% 3.95
Swallow Wind 30.3% 3.22
The Bear Farmers of Burnham 16.5% 4.81
Remove Your Footwear 4.6% 4.59
15 of Al's 21 albums qualified. Happily, they were his 15 commercially-released studio and concert albums. The six that did not qualify were:
NON-QUALIFYING ALBUM RECOG SCORE
The Best Of Al Stewart 58.7% 4.29
To Whom It May Concern 57.8% 5.08
The Early Years (double) 56.0% 4.28
Chronicles 31.2% 4.78
The Blue Album 21.1% 4.80
The Early Years (single) 19.3% 3.99
The 110 Al connoisseurs who made up the 1995 BOAP blue ribbon panel are listed below. (A '*' means the voter participated in the 1994 BOAP, a '+' means he or she was part of the 1992 BOAP.)
Kari Ahtiala yykaah@uta.fi
Randall Armor randarm@snip.net
* Geoff Arnold geoff.arnold@sun.com
David Artus davida@sbil.co.uk
Jonathan Artus davida@sbil.co.uk
Linda Artus davida@sbil.co.uk
Pete Ball ballp@hpohp2.wgw.bt.co.uk
Shane Barbera s.barbera@cowan.edu.au
Mark Brady mbrady@msmail.his.tch.tmc.edu
Emily Branson ebranson@sage.cc.purdue.edu
Carl Brooks cbrooks@southend.com
Susan Browning susan@phoenix.net
Rob Callum callum@minerva.cis.yale.edu
* David Capeloto capeloto@mamcass.sp.trw.com
Mary Ann Carberry macarber@lynx.neu.edu
+* John Carr jfc@mit.edu
Brian Carroll bcarroll@telebit.com
Brian Chaffin chaff002@maroon.tc.umn.edu
Edmund Chattoe econec@vax.oxford.ac.uk
Richard Clayton richard@turnpike.com
Robin Coles robinc@cix.compulink.co.uk
* Chris Conway wombat@aquilagroup.com
Lisa Davis ldavis@connix.com
Will Dossel wdossel@infi.net
Paul Dreifus pdreifus@cup.org
Steve Dreksler dreksler@nrlmry.navy.mil
Marina Drobnic marid@riter.computize.com
+* Kim Dyer kimbis@snip.net
Colin Edwards cedwards@pavilion.co.uk
Thomas Faison tfaison@bbt.com
Paul Feldman trotglobe@snip.net
Mike Flavin mmfjr@snip.net
Steve Fletcher sqf8r@virginia.edu
Jeff Francom jfrancom@novell.com
AEleen Frisch aefrisch@lorentzian.com
Martin Frischer 100260.3152@compuserve.com
Erik Frost kberg@teleport.com
* Thomas Gill tegill@ucdavis.edu
Jackie Granja sjgranja@ingr.com
Don Greenop memphian@cris.com
Glyn Harmon harman@cf.ac.uk
+* Ben Henwood ben@banjo.apl.washington.edu
Ela Hnizdo ers11@delphi.com
* Bob Hofkin hofkin@software.org
Neil Hogan neilhogan@snip.net
Veronica Honer vmh1954@snip.net
* Richard Horton rhorto01@mail.win.org
Charlie Hulme charlie.hulme@man.ac.uk
* John James chessieres@snip.net
Matt Jarvis matt.jarvis@gsfc.nasa.gov
+* Jeff Johannes jjohanne@indiana.edu
+* Chris Kaiser chrisk@dse.govt.nz
+* Asya Kamsky asya@remarque.berkeley.edu
Noel Kaplan april5@snip.net
* Lowell Kaufman cactus@crl.com
Alana Landey alana2@snip.net
Ken La Salle klacelle@snip.net
Sue Lindley vyper@cris.com
Juhani Lindqvist juhani.lindqvist@jorvi.elisa.fi
Max Logan logan@ctron.com
Douglas MacPhail (none)
Cort McMurray cortmc@ix.net.com
Vicky Millar n0vmr@snip.net
Mark Moerman mjm4@ix.netcom.com
Rona Montag stevenm2@snip.net
* Dave Nachmanoff denachmanoff@ucdavis.edu
Bill Nelson william.e.nelson@sprint.sprint.com
Michael Ozer (none)
Denis Palmer denis@dpalm.demon.co.uk
Keith Palmer kdp@dmu.ac.uk
Alan Parker mkmf51a@prodigy.com
Ken Paulsen chct36b@prodigy.com
Andy Peterman andyp@treehouse.org
* Russell Pierce rkp@druak.att.com
Jeff Porter at449@freenet.carleton.ca
Allan Pratt apratt@taligent.com
Anne Rasmussen zoot@syspac.com
Chris Read chris@gruffin.demon.co.uk
John Reid gozzo@mary.fordham.edu
Christian Resseman ressema@uni-muenster.de
Andy Rex (none)
Graham Rhodes rhodes@sed.ara.com
Deborah Riel driel@black.clarku.edu
Jochen Rindfrey rindfrey@igd.fhg.de
* Don St. John dstjohn@iftw.com
Kurt Schmidt kas61@snip.net
John Shannon jshannon@indirect.com
* Steven Sharpe sbsharpe@informatics.jax.org
Steve Smoot smoot@cs.berkeley.edu
Bob Spector spector@sirius.com
Stephen Spero (none)
Paul Stangas (none)
Doug Steele steele@swttools.fc.hp.com
+* Nick Straguzzi nstraguzzi@snip.net
Kenneth Swope kenswope@umich.edu
Steve Thoburn sthoburn@snip.net
Michael Timmons timmons_michael@mailmn.tandem.com
Mark Twomey mtwomey@dttus.com
Glenn Vallen vallen@ozemail.com.au
Robert Van Howe vanhower@mfldclin.edu
Cesare Vecchio ces.vec@iol.it
Mary Venable venable@nevada.edu
* Steven Vest (none)
Christine Unruh lsroger@dorsai.org
Ken Weiher kenweiher@snip.net
Stephanie Willen swillen@unec.edu
Marshall Wood marshall@fs09.webo.dg.com
Adam Woolfe aw120@cus.cam.ac.uk
Alan Wyckoff jawyckof@msmacab.remnet.ab.com
Nick Yutko (none)
Not surprisingly, the majority of voters are from the United States: 83 in all. 15 are from the United Kingdom, two each hail from Australia, Finland, and Germany, and one each from Canada, Hong Kong, Italy, and New Zealand.
The male-to-female split was 89 to 21, roughly the same ratio as the last poll.
The age distribution: Four voters are 25 years of age or under, 17 are from 25-29 years, 35 from 30-34, 28 from 35-39, 18 from 40-44, and six voters are 45 or older. (Two voters declined to give their age.)
The average voter is 34.4 years old, with a standard deviation of 6.13. He or she has been a fan for Al for 16.7 years (s.d.=5.58) and a member of the ASML for 0.90 years (s.d.=1.00). Finally, he or she has been to an average of 4.4 Al concerts (s.d.=6.27).
Al Poll Central is not a one-man operation, as many mistakenly believe. A whole lot of effort by a lot of far-flung people made the 1995 BOAP possible.
Heartfelt thanks to the Al Poll Central quality control department for proofing the ballot before its general release: Geoff Arnold, Matt Braun, Kim Dyer, and Tom Gill. Thanks also to our crack public relations department of Terry Carroll, who made the ballot available to voters via ftp, and Muffy Barkocy, for advertising the poll on the Al Stewart Web Page and publishing the results as they're released. And, thanks to our chief engineer, Chris Conway, for writing the song and album ranking program used by many voters.
Hats off to Brian Chaffin for running the Great Ballot Exchange in which voters got to view each others' ballots after they were submitted. And of course, thanks to the ASML's most famous member: Dan Farmer, our founding father.
The ballots are tallied, the results are tabulated, and the curtain slowly rises. It seems to us as though we've been upon this stage before....
Nick Straguzzi
CEO, Al Poll Central
nstraguzzi@snip.net