August 11, 2018

Harvard statistician developed an algorithm to determine the likelihood that a song was written by John Lennon or Paul McCartney...

... and pretty much resolves the longstanding controversy over who wrote the music to "In My Life." There's no controversy about the lyrics — only about the music. This Science Friday podcast goes through some of the characteristics of Lennon music and McCartney music, and it's mainly that John was more conventional and ordinary and Paul was creative (for example, jumping whole octaves in songs like "Love Me Do" and "Eleanor Rigby"). But the statistician, Mark Glickman, is not given enough time to tell us about the detail of the music in "In My Life," so I don't know why his algorithm gave that song to John (with a 98.2% likelihood). Isn't there a big jump from one note to the next in the final "in MY-EYE-EYE-EYE life"?

43 comments:

Heartless Aztec said...

Taking the magical mystery out of the music one shitty math equation at a time. I asked one of my old mystic muses I kept - a matchbox from 1968 - and it told me to toke up, kick back and put on the mono mix of Rubber Soul and leave it at that.

rhhardin said...

Statistics gets everything backwards. It omits that if reality follows our models and we use this statistic, we'll misidentify 1.8% of the time.

They do not have reality follows our models, nor do they have a probability for it.

The Crack Emcee said...

Lennon was a rhythm guitarist, so, yeah, he'd probably play more conventional riffs, but he still had the biggest imagination, which is what propelled them. Paul could never have created "I Am The Walrus". Paul also got on with George Martin, so that helped his compositions more than anyone else.

I've always been a fan of McCartney's lyrical baselines. Following his playing always had me imagining some kind of H.R. PuffinStuff type play land, which is odd for a bass player, but he does it.

I don't really listen to The Beatles anymore, which is odd, because I can remember wondering when in God's name they'd wear off.

Francisco D said...

Paul always seemed to be the happy, peppy one.

John seemed more contemplative, edgy and somewhat sad.

Didn't John have a much more difficult upbringing?

Temujin said...

Not sure this was anything more than a navel-gazing exercise by this statistician. Not a study of the music itself. For some of that, you might enjoy Rick Beato's "What makes this song great" videos on You Tube. He's not done any Beatles tunes yet, but he's got some very interesting dissections of great tunes.

Psota said...

John may have been the cool one, but if he had been the leader, the Beatles would have been done after 2-3 albums.

Paul was a music machine who has never really stopped producing/playing at a very high level

Big Mike said...

The fact that they’re from Harvard doesn’t increase the likelihood that their algorithm is correct.

NorthOfTheOneOhOne said...

Why can't people just leave stuff alone and enjoy it for what it is?

Fritz said...

I doubt that he could adequately explain it to someone who has not been trained in multivariate statistics.

"First, imagine a multidimensional football in hyperspace".

On the other hand, statistics is also commonly used to "prove" that which you already think you know, so take it with a dose of salt.

mccullough said...

It would be easier to ask Paul who wrote the song but some people want to try to learn the hard way.

Carol said...

For years I couldn't even tell the singers apart, aside from Ringo. After reading some bio stuff I got so I could tell who wrote what.

But mainly, John was the rocker, and thought the rest was BS. That's more limiting, yet his vision lived on with punk, hair bands, grunge and metal.

MayBee said...

I love that song.

Ann Althouse said...

"It would be easier to ask Paul who wrote the song but some people want to try to learn the hard way."

That's why there's a controversy about it. Paul says he did.

rhhardin said...

It's more than laymen not understanding multivariate statistic. The statisticians don't understand the meaning of confidence intervals.

Ralph L said...

Isn't there a big jump from one note to the next in the final "in MY-EYE-EYE-EYE life"?

Maybe Paul couldn't sing it. I assume John sings it.

Fernandinande said...

Ozzy's "In My Life"

Fernandinande said...

When they don't show a picture, it's probably a Markov chain.

Lyrics: Mark Glickman

No time for a Markov chain
or for models unconstrained
I ran the sampler, it's so slow
this inference I'll never know
this inference I'll never know
No time left for you
No time left for you

tcrosse said...

It would be interesting to apply this algorithm to songs which we know for sure neither John nor Paul wrote, to test for false positives.

Big Mike said...

@tcrosse, I’m sure that’s how they trained the algorithm.

Bleach Drinkers Curing Coronavirus Together said...

Isn't there a big jump from one note to the next in the final "in MY-EYE-EYE-EYE life"

Not sure a one-note transition was meant to be "statistics."

Mike (MJB Wolf) said...

Has he considered collaboration? Snip: "[E]arly on, they would collaborate extensively when writing songs, working 'eyeball to eyeball' as Lennon put it." (From their wikipedia entry). I mean 1965 was halfway into the run of the Beatles, but to me it sounds like them working together, and obviously that high ending was Paul. I like that they used the "by Lennon and McCartney" across-the-board for Beatles songs. It gives doofuses like this professor something to argue about, and provides fodder for those of us still fascinated by their music and how it was made.

Fernandinande said...

Big Mike said...
@tcrosse, I’m sure that’s how they trained the algorithm.


But maybe not - note that tcrosse referred to songs which neither wrote.

This more complete description of their technique refers only to "each Beatles song from 1962 to 1966". (The article's statement about the Unabomber is misleading if not actually false, so perhaps take the rest of it with a grain of salt).


Out-of-sample classification accuracy for [Beatles?] songs with known authorship was 80%.

tcrosse said...

Let's try their algorithm against this:
Knickerbockers Lies
or this:
Badfinger No Matter What

Ralph L said...

apply this algorithm to songs which we know for sure neither John nor Paul wrote

Like "Here Comes the SUM?" That wasn't George, either, a court found.

Bay Area Guy said...

As a general rule of thumb, Harvard statisticians are nerdy Incels who really need to get laid.

Sadly, the statistical probability of that event is quite low.

Pugsley the Pug said...

My favorite song from the Fab Four. A beautiful introspective ballad by Lennon. My wife & I had this song performed at our wedding 24 years ago as it is an excellent song for such an event. Being that it was written and recorded 53 years ago and that Lennon & McCartney added bits to each other’s songs at that point in their careers, I can believe that Paul did contribute a small part to the song but maybe not as much as he remembers 53 years later. Plus George Martin contributed to many of their songs that went uncredited to him. Martin has a beautiful piano solo near the end of this song as well.

Ann Althouse said...

"Not sure a one-note transition was meant to be "statistics.""

What do you think they were using? That's the main Paul-connected factor mentioned in the podcast.

Ralph L said...

No offense, but the song could be heard as a bit of a put-down to wedding guests, even as it expresses the essence of marriage. That isn't necessarily a bad thing, depending on the guests.

tim in vermont said...

I always assumed from the beauty of the progressions and base line that Paul wrote it, but I have no statistics to back it up.

Bay Area Guy said...

Bean counters add very little to the joys of music.

tcrosse said...

Bean counters add very little to the joys of music, but they sure can dance about Architecture.

SteveBrooklineMA said...

I knew Mark Glickman thirty years ago in grad school. Very nice, very smart guy. It made me smile to see his name in the news. As most academics in this situation would, he’s probably thinking “so *this*, of all things, is going to be my claim to fame??”

Mary Beth said...

Not a study of the music itself. For some of that, you might enjoy Rick Beato's "What makes this song great" videos on You Tube.

Last night I was looking at videos YouTube recommends for me and his was one of them. It was the first time I had seen any of his videos. Now that you reminded me of it, I'm going to go back and watch some more.

TWW said...

He also created an algorithm that concluded that John and Paul person were the same and that they were the original Bee Gees.

Narayanan said...

Did he also develop Asian exclusion algorithm for admission office?
If not can he do one to show how it could be done!

Pugsley the Pug said...

Ralph L said...
No offense, but the song could be heard as a bit of a put-down to wedding guests, even as it expresses the essence of marriage. That isn't necessarily a bad thing, depending on the guests.

A put down to wedding guests??? In what way? If the groom invites a former lover as referenced in the first verse and that ex-lover attends, that means they are on good terms and that she is cool with her ex-lover getting married. Lennon is reminiscing in the first verse about people he has known throughout his life in a positive way. In the second verse he is focusing on his current love as his soul-mate (as we know now, that wasn’t the case in real life). Lennon considered this song as he wrote it, a turning point in how he wrote songs as more of a mature writing style and growth as a song writer. This song was considered the #1 love song of the rock era at the turn of this past century in a poll by Billboard (I don’t recall the details from 18 years ago).

southcentralpa said...

That EYE-EYE-EYE jump you talk about really feels more like a last-verse embellishment added in the studio, rather than a dynamic feature of the song as written ...

Birches said...

There is nothing in the song to make me believe John wrote it. It is such a Paul song.

help said...

pubg mobile lite
https://goo.gl/eoQFke

Bleach Drinkers Curing Coronavirus Together said...

What do you think they were using? That's the main Paul-connected factor mentioned in the podcast.

A number of notes or note transitions - way more than one. I thought the Paul contribution was the melody for the "middle eight" section - all these places have their moments. Way up-and-down - singsong, Paul. John said as much.

Bleach Drinkers Curing Coronavirus Together said...

With lovers and friends, I still can recall... that finishes it off. The study author also noted that the section in question was syncopated, another hallmark of Paul's melodies. Compare Penny Lane (syncopated) to Strawberry Fields (un-syncopated). Perhaps the best example of a heavily syncopated Paul tune was Got to Get You into My Life. I think he was copying Motown or Stevie Wonder with it, but it's still syncopated as hell. So is Lovely Rita and Sergeant Pepper. Contrast that to John's Lucy in the Sky with Diamonds or A Day in the Life. The latter was a collaboration and you can hear Paul syncopating the hell out of the middle eight once again - "Woke UP, fell out of BED, dragged a COMB across my HEAD... - all accented on the upbeat/offbeat.

RigelDog said...

I would assume Paul wrote the melody due to the swooping intervals...I don't hear this song as following any sort of rote progression.

Aussie Pundit said...

These two guys were both musical geniuses. I'm confident they could easily do each others' style if required, if the moment or a particular song called for it.

It's so personal that even if Paul had contributed to the melody, John may have felt possessive of this one and liked to think of it as 'his'. On the other hand, it resembles Imagine a lot, which is 100 percent John. (or rather John and Yoko).