Songwriting · Module One
Seeds &
Implications
A two-week, self-directed programme for getting a song out of your hands instead of out of your judgment. Built from the August 16 lesson.
The diagnosis
You don't have a talent problem. You have a sequencing problem.
You brought me three complaints, and they are all the same complaint wearing different coats.
First: when you start from chords, you only hear the song that already owns those chords. Second: when you start from melody, you can't tell whether it's any good until you hear the harmony underneath it. Third: whatever does come out sounds either like a nursery rhyme or like it's trying too hard.
Underneath all three is one habit — you are judging the sketch as though it were the finished painting. You are asking a four-note fragment to prove it deserves to exist, and of course it can't, because nothing that small ever can. Fragments aren't good. Fragments are fertile. Different test entirely.
So the fix isn't inspiration, and it isn't rage, and it isn't waiting for something to happen to you. It's a smaller starting unit and a rule against judging it for forty-eight hours.
Idea one
The seed
A seed is the smallest thing you can start from that isn't already a song. It has no obligation to be interesting. Its only job is to imply a next move.
- A tempo. Just a click, and a feel about that click.
- A chord progression. Two chords is a seed. Four is generous.
- A rhythmic motif. No pitches at all. Three or four hits and the spaces between them.
- An interval. Two notes and the distance between them.
The rhythmic motif is the one you haven't tried, and it happens to be the exact antidote to your block. Here's the mechanism: a rhythm cannot remind you of another song's melody, because it doesn't have any notes yet. You cannot accidentally write Last Nite out of a clapping pattern. Your trap is a pitch trap. So start in a place where pitch does not yet exist, and arrive at pitch late, when you're already committed to a shape that's yours.
Headphones or speakers on. The first button press starts the audio engine — browsers require a click before they'll make noise.
Interactive · One
The Seed Machine
One bar, eight eighth-notes. Three or four hits, no pitches. Generate a few, or tap your own — then notice how quickly you start having opinions about something with no melody in it at all.
You can also click the boxes directly to turn hits on and off.
Growing it
Once a rhythm is stuck in your body, pitch is almost an afterthought — and that is the point. You are no longer choosing notes in a vacuum; you're dressing a shape that already exists. Five moves, in order:
Interactive · Two
From rhythm to phrase, in five moves
This uses whatever seed is sitting in the machine above. Work down the list — each button does one thing, and you hear exactly what that one thing bought you.
- Lock the rhythm. Repeat it until it's automatic.
- Assign notes to the hits — one key, any degrees. This gives it contour.
- Repeat the figure with different notes. Same rhythm. That's bar two.
- Add one extra note at the end of a cluster. This is what makes it stop sounding like a loop.
- Harmonise last. Chords are the consequence of your melody, not its cause.
Press Different notes, same rhythm five or six times in a row. Some of them will be good. That is the entire discovery — you don't find the melody, you generate twelve and recognise one.
Idea two
The library of implications
Everything you already know about music — every record, every show, every piece of trivia you thought was just enthusiasm — is not decoration. It is a lookup table, and you have never been asked to query it on purpose.
Once you have a seed, the question is never what should happen next? That question is unanswerable and it will freeze you every time. The real question is: what do the songs I already love do at exactly this point?
That INXS song — what do they do? Verse once, repeat it the second time, then go somewhere new. Maybe I try that, but instead of going here, I go there.
McCartney says he still checks in with John when he writes. Every interviewer wants to hear that and he knows it, but what he actually means is mechanical: after all those years in a room together, he can predict whether John would have found a given move stupid, cheesy, or good. John has been dead for forty years. The lookup table has not.
This is also permission
Springsteen's Independence Day and the Wallflowers' One Headlight share melodic bones. Not a coincidence — Jakob Dylan put the words "Independence Day" in the lyric, and when Bruce played it with them at the VMAs, Jakob told him where it came from. Bruce said: I know.
And yet they are not remotely the same song, because everything grown from those bones diverges:
Interactive · Three
Same bones, different body
One melodic shape. Two treatments. Listen back to back — the harmony and the placement of the high note do almost all the work of making these feel like different songs.
Wallflowers: IV–I–♭VII–vi–V. Moves somewhere new every bar; the payoff note arrives almost immediately, up front.
And the choruses have nothing to do with each other at all — Bruce's barely develops, Jakob's does. (In the lesson I called that third chord the "seven." In practice, in rock, it's the flat-seven — F in the key of G. A true diminished vii would sound wrong here, and you'd hear it instantly.)
Somebody heard a good part, thought they didn't take that where it could go, and took it there. The Killers have done it. Bowie did it constantly. Fans get upset about this; writers do not.
So when a progression makes you hear the Strokes — good. That is your library working correctly, on time, as designed. Don't back away from it. Ask the only question that matters: what did they not do with it?
Check yourself
Six questions. You already know the answers.
This isn't a test and there's no grade going in a file anywhere. It's here because you tend to assume you don't know things you demonstrably do know, and the fastest cure for that is watching yourself get them right.
Quiz · Concepts
Quiz · Ear
Same motif, or different?
Two phrases play in sequence. Sometimes they share a rhythm and differ only in pitch — which, per Idea One, makes them the same motif. Sometimes the rhythm itself is different. Say which.
Most people get this near-perfectly within four or five tries and are surprised by that. Rhythmic identity is one of the loudest things in music and one of the least talked about.
The work
Two weeks. Forty minutes a day. Volume, not quality.
You identified the pottery experiment yourself — the class graded on quantity out-produces and out-performs the class graded on the single perfect pot. Good. Now actually run it, which means accepting that most of what follows will be bad on purpose.
Week one — generate
Rhythm only. No pitch.
Metronome on. Record ten rhythmic motifs into voice memo — tap them, clap them, or bang one repeated note. Ninety seconds each, then move on whether you like it or not. Ten.
Add contour.
Take your four favourites from yesterday. Give each one pitches in a single key. Repeat the figure with different pitches. Add the extra note that makes it end. Four phrases, all recorded.
Harmonise backwards.
Now you're allowed chords. Find what fits under each phrase. Same key as the melody. Notice what's happened: the melody came first, so nobody else's progression got to lead you anywhere. This is the answer to your chords-or-melody question — melody first, harmony second, and the melody itself grew out of rhythm rather than out of nowhere.
Off.
Don't listen back. Not once.
Query the library.
Pick the two phrases that survived the silence. For each, ask what a song you love does after a phrase like that — repeat it? climb? drop the band out? Write down three specific songs and the specific move each one makes. Then try the least obvious of the three.
Off — or keep going if you're in it.
If you're in a zone, stay in it. Zones don't take appointments.
Week two — sing it
The voice memo rule.
Take your two strongest sketches and sing them. Into your phone. Nonsense syllables are fine. This is not about wanting to be a singer — it's that piano makes everything sweeter and more childlike than it actually is. When You Were Young was nearly thrown out for being too sweet; written at a piano it's a lullaby, played by a band with a guitar tone it's the record. Your nursery-rhyme problem is very probably a piano problem, and a human voice diagnoses it in ten seconds, because every voice is unique and no piano is.
Change the instrument.
Same sketches, different sound — guitar patch, synth, anything that isn't a piano. Watch your opinion of the melody change. That change is information about the instrument, not about the song.
Off.
Again. This one's harder than the first one.
Now the critic speaks.
Listen to everything. Rank it. Throw things out with enthusiasm. You have an excellent critic — it was never the problem. It was just showing up ten days early and strangling everything in the cradle.
Take one further.
One sketch. Get it from a verse into something that isn't the verse. That's the whole assignment.
If a sketch still sounds like a nursery rhyme
Keep it. Then write words over it that mean something. The gap between a sing-song melody and a lyric with teeth is not a bug — it's a technique, it's most of Springsteen, and it's a fair amount of Sam's Town.
Ground rules
Five, and only one of them is negotiable.
- Melody and chords in the same key. The only hard rule on this page. Everything else here is a suggestion.
- Record everything. The phone is better than the laptop, not worse — it's always with you, and it works at the office where Logic isn't.
- Nothing counts until you've sung it into a voice memo. Until then it's a sketch on a napkin, and napkins are supposed to look like napkins.
- Suppress the inner critic for forty-eight hours minimum. (I said this backwards on the call — I told you not to suppress it, which contradicts everything either side of it. To be unambiguous: silence it while generating, then let it back in on the listening day. It is a very good critic. It is simply early.)
- When you hear another song in your own idea, dissect rather than retreat. What was their seed? How did they grow it? What did they refuse to do with it?
September 2
What to bring.
- Voice memos of at least four sketches — sung, not played.
- One page of notes: three songs you consulted from the library, and the specific move you took from each.
- One sketch you're certain is bad, and the sentence explaining why. This is the most useful of the three and the one you'll be most tempted to skip.