Why I'd rather a student pulled apart a bad song than a great one

Rick Beato asks what makes a song great. Pat Finnerty asks what makes a song stink, and has built a substantial audience doing it.

The joke conceals a real pedagogical claim, and I think it is the right one. A song that works hides its decisions — everything about it sounds inevitable, which is precisely what "great" means and precisely why it teaches so poorly. A song that fails exposes them. The modulation that lands nowhere, the lyric that will not scan, the arrangement with no space left in it: each one makes a compositional choice visible because it went wrong.

There is research behind this, and I want to be careful with it, because the popular version of "we learn from mistakes" is tidier than the evidence.

Who Finnerty is, and why the format works

Patrick Quinn Finnerty is a Philadelphia-based musician, born in Scranton in 1980, who has been playing in bands since he was sixteen — Okay Paddy, And the Moneynotes, Pat Finnerty and the Full Band — and has worked as a sideman guitarist with Dr. Dog and Strand of Oaks. That matters to the argument. He is not a critic who learned some theory. He is a working guitarist who can hear what a session player was asked to do and why it did not work.

What Makes This Song Stink runs roughly 45-minute episodes and is an explicit parody of Beato's What Makes This Song Great?. By late 2023 the channel had passed 166,000 subscribers and 7.74 million views. He adapted it to a podcast in 2022 and makes shorter Little Stinkers alongside. Dave Grohl turned up on his livestream in 2020. His 2021 Weezer episode is credited with popularising "the Weezer conversation" among fans arguing about that band's trajectory.

He has been open that the audience overlap is the point — telling Guitar World he knew millions of people would get the joke because millions of people watch Beato. The parody only functions if the original is the reference text. Which is a neat description of what the format actually does: it inverts the standard analytical move and, in inverting it, shows you the move.

The convention it inverts

Almost all music teaching runs on exemplars. Listen to this recording. Study this score. Learn this piece. The implicit theory is that exposure to excellence transmits excellence.

There is a lot to that, and I am not arguing against repertoire. But it has a blind spot that anyone who has taught for a while will recognise: a student can admire something at length without being able to say what was decided. They can tell you a passage is beautiful. Ask why and you get adjectives.

Hand the same student something that does not work, and something changes. They become specific immediately. The chorus doesn't lift. The bass is doing the same thing as the guitar. It's too long. Those are structural observations, and they arrived without prompting — because failure is legible in a way success is not.

[Lincoln — you'll have a real instance of this. A student who could not say anything useful about a piece they loved, then dismantled something they hated in thirty seconds. Worth putting in your own words.]

What the research actually says

Here is where I want to slow down, because this is usually oversold.

Janet Metcalfe's 2017 review in the Annual Review of Psychology sets out the core finding plainly: error avoidance is the operating norm in classrooms, and laboratory evidence suggests that norm may be counterproductive. Errorful learning followed by corrective feedback is beneficial to learning.

The most interesting result in that literature is the hypercorrection effect — the benefit is strongest when the learner strongly believed their error was correct. Being confidently wrong and then corrected produces unusually durable learning. Being tentatively wrong produces much less. That is precisely backwards from how most teaching is structured, where we work hard to prevent the confident error from ever occurring.

Applied to listening: a student who is sure a song is good, and is then walked through the specific thing that is not working in it, is in the highest-value state that literature describes.

But. The evidence on erroneous examples specifically — deliberately showing a learner a worked example containing a mistake — is genuinely mixed. Research on erroneous worked examples finds they can outperform correct ones, but that the effect is moderated by prompting, scaffolding and the learner's prior knowledge. A systematic review of the conditions for effective learning from erroneous examples concludes that the area remains under-theorised, with contradictory evidence on prior knowledge, cognitive load and long-term effects.

So the honest position is not "negative analysis beats positive analysis." It is: studying failure works under conditions, and the main condition is that the learner knows enough to see what went wrong.

That maps onto teaching exactly. A grade two student listening to a bad song learns nothing, because they cannot separate the fault from the whole. A grade seven student learns a great deal. The technique has a floor.

The counter-argument, at its strongest

The best objection is not that criticism is unkind. It is that it is easy, and easy things crowd out hard ones.

Any teenager can tell you a song is bad. Very few can tell you why a great one is great, and that second skill is harder, rarer and more useful. If analysing failure becomes the whole diet, you produce students fluent in complaint who cannot construct anything — and there is no shortage of those already.

There is a second, sharper version: contempt is a substitute for attention. It is possible to spend forty-five minutes on why a song stinks and learn nothing, because the conclusion preceded the analysis. That is not error-based learning. It is taste confirmation wearing analytical clothes.

Both objections land. Neither is an argument against the technique — they are arguments about proportion and about how it is run.

How I would actually use it

Pick something the student likes. This is the hypercorrection point and it is the whole game. Analysing a song they already disdain teaches nothing, because there is no confident error to correct. Analysing something they defend is where the durable learning is — and it requires you to be genuinely fair about it.

Ask for the decision, not the verdict. "What is wrong with this" invites an opinion. "What did somebody choose here, and what else could they have chosen" invites analysis. Same passage, entirely different cognitive work.

Make them fix it. This is the step most critical listening skips, and it converts the exercise from complaint into composition. Rewrite the pre-chorus. Re-voice the chord. Take out an instrument. Now they own a decision instead of a judgement.

Scaffold it, because the research says the scaffolding is doing much of the work. Do not hand a student a song and an open question. Narrow it: listen only to the bass, only to the drum fill going into the chorus, only to where the vocal sits.

Run both directions. Same week, one song that works and one that does not, ideally sharing a device. The contrast teaches more than either alone.

Never punch down. Analyse a song by someone with a career, not a classmate's composition. The moment this becomes a technique students fear being subjected to, it stops working entirely.

What it is really teaching

The point is not that bad songs are instructive because they are bad. It is that failure exposes intention, and intention is the thing we are actually trying to teach.

When a student can hear that a chorus does not lift, they have understood that choruses are supposed to lift and that specific mechanisms make them do so. They learned the mechanism from its absence — which, on a good day, is faster than learning it from a hundred examples where it worked so well they never noticed it was there.

Finnerty's series works for the same reason a Beato episode works, running in reverse. Both are teaching you to hear decisions. One does it by showing you a decision that paid off; the other by showing you one that did not. The second is funnier, which is why it has an audience — but it is not less serious, and for a student who has been handed a canon and told to admire it, it may be the more useful of the two.

If you want the positive-analysis version alongside it, teaching modes by what they sound like is the same instinct pointed the other way.

Common questions

What is What Makes This Song Stink?
A YouTube series by Philadelphia musician Pat Finnerty that parodies Rick Beato's What Makes This Song Great? Episodes run around 45 minutes and combine genuine musical analysis with commentary on why a song fails. He adapted it into a podcast in 2022 and also makes shorter videos called Little Stinkers.

Do students learn more from mistakes than from correct examples?
Sometimes. Research on learning from errors finds errorful learning followed by corrective feedback beats error avoidance, and the benefit is strongest when a learner was confident they were right. But studies on erroneous worked examples are mixed, and the effect depends heavily on scaffolding, prompting and prior knowledge.

Why is analysing a bad song useful?
Because a song that works conceals its decisions — everything sounds inevitable. A song that fails exposes them. The modulation that does not land, the lyric that will not scan, the arrangement with no space in it: each makes a compositional choice visible precisely because it went wrong.

What is the hypercorrection effect?
The finding that errors made with high confidence are more likely to be corrected and retained after feedback than errors made with low confidence. Being confidently wrong, then corrected, produces unusually durable learning — the opposite of what an error-avoidant classroom assumes.

Is criticising music a useful teaching technique?
Analysis is; contempt is not. The useful version asks what decision produced this result and what the alternatives were. The useless version confirms a taste hierarchy the student already held. The test is whether they come away able to name a mechanism rather than a verdict.