Yesterday I wrote about the camera-off problem — a professional development system that counts hours instead of competency. A provider logs in, turns the camera off, collects the hours. The system registers completion. Nothing changed.

Today I want to go deeper into what that pattern reveals. And I want to start with something from our own field — a finding about how children actually learn language — because it turns out to be the perfect metaphor for what’s broken in how we train the adults who care for them.

What a Baby Taught Us About Learning

For decades, the most influential idea in early language development came from Hart and Risley’s 1995 study, Meaningful Differences in the Everyday Experience of Young American Children. Their finding was striking: by age three, children from higher-income families had heard roughly 30 million more words than children from lower-income families. The conclusion seemed clear. More words in, better language out. Volume of input drives development.

That idea became common sense. It shaped policy, parenting advice, and early childhood programs for years. Expose children to more language, and they will learn more language.

But newer research tells a different story.

In Po Bronson and Ashley Merryman’s NurtureShock, Chapter 10 — “Why Hannah Talks and Alyssa Doesn’t” — they describe a shift in the science that upends the volume model. The research they highlight, including work by Michael Goldstein at Cornell (Goldstein & Schwade, 2008, Psychological Science), found that what drives language development isn’t how much a child hears — it’s whether someone responds when the child speaks.

A baby babbles. A caregiver responds — immediately, contingently — with eye contact, with words, with a touch. That response is what builds language. Not the quantity of words floating around the room, but the quality of the exchange. Serve and return. When researchers compared high-responding mothers to low-responding mothers, the children of responsive caregivers hit language milestones months earlier. Not because they heard more words. Because someone noticed what they were doing and responded to it.

A child who hears ten thousand words from a television learns almost nothing. A child who makes a sound and gets a genuine human response learns to speak.

The old model was: input creates learning. The new model is: responsiveness creates learning.

Now apply that to how we train the adults who care for children.

The Camera-Off Problem, Revisited

Our professional development system is built on the old model. We assume that if we deliver enough content — enough hours, enough Zoom sessions, enough modules — providers will absorb it and their practice will improve. More input, better outcomes. The same assumption we once made about children and language.

But just as a child watching a screen doesn’t develop language the way a child in a real conversation does, a provider sitting through an online training — distracted, multitasking, camera off, no one noticing whether they’re engaged — doesn’t develop practice. The responsiveness isn’t there. The trainer can’t see what the provider is struggling with. The provider can’t get immediate feedback on their real questions. There is no serve-and-return.

We learned that passive exposure doesn’t build language in children. Why do we still believe passive exposure builds competency in adults?

This is the knowing-doing gap I wrote about earlier — the distance between what people learn and what they actually implement. But the language development research gives us something sharper: it’s not just a gap in motivation or willpower. It’s a structural design flaw. We designed the system around input when the science says responsiveness is what matters.

The Incentive Problem Underneath

Why the silence? Because the professional development ecosystem has a market failure built into it.

The organizations delivering training get funded based on participation — seats filled, hours delivered. Not on whether practice actually changed. A provider who logs into a Zoom training and turns their camera off for two hours counts exactly the same as one who is deeply engaged. And the training organization has no incentive to tighten that up, because stricter standards means fewer completions means less to report to funders.

There are people gaming this system — doing the minimum, collecting hours without meaningful engagement. And here is where it gets morally complicated, because those people have reasons. They’ll tell you: “The pay is too low. I need to keep my business running. I can’t afford to close for a full day of training. This is what I have to do to survive.” And those pressures are real. Nobody in this field is getting rich. The economic squeeze on providers — especially home-based providers — is genuine and relentless.

But real pressure doesn’t make the response right. You can understand why someone cuts corners and still say: that’s not okay. The children in your care deserve a provider who is actually present — in the training and in the work. When someone logs in, turns the camera off, and collects their hours, they are taking something they didn’t earn. The fact that the system makes it easy doesn’t make it ethical. The fact that the pay is low doesn’t make it justified.

This is hard to say in our field, because we rightly want to protect and uplift providers. But protecting people doesn’t mean accepting every rationalization. If we can’t hold both truths — that providers are underpaid AND that gaming the system is wrong — then we aren’t being honest advocates. We’re just being comfortable ones.

And the people who could redesign the system to close these gaps won’t, because the current design serves their institutional interests. Everyone is protecting their own bowl of rice.

The Expectation Problem

There’s a study in education that’s been replicated dozens of times. In 1968, Robert Rosenthal and Lenore Jacobson told elementary school teachers that certain students had been identified as “intellectual bloomers” — kids on the verge of a growth spurt. The students were actually chosen at random. But by the end of the year, those students showed significantly greater gains in IQ. Not because they were smarter. Because the teachers treated them differently. Higher expectations changed behavior — not the students’ behavior, but the teachers’.

This is the Pygmalion Effect. The striking finding wasn’t just that expectations helped — it was that ordinary students with high expectations outperformed talented students with no expectations. Two classrooms. Same school. The difference wasn’t ability. It was what someone believed was possible.

I think this explains something about the ECE workforce that nobody wants to say out loud.

The professional development system asks for hours. It asks for attendance. It asks for completion. It does not ask for transformation. It does not ask providers to demonstrate that their practice changed, that their interactions with children deepened, that they can do something today they couldn’t do before the training. The bar is: show up. And people rise to exactly the bar that’s set for them.

This isn’t a character flaw. It’s a systems outcome. When you design a PD system around the minimum — when the expectation is compliance, not growth — you get compliance. When you measure hours instead of competency, people optimize for hours. The workforce is doing exactly what the system asked them to do. And then we act surprised when quality doesn’t improve.

This landed for me recently in a conversation with my supervisor about hiring. We were reflecting on a previous hire and what we want in the next person coming into a customer service role. We talked about the reality that sometimes the work is slow — there are fewer things to do, and you might need to find things for someone to occupy their time. But what we both kept coming back to wasn’t the task list. It was whether this person would find meaning in the work. Whether they would see a slow day as a chance to improve something, learn something, organize something — not because someone told them to, but because they care about doing the work well.

We want to pay fairly. Of course. But we also want the person to feel that this is a great place to work — and we want them to bring something that a paycheck alone can’t produce. Intrinsic motivation. The desire to do more and do better, not for the monetary reward, but because the work matters to them. I’ve seen what that looks like — a provider in her eighties who still shows up, not for herself, but for the community. That kind of commitment can’t be mandated or purchased. But the system does nothing to nurture it.

And then we said something that stuck with me: the expectation we set as employers shapes what we get. If we set the bar low — if we communicate through our systems and our culture that the minimum is acceptable — that’s what we’ll receive. Even from people who are capable of much more. The Pygmalion Effect doesn’t just apply to classrooms. It applies to every workplace. It applies to every field.

Now scale that to early childhood education. The system has set its expectations low for an entire workforce. Not intentionally — but structurally. And the consequences are showing up everywhere.

Pay is going up in some places — and that’s necessary and overdue. But pay alone doesn’t produce passion. A provider who was going through the motions at $18 an hour will go through the motions at $25 an hour if nothing else changes. The balance between expectation and reward is off. The reward is increasing, but the expectation hasn’t moved. The transactional relationship with the work stays transactional.

I’ve heard this frustration from directors. At a recent AECEA meeting, several directors described feeling stuck with staff who underperform — teachers who show up but don’t engage, who have the credentials on paper but not the practice in the room. “You can’t teach an old dog new tricks,” one director said. And this wasn’t one frustrated person venting. It was a pattern. Multiple directors, multiple programs, the same story.

I understand the frustration. But I also hear the system in that complaint. Those teachers came through the same PD pipeline that asked for hours instead of growth. They were shaped by low expectations — from the credentialing system, from the training culture, and sometimes from the programs that hired them without articulating what excellence looks like. The directors are inheriting a gap that was created upstream — and blaming the individual instead of the structure that formed them.

The full loop looks like this: the system undervalues care → so it builds a credentialing proxy → the proxy sets low expectations → providers rise to that low bar → some rationalize gaming it → intrinsic motivation gets neither cultivated nor rewarded → pay goes up but performance doesn’t follow → directors get frustrated → and everyone blames the individual worker instead of the system that shaped them.

The Pygmalion Effect tells us that expectations shape outcomes. Ordinary people with high expectations outperform talented people with no expectations. What happens when an entire field sets its expectations at “complete the hours”?

The Deeper Question: Why Credentials at All?

Here is where I want to push further than most people in this field are willing to go.

Why did we build this entire credentialing apparatus in the first place?

Care work produces something real and essential — safety, attachment, development, belonging. But the field has never been able to make that value legible in economic or political terms. A child who is securely attached doesn’t show up on a balance sheet. A toddler who learned to regulate their emotions in a family child care living room doesn’t generate a metric anyone reports to a funder.

So the field reached for what is countable: units, degrees, permits, professional development hours. The implicit bargain became: “We can’t prove the work itself is valuable in terms you’ll accept, so we’ll prove the workers are credentialed in terms you understand.”

Credentials became the stand-in for value because the system couldn’t figure out how to price care directly.

And once that proxy gets locked in, the whole ecosystem orients around it. PD providers sell hours. Colleges sell units. Policy advocates argue for wage scales tied to educational attainment. And the actual care work — the invisible work behind what looks easy, the holding, the feeding, the patient repetition, the cultural grounding, the love — stays invisible underneath all of it.

The irony is painful. This strategy was meant to elevate the workforce. But it actually reinforces the original problem. It concedes the frame that the work by itself isn’t enough — that providers need external proof of worth. A plumber doesn’t need a bachelor’s degree for society to agree their work has value. The price signal does that. But because care work has been systematically undervalued — because it’s historically women’s work, disproportionately immigrant women’s work, women of color’s work — the field felt it had no choice but to build a credentialing system as a substitute for direct valuation.

What AI Is About to Expose

Now zoom out.

With the rise of AI, this same problem is about to hit the broader workforce — hard. Lawyers, coders, analysts, marketers — they’ve all operated under the assumption that their degrees and expertise justify their compensation. When AI can produce a competent legal brief or a working codebase, those workers face the exact question early childhood educators have always faced: if something else can produce the output, what justifies paying you?

The proxy of credentials collapses for them too.

In a sense, care workers have been living in the future this whole time. They’ve always been in the position that everyone else is about to be in. I wrote earlier that AI won’t replace care — that its real potential is reducing structural friction, not substituting for human connection. I still believe that. But this essay goes further.

Here is the strange inversion that gives me pause: the thing AI cannot do is be physically present with a child. A two-year-old doesn’t need someone who knows about attachment theory. They need a warm body, a calm voice, a person who shows up at 7am and holds them when they cry. That’s not a proxy for value — that IS the value. And it’s the one kind of value AI genuinely cannot replicate.

The work that was always seen as the least valuable — bodily, relational care — might turn out to be the most irreplaceable. Meanwhile, the “high-value” knowledge work is the most vulnerable to displacement.

Power, Not Value, Is the Problem

So does this realization change anything?

Here is my honest worry. We live in a capitalistic system where money is the tool that organizes resource allocation. If AI makes most knowledge work reproducible at near-zero marginal cost, the value doesn’t disappear — it flows upward to whoever owns the systems. The coder doesn’t capture the value; the platform does. The teacher doesn’t capture the value; the curriculum company does. And care workers, who were already at the bottom of that value chain, get squeezed even further — not because their work matters less, but because they have the least structural power to demand a share.

This is not a new dynamic. It is the dynamic early care and education has always lived in. Families need care desperately. Providers deliver it. And yet the economic value gets captured everywhere else — by landlords, by credentialing systems, by administrative overhead — while the person doing the actual work stays poor.

AI doesn’t create this problem. It accelerates and generalizes it across the entire economy.

What I’m Still Sitting With

I don’t have a clean conclusion. I’m not sure the field is ready to hear that its credentialing infrastructure might be more about institutional self-preservation than quality improvement. I’m not sure the broader economy is ready to hear that care workers have been canaries in the coal mine all along.

But I do believe this: the question of how we value human care — real, physical, relational care — is no longer just an early childhood education question. It’s the question for an economy where machines can handle information but cannot hold a child.

If we couldn’t answer it for child care providers, I’m not confident we’ll answer it for anyone else either.

Unless we start the conversation differently.

I’d welcome the exchange.