Weekend greetings, CharterFolk.
I hope you’re all having a great Labor Day.
My topic today is alphas and betas.
A confluence of developments has emerged in recent days on the theme.
The week began with Mamdani banning AI use in NYC through middle school.
The next day, Los Angeles announced an even more sweeping prohibition.
The general tech-lash in K-12 continues to capture even more attention.
A couple of weeks ago, Andy, Reed and Mike discussed AI’s likely impact in education in the next decade.
Andy and I debriefed the discussion early last week.
The point of their conversation where I thought the issue came into sharpest focus?
When they made their over-under wager on whether Alpha Schools would grow to serve at least 12,000 kids in the next three years.
Reed took the over.
Mike and Andy the under.
Alpha, as CharterFolk readers will know from prior posts I have written here, has captured a lot of attention.
Some of the most glowing came from a long piece written by a parent that was published on Scott Alexander’s much-read blog Astral Codex.
In it, the parent said he believed something profound was happening in terms of learning at the school, though the strengths he identified weren’t some breakthrough in AI instruction. They included intensive adult support, school culture and incentives.
Now the word goes out that Alpha is opening 27 campuses this year …
… supported by the $1 billion in backing from Joe Liemandt, the billionaire founder of Trilogy Software and ESW Capital.
As one would expect given its targeting of the high end of the private school market, Alpha has chosen to open in more affluent areas.
One is a place I know well, just down the block from Crossroads – a high-end private school famous for its attractive campus and progressive teaching methods in Santa Monica.
Tuition is going to be $65,000, roughly comparable to what Crossroads charges.
One of the other new schools is opening in Miami, which seems to make sense.
The rush of wealthy families to the area, and the lack of high-end private school options available there, make it a natural fit.
Where Alpha hasn’t yet proven itself much of a fit is in the charter school space.
This week, ProPublica, no friend to charter schools generally, wrote a highly critical piece.
It lamented that a proposed school – not one proposing to bring the whole Alpha model to CharterLand, but one under a different brand and operated by a different party that would contract to access Alpha’s online AI instructional platform in order to offer an entirely virtual program – was denied a charter by the Texas State Board of Education, but then was allowed to open as a school that would be funded under the state’s new ESA program.
For ProPublica, this proved how broken and unaccountable the whole ESA approach is.
It broadened its critique to cite several other states across the country – most of which are clearly pro-charter states – which have denied applications along the lines of what was proposed in Texas.
Only one state – Arizona – has approved a charter for a virtual school using Alpha’s AI instructional platform.
Now its results are out and are being pored over.
A thoughtful Substack post was released last month by Dan Meyer that’s worth checking out.
It doesn’t purport to provide the definitive final word, but it shows that results at this first virtual site are far different from what seems to be emerging at Alpha’s in-person schools serving highly affluent families.
Meanwhile, in recent weeks, AEI/CERN has questioned whether the entire charter school model may be antiquated because ESA- and voucher-supported private schools are now being allowed to grow and innovate with far greater freedom than charter schools currently have.
I would call it the anti-ProPublica piece.
ProPublica looks at the charter denials and says: exactly right.
Weak applications are being screened out by responsible public oversight only to reappear as publicly funded private schools under far looser rules.
AEI/CERN looks at the same facts and says: exactly wrong.
If innovative schools can get opened only through ESAs and vouchers, while charter authorizing keeps turning them away, that is evidence the charter model itself has become too restrictive.
Indeed, Alpha founder MacKenzie Price went to Instagram last month to cite as further evidence of public education’s brokenness the fact that Alpha proposed to open 10 free virtual charter schools across the country and was denied in every instance but one.
So what is all this?
Evidence that charter schools are responsibly regulated?
Or hopelessly over so?
Of Alpha and Beta
It calls to mind the naming of the new venture.
Alpha.
I’ve looked for a rationale as to why the organization chose that branding and haven’t been able to find anything.
First, as in the Greek alphabet?
Dominant, as in lead dog?
The internally tested version of software that precedes the outside-tested version?
None seem likely.
Making me speculate that perhaps the most likely meaning of Alpha the creators found compelling came from the world of finance.
Alpha is the unexpected positive return an investor receives relative to the amount of risk taken.
In the famous CAPM structure, an investment’s expected return is related to its beta – its sensitivity to movements in the broader market.
Take on more beta and, generally speaking, an investor should expect more return.
But that higher return isn’t alpha.
It’s simply the return an investor should have expected for accepting more risk.
Say that, given the amount of beta you took on, you should have expected a 15 percent return.
But you got 25 percent.
That extra 10 percentage points?
That’s alpha.
Not simply a good return.
A return beyond what the risk you took would have predicted.
Which brings us to the other meaning of beta.
The one Silicon Valley has made common in the vernacular.
Beta testing.
Putting something new and not yet fully proven into real-world conditions so that you can find out whether it actually works.
Suddenly a vocabulary is emerging that is resonant to the moment we’re in.
Because all of us in the ed reform world say we want alpha.
Innovation that greatly improves learning.
Breakthrough approaches.
But you don’t get alpha without taking on beta.
The challenge is figuring out how to wisely assess beta in our pursuit of alpha.
Up the income ladder, we assess the risk to be lower.
Somebody wants to spend $65,000 to enroll their child in an environment wherein at least parts of the model remain unproven, while all the trappings around that model – high adult interaction, high-achieving peers, strong school culture – are hallmarks of what the elite are already getting for their kids in other luxury-end private schools?
Seems pretty low risk.
The family has resources. The school has resources. There are lots of adults around. If something isn’t working, everyone has options.
But take one unproven sliver of that program and make it the totality of a student’s school experience in a virtual context, potentially serving very large numbers of students, many of them high-need?
That’s an entirely different beta.
And the authorizers turning down these Unbounded schools in red states don’t seem to me excessively inclined toward charter school opposition.
Indeed, the assessments I’m getting from various Folk closer to these denials suggest Alpha wasn’t putting as much into these applications as it appears to have put into the opening of its 27 new private schools, and it was that lack of commitment as much as anything else that resulted in their denials.
Authorizer efforts to test for commitment are fraught and open to interpretation.
One person can see a reasonable test to see that a school development team is all in.
Another can see needless friction making charter school development needlessly difficult.
But again, in the context of these particular authorizers who aren’t known to be in the category of those requiring thousand-page petitions, applications being withdrawn halfway through the process does not instill confidence.
That isn’t necessarily evidence of hostility to innovation.
It could be reasonably assessed beta.
Exactly what good authorizers are supposed to do.
And commitment really matters.
It’s a part of what distinguishes Alpha to begin with.
Look at the contrast against what happened with AltSchool.
As a proposed entrant into the high-end private school market, AltSchool had extraordinary buzz a few years back.
As well as a focus on AI – not to directly provide instruction to students, but to make it easier for teachers to customize instruction to individual student need.
A few years later, AltSchool had blown through its $175 million in venture capital and had shut down altogether.
Alpha’s got a billion dollars in backing from a funder who certainly projects being all-in.
Twenty-seven new sites are opening this year.
Compared to AltSchool, that’s enormously reassuring.
At the macro level, commitment hardly seems to be the problem.
But commitment in the macro isn’t necessarily commitment in the particular.
An authorizer doesn’t approve a billion-dollar enterprise.
It approves the school application sitting in front of it.
And if what Alpha brings to CharterLand is a virtual school operated by someone else using only one piece of the Alpha recipe – and the team then withdraws when the process gets difficult – an authorizer is perfectly entitled to ask:
How committed are you to this?
That distinction matters because long-term commitment is exactly what we should want from organizations attempting something this ambitious.
And it stands in sharp contrast to many of the gestures we see happening in education right now.
Most of them strike me as flails not meant to last.
AI bans?
Yeah.
See how stable that proves.
Confidence that ESAs and vouchers have the accountability and freedom trade-off well calibrated?
The proof will be in the pudding years from now.
Predictions of the charter school movement’s demise?
Again, wait that one out, CharterFolk.
What we need is a longer-term focus and the capacity to hold two ideas in our minds at once.
Yes, charter schools have become over-regulated generally, and we need it to be much easier for committed developers to get new charters approved.
And yes, some proposed new charter schools being denied is not, by itself, evidence that the movement has lost its way.
What we need is a fundamentally better way to assess beta.
And so much of the challenge comes down to the authorizers, and to the beta they are able and willing to take on.
One school fails.
One school deviates from acceptable governance or fiscally responsible practice.
The authorizer gets accused of being asleep at the switch.
What happens?
The mountains of new requirements and constraints.
Or even worse?
Authorizers who aren’t really authorizers.
Most school districts that oversee charter schools aren’t really authorizers but are protectors of their own domain.
Their decisions have nothing to do with assessing beta.
And they cloud up any discussion of how authorizers responsibly do it.
Ultimately eroding trust.
And in a trustless landscape, the default becomes distrust.
If not mistrust.
Getting the Charter School Recipe Right
Much of the challenge can be traced back to the charter school movement’s origins.
We were designed to bring both forms of beta into the system – an introduction of risk and a way of testing new approaches – such that we might uncover Alpha that education would never have found.
The proposition was higher levels of freedom in exchange for high levels of accountability.
It’s tempting to say we need to get that recipe back.
But maybe the more accurate way to put it is that we need to get it right for the first time.
Over the years, having entrusted so much authorizing responsibility to parties with conflicts that no one trusts, the proposition has eroded.
Not enough accountability.
Not enough freedom.
Which is compromising our prospects for generating Alpha.
It should be one of our biggest priorities right now.
Getting the authorizing excellence we need into the landscape.
It starts with greater accountability for academic performance.
That’s what gives us the confidence to take more beta on the front end – knowing we can course correct when one of those difficult judgments proves wrong.
As to Alpha itself in the shorthand?
I’m not that worried.
Andy’s, Mike’s and Reed’s bet was about how the organization will do in the private market.
That’s still free to play out.
It’s a bet I’m not that interested in actually.
Whether they can make the model resonate with large numbers of the elite matters much less to me than whether they can make it work with small numbers of those who need better education most.
To some extent you could say, let the beta testing happen with the wealthy and, if it works, bring it to the masses.
But on the other hand, thoughtful writers seem to think Alpha’s whole recipe right now – even apart from whatever breakthrough its AI instructional platform may ultimately represent – could be on to something.
Why not test that right now?
Not some sliver of it, but the full Alpha recipe.
One of the dozens of new Alphas that will open in 2027?
Make it a charter.
Some of the per-pupil funding levels in big cities across the country approach what Alpha’s charging just down the block from Crossroads.
Certainly enough money that the remaining shortfall can be covered by a small portion of that billion dollars in backing.
Not some sliver of the model.
The thing itself.
Then we’d finally be testing the beta that matters:
Are Alpha’s results a function of the program it provides?
Or simply a reflection of the students it serves?
Which would finally let us answer the question that matters most:
Is Alpha actually generating alpha?