A few months ago, I looked at the quite laudable CZI/Gates “Moonshot” effort.

There I made a technical point about their somewhat randomly chosen goal: moving students from the fiftieth to the ninety-eighth percentile on math over the course of ten years. It was too big.

Yes, I get it: moonshot. Audacious! But if the goal in 1969 was, say, men landing on Jupiter, I don’t think we’d have made much progress. Even moonshot goals benefit from plausibility. There’s a just right “Goldilocks” quality to them.

Running a 3:40 mile would not be a moonshot. It’s too close to the current record. Running a three-minute mile is probably not biologically possible. That’s landing a man on Jupiter in 1969. Somewhere between those numbers is the right goal.

Today I look at another ed reform moonshot. This comes from the good people at the Thomas B. Fordham Institute and the Center for American Progress. Again, laudable. Even better: Bipartisan! I’m a big fan.

This effort is backed by Schmidt Futures, a fascinating philanthropy (in my view) because their top belief is betting early on exceptional people. Smart. Sounds like a venture capital approach, and a fun departure from the typical plan, which is to pick a favorite strategy and then fumble around in the seat cushions to find an able leader. In terms of their education goals, Schmidt Futures, like CZI/Gates, hopes for digital math tutors. Me too. But it’s an area where my insider friends (who actually work on these things) tell me we’re a long way away from stuff that really works to help struggling kids.

Okay, that’s the background. Now onto this moonshot:

We’ve now launched the centerpiece of the project: a competition to unearth examples of such bold, new, evidence-based ideas.

Hmm. Does “evidence-based” cut against “bold, new”?

I think I’ve initiated more randomized trials in different arenas that just about any ed reformer (teacher coaching, college counseling, teacher-parent communication, high-dosage tutoring, teacher prep, charter schools, etc., with a few successes and lots of failures)—so let me be clear that I’m very pro evidence! Pro, pro evidence.

However, I’m less sure that we want evidence-based in this context. If we’re talking about empirically strong evidence, then it’s something that won’t be bold and new. If we’re talking about b.s./weak/typical correlation evidence that doubles as marketing, then what’s the point of asking for it?

We are seeking the equivalent of education’s self-driving car, i.e., an idea that will revolutionize schooling.

The examples we seek should not be policy proposals. Instead, we aim to collect applied research and development ideas that will unleash innovation, support educators—and dramatically improve student outcomes.

What is a technological or developmental effort that will spark game-changing differences in student learning?

I’m trying to parse this out.

The self-driving car is close to happening in real life. Next few years. See here for those timelines. That moonshot started a while ago.

Moonshots take a while from inception, and that affects the types of ideas you can bring to the table. What’s the timeline for this one?

We are seeking ideas that would help the U.S. achieve one of the following big goals (your choice) with the help of a public or private investment up to $1 billion.

Most of the current ed tech marketplace is software, apps. Most of the stretch efforts simply replace adults with robots—whether parents, teachers, or tutors.

Cut in half the number of fourth graders reading “below basic.”

Existing phonics apps are already somewhat useful for this group of strugglers. Eventually there will be a cute robot who reads Curious George to/with a four-year-old instead of a parent. Whether that’s a good thing or not, we shall see.

Mischievously, a “friendly” EMP device that turns off televisions and game consoles so kids have less screen time might help increase reading scores.

Double the number of eighth graders who can write an effective persuasive essay.

Double the amount of high-quality feedback the average middle schooler receives on their academic work.

Hmm. One doubles an input. The other doubles an output, which is way harder.

There is existing essay-grading software, which is promising, and interesting grammar products like Quill. (Paging Peter Gault: You should apply for this moonshot.)

Eventually this will be a robot teacher who does writing conferences while making cringy dad jokes. A typical eighth grader now gets perhaps five “longer pieces of feedback” per school year. By that I mean an essay that comes back with several teacher sentences combined with red ink corrections, and perhaps a couple brief writing conferences. The dosage of feedback is currently way, way off. It needs to be more than doubled—more like increased twenty-fold.

Shrink by 30 percent the average time a student spends in English-language-learner status.

Ha. I know that policy solutions are forbidden here, or I would say, “Don’t incentivize schools to label and keep kids in English language learner (ELL) status.”

And I know that political solutions are also forbidden here, but when we created a new school for ELLs in Boston, we found this weird dynamic: Many of the best veteran ELL teachers told us that many (not all!) of the same strategies that work for struggling “native born” kids were the right ones for ELLs. But there was a political context which made this hard to say at times.

If the true quest here is at least partially helping new immigrants speak English, then one underleveraged strategy is helping them make friends whose native language is English. Otherwise, the new kid naturally seeks out others like himself, and it slows down progress in conversational English. Eharmony-for-friends-type website at the school level?

Ensure that every student receives high-quality college and career advising by ninth grade.

“Every student” implies to me “one product.” Here, however, we need to break the challenge into component parts.

One, a small number of academically strong kids “under-match” and could go to a more competitive college (thereby displacing other kids, though).

Two, the return-on-investment question of college for mediocre high school students seems unresolved to me. I have a number of ed reform friends sort of yelling, “If you want college for your own kid, how dare you suggest anything but that for other kids.” I’m guilty of saying such things years ago. Yet evidence on precisely which students benefit from going to noncompetitive colleges is murky.

And three, finally, career guidance for non-college-goers seems utterly broken. It’s not ready for a tech replacement because the human version is so messed up. Before this one can be digitized, it needs a reboot in terms of “What are the humans telling the kids if kids are lucky enough to have a one-on-one conversation with any sort of counselor?”

Double the number of students from low-income families and students of color who graduate from high school with remediation-free scores on the SAT, ACT, or similar exams.

That is tougher than most people think. Randomized controlled trials have shown that even well-regarded in-person tutoring has not actually bolstered SAT or ACT scores. My priors here lean me toward human tutoring with unusually strong management teams, combined with virtual delivery. A nice experiment is happening soon with Saga Tutoring and virtual SAT help. Alan Safran and Antonio Gutierrez, you should go pitch for this moonshot.

Two final thoughts.

First, I’d love to see more about what “developmental” moonshots could look like. I interpret that to mean non-technology “processes for humans”—combinations of “moves” for individual teachers in particular.

  • If the baseline “Joy Factor” score of a typical class is a three out of ten, how can a teacher double that to six out of ten without harming achievement?
  • What would it take to reduce a sixty-hour-a-week teacher’s effort to fifty hours without harming achievement or student ratings? This could go a long way to increase teacher retention.
  • What steps does it take to “flip” a disenchanted third grader into the single most helpful, thriving kid in the class?
  • Is it possible to have a brutally honest conversation, like doctors do, with a parent about an academically weak student, such that, when combined with an action plan, it improves the kid’s short, medium, or long-term outcomes?

By better describing which “developmental processes” would be eligible for the moonshot, and towards which types of goals, you’d potentially gain non-ed tech entrants—particularly if those processes were meant for individual teachers. Many are hungry to add to their toolboxes.

Second, I’d cut the evidence-based requirement entirely. Just make it “a competition to unearth examples of bold, new ideas.”

Michael Goldstein is co-founder at Oxford Street Education and a visiting scholar at the Harvard University Graduate School of Education.