In the Math Corps, we follow a Kids Teaching Kids — and, more importantly, Kids Caring About Kids — model. At its heart is the belief that every child has greatness within them, whatever that means for them: reaching their full potential, and getting access to the opportunities that can make that a reality.
The program has been running in Detroit for 30+ years (it was started at Wayne State by Steve Kahn and Len Boehm) and is now in its third year at Rutgers. It’s completely free to every camper.
This year, we have about 100 kids: 60 campers (rising 7th, 8th, and 9th graders), 30 TAs (high schoolers entering 10th, 11th, and 12th grades), and 12 college students (half instructors, or CIs, and half “assistants”).
Idea 1: Look at the NJ Department of Education’s report on New Brunswick High School: 85% of the student body is economically disadvantaged, and fewer than 10% of students are proficient in grade-level math. This is all happening across the street from an R1 research university. It’s just not right. We have to do something!
So what should we do? Idea 2: The so-called “Bloom two-sigma effect” says that: students who receive individualized instruction with mastery learning from an experienced private tutor perform, on average, two standard deviations above the mean. Pop quiz: what does that actually mean? If the average performance is at the 50th percentile, then two standard deviations above that is…?
(Answer: the 97.7th percentile!)
So here’s how we structure the camp: we don’t go quite down to 1:1, but we find that 2:1 instruction is enough. Still, hiring a tutor for every two students would be prohibitively expensive! Well, maybe our instructors don’t need expensive master’s degrees from schools of education. In fact, maybe they don’t need college degrees. In fact, maybe they don’t even need high school diplomas! What if the tutors were high school students themselves? (Selected, of course, first for their technical mastery of the subject, and also for being kind, empathetic people; then trained to align with the program’s goals.) This is the experiment the Math Corps has been running for 30+ years, and it’s been a resounding success: YES, high schoolers can be extremely effective tutors to middle schoolers, both in terms of instructional outcomes and cost! So each “Team” consists of 10 middle school campers, five high school TAs, and one CI (college instructor) overseeing the team’s operations. We also give the TAs coursework of their own, to build their mathematical skills even further. (They receive a modest stipend, in addition to courses taught by university professors.)
Idea 3: How should we respond to the “math wars”? Should students focus on mastering skills, or on critical thinking? Will we insist that kids know their times tables, or that they engage in productive struggle and inquiry? This is a false dichotomy; the Math Corps answer is: YES and YES! We make it explicit that these are two different skill sets, and that they need to be taught in different courses. So everyone — campers and TAs — get both a “Broccoli” class (raw skill acquisition, mandatory daily homework, and weekly timed quizzes, all aligned with the school curriculum) and an “Ice Cream” class (no homework, no quizzes, just a final presentation at our Closing Ceremony; the topics are college-level, delivered in an age-appropriate way).
For example, in Ice Cream, our rising 7th graders take a course in Topology. They learn about simple closed curves and duals of planar graphs, and work up to Euler characteristic. (This just involves drawing fun shapes on the blackboard and counting on your fingers.)
Idea 4 (new this year): We’ve implemented a number of ideas from the Science of Learning, developing (thanks to AI) a platform to track each individual camper’s progress. We first mapped out the entire Broccoli curriculum, then broke down each skillset (e.g, “Algebra 1”) into individual skills, organized into a dependency graph showing which skills are prerequisites for which others. Every day, we record the skills each camper has mastered, which lets us map their “Zone of Proximal Development” (ZPD): the skills whose prerequisites they’ve mastered, but which they haven’t yet mastered themselves. These are the skills we serve up in the next day’s instruction and homework.
Problem: we’re a completely tech-free camp: no phones, no tablets, no screens whatsoever. (I see what happens to kids on screens, and I will not have it on my watch.) So what are we going to do? Answer: Every night, once the mastery graphs have been updated based on what happened that day, we print ~250 pages of individualized worksheets and homework for the campers, and answer keys for the TAs. The tech is all in the back of the house, while the front is pen+paper, chalk+blackboard, and kids face to face.
The worksheets have three parts: 1. core work in each camper’s ZPD; 2. review skills in a spaced-repetition and retrieval-practice format—four days after mastery, and then again seven days later—so they become part of the camper’s long-term memory; and 3. a few “reach” skills that lie beyond the camper’s ZPD, but plant the seeds of what’s coming a few days later (so they can sleep on it, and won’t be seeing it for the first time when they actually get to work on it). Once a week, the AI also generates short, individualized 15-minute quizzes (pen and paper, of course!) drawn from skills the camper has already mastered. So the quizzes in general are very stress-free; they just check that mastery is real and being preserved. Students move through different skillsets at their own pace, and when they finish one, they automatically take the pretest on the next. We don’t do “tracking”: every student gets exactly the same treatment. That is, they each get something completely different, depending on precisely what they need that day to maximize their learning!
So. How’s it going so far? It’s day 8 (two weeks, four days a week), and here’s a plot of the different skill sets (“Real Number Line” → “Operations” → “Pre-Algebra” → “Algebra 1” → “Algebra 2” → “Precalculus”), with about 20 skills in each set. The blue bar above a given skill shows the number of campers, out of 60, who had already mastered that skill when they arrived (as determined by our pre-tests). The green bar above it is our value-added: how many kids mastered that skill since arriving. You can see that the total number of camper-skills added (the total green volume) is about 2,000. That’s an average of about 34 skills per camper in 8 days (really 6 days, since day 1 is ice breakers, and they haven’t done their work yet for day 8), or about 6 skills per camper per day. I expect this to taper off by the end of camp as they hit harder skills, settling at a final average of about 3 skills per camper per day. Over the four weeks—or 16 days—of camp, that comes to about 50 skills per camper, which is roughly two grade levels!
Now, none of this is surprising. We’re simply implementing well-known principles from the science of learning, along with the Bloom two-sigma effect! There’s no magic here. (The real magic is seeing what happens to the kids as they go through the program. This is really a values program masquerading as a math camp!)
For our first two years at Rutgers, we were supported entirely by my standard NSF research grant, which expires next month. (My new NSF grant has no budget for Math Corps.) So this, our third year, was the first time that we had to figure out how to raise funding on our own. We are grateful to the many sponsors and donors who made this summer possible!
But the situation is rather precarious. Our 2026 fundraising came largely from a combination of: generous Math+AI startups, about $10K remaining in the expiring NSF grant, and other friends I knew personally through different contexts. Our crowdfunding brought in a small additional contribution, for a total of about $79K raised. Our budget, on the other hand, comes to $93K, with the high school TAs as the largest ticket item. Our senior staff — our amazing Site Director, Program Coordinator, and Dean of Students (I don’t take a penny) — are another significant expense. Note that our total cost comes to about $1K per camper, which is extremely efficient for what the program provides. I was able to make up the shortfall this year with my Chair’s discretionary funds, but obviously none of this is sustainable. So our biggest challenge going forward is to find more reliable sources of funding.
That said, I’m very happy that we had many contributions in the $5K–$15K range, and not just one or two very large donors. Next year, we hope to add one more Team, for a total of 105 kids and a budget of $106K.
If you would like to donate to the Rutgers Math Corps program, or just learn more about it, please visit: http://rutgersmathcorps.org. Thank you for your support!