Yesterday marked one year to the day since I enrolled in my first Math Academy course, and it’s a good time to reflect on my Math Academy experience thus far. (For past posts about my experiences, see my first Math Academy post, which contains links to all my other posts and updates.)

My progress thus far

Here’s where things stand after one year: First, I’ve completed the following courses:

I’m also 91% of the way through Calculus II. I had hoped to combine this post with a post celebrating my completion of Calculus II, but unfortunately I got sick in late November and ended up skipping several days of Math Academy work in December. I should finish up Calculus II later this month.

I’ve accumulated a total of over 19,500 XP, averaging about 53 XP per day over the course of the year. If you ascribe to the Math Academy rle of thumb equating 1 XP to 1 minute of focused work, then I’ve spent about 325 hours working on Math Academy lessons, reviews, and quizzes, the equivalent of two months of a full-time job.

Why I’m doing this

A few weeks ago someone on X or on Discord asked me what my ultimate goal was in doing Math Academy. My initial interest, as I’ve written, was prompted by my failure to study linear algebra to a point where I knew what eigenvalues and eigenvectors were. But that wasn’t my ultimate goal.

The reason I started trying to learn linear algebra (way back when) was as an adjunct to learning probability and statistics. And that in turn was motivated by my desire to get to a point where I could read social science papers and understand the techniques being used. I’ve always enjoyed writing blog posts about various topics for which academic literatures exist, and I’ve always been frustrated by having to skim over the sections in papers where they discuss things like study designs, statistics methodology, causal inference, and the like.

So my goal is to be able to get to a point where I can go through books like Bayesian Data Analysis or Statistical Rethinking, follow the math and the R code, and maybe even try out some of the techniques on things I’m interested in for which I can get data.1

A related goal is to improve my R and Python knowledge. I can do exploratory data analysis in R (including using ggplot2) and can do Python coding to a low intermediate level. I’d like to skill up to a more advanced level in both languages.2 I‘m really looking forward to Math Academy adding a computer science course, but as noted below I’m not expecting to see it any time soon.

Math Academy then and now

As noted in my very first Math Academy post, I started my Math Academy journey not by enrolling in the service but by reading Justin Skycak’s book-in-progress The Math Academy Way and taking copious notes. That reading convinced me that Math Academy had what looked to be some reasonably sound learning science behind the service, and was therefore worth trying out.

The current version of the The Math Academy Way is about 12% more in length than the version I read originally, and is still not finished: There are five chapters marked as “in progress,” and 12 pages of “notes for future additions.” I suspect it will be some time before we see a complete first edition of the book.

The same can be said of the Math Academy service itself. The service is still described as having “beta” status, with no clear indication as to when it might shed that label. As with Gmail (which was labeled as “beta” for 5 years even as Google onboarded tens of millions of users), it’s possible that Math Academy won’t reach 1.0 status for a while.

That’s really just a minor semantic issue. The Math Academy service has been reasonably stable for the year I’ve been using it. The only thing out of the ordinary I’ve noticed is that the system takes a relatively long time to respond (several seconds at least) when you complete a lesson or a review. This may indicate increased system load due to more people using the service. Alternatively, it may (or may also) reflect the fact that choosing the next action may be increasingly time-consuming for students like me who‘ve already traversed a significant portion of the overall Math Academy knowledge graph. (In particular, I can imagine that deciding what topics have been reviewed thus far, including implicit review, may take a while.)

To my recollection, the only major product feature that’s been added during the past year is the “gravity” feature, which allows students to designate topics that they want to focus their learning toward. Math Academy then prioritizes lessons on the paths in the knowledge graph leading to those topics. Gravity is a beta within a beta, as you have to request access to it. I haven’t used the feature myself, because my interest is in completing entire courses, but the feedback I’ve seen on X and Discord is generally positive.

Most of the effort of the Math Academy team seems to be going toward creating new courses and making the backend and UI changes that will be necessary for at least some of those courses. To the best of my knowledge, the only completely new course added in 2025 was Discrete Mathematics, a university-level course.3 However, there were multiple updates to existing courses, including the addition of many free response questions, for which the student needs to type in an answer instead of selecting from multiple choices.

Several other courses were mooted as potentially appearing in 2025, including Differential Equations, Machine Learning (proposed as a two-part course), Computer Science (also proposed as a two-part course), and Abstract Algebra in the university-level courses, and SAT Test Prep and ACT Test Prep in the high school courses. Of those, the only one that appears likely to be released in the near future is Differential Equations.

There were also significant updates and expansions planned for SAT Math Fundamentals and Mathematics for Machine Learning. To my knowledge the Mathematics for Machine Learning expansion has not yet been released; that will presumably be done in conjunction with release of the first Machine Learning course. As for SAT Math Fundamentals, per Justin Skycak on Discord, “We added about 115 new ‘missing middle’ topics that appear on the SAT but are not covered in any standard school curriculum. For scale, 115 topics is roughly half the size of a full-year course like Algebra 1.”

The Math Academy team remains relatively lean. The only named individuals known to be employees are Jason Roberts (UI and backend development), Justin Skycak (analytics and algorithms), Alex Smith (content development), and Sandy Roberts (technical support and general administration). There’s also a group of people (presumably contractors) creating lesson content; their number and names are unknown.4

As discussed on one of Jason Robert’s and Justin Skycak’s podcasts, they don’t follow a set schedule for development, but rather concentrate their development efforts on whatever part of the product seems most urgent to focus on at the time. As a result there is no fixed roadmap for future courses, and I think it’s fruitless to speculate on what might be released when.

Speaking of podcasts, there are now a whole series of Math Academy podcasts featuring Jason Roberts, Justin Skycak, and (occasionally) Alex Smith. Some of them have interesting insights about the origins of Math Academy and its internal workings, but they’re pretty long and often rambling; making transcripts of them and having an LLM summarize the transcripts might be the best way to approach them.

Beyond Math Academy

As a result of using Math Academy, and especially of reading The Math Academy Way, I picked up a few other interests in 2025.

The first is PhysicsGraph, an online education service inspired by and modeled on Math Academy. I wrote an earlier post about the PhysicsGraph quantum computing course; I plan to write another post about their Physics I course. The PhysicsGraph folks are being much more aggressive than Math Academy in introducing new features in the UI and elsewhere; whether Math Academy will be inspired to adopt some of them is an open question.

Since I was following folks from Math Academy, PhysicsGraph, and related ventures on social media, I also stumbled upon the early reviews and articles about Alpha School. As it happens, Alpha School uses Math Academy as part of its educational software stack, thus highlighting another aspect of the Math Academy business model, i.e., licensing the service to schools instead of just serving individual subscribers. I wrote about the first podcast by Joe Liemandt, the funder of Alpha School and related ventures, and plan to write more about Alpha School in future.5

Finally, all this got me interested in developments in learning science, and so I started following Carl Hendrick, an academic working in the field who’s had some involvement with Alpha School. He has a newsletter that I’m subscribed to, and is also co-author or co-editor of several books on teaching and educational psychology, at least one of which I plan to read this year.

That’s it for now, I hope to be back here with another update in a few weeks once I finish Calculus II.

  1. Before anyone interjects, yes, I could enlist the help of an LLM in doing such analyses. But even if I do that I’d still like to be able to evaluate the approaches used and determine whether the LLM’s output makes sense. ↩︎

  2. “But you could use an LLM . . .”—see the previous footnote. ↩︎

  3. Unfortunately, the Math Academy website doesn’t have a page for course announcements and related news, so I’m basing this statement on a comparison of the current set of courses to an archived snapshot of the courses page from a year ago. ↩︎

  4. The current version of The Math Academy Way does acknowledge the contributions of a Yurii Leschenko. I have not been able to confirm his connection with Math Academy; however, on LinkedIn there is a person of a similar name whose profile states that they are a “Math & E-Learning Expert” working as a remote employee of a California startup creating math educational content for K-12 and university courses. ↩︎

  5. I already name-checked Alpha School and its variants and imitators as a possible alternative private school ecosystem for a 21st century techno-nationalist elite. ↩︎