More on Alpha School

A profile of a key technologist

Jeremy Stern writes,

The one thing [Joe] Liemandt will talk about for hours on end is Alpha School: the teacherless, homeworkless, K-12 private school in Austin, Texas, where students have been testing in the top 0.1% nationally by self-directing coursework with AI tutoring apps for two hours a day.

Note: Alpha school is not really teacherless. Teachers Students are helped by highly-paid adults, even if they are not called teachers.

Most of the first half of the long article is about MacKenzie Price, the founder of Alpha School. She met Liemandt working for his company, Trilogy Software.

Liemandt’s story has classic elements of founding a startup in the days before VC’s were just throwing money at them.

by 1990 credit cards just appeared in your mailbox in an envelope with the words, “You’re Pre-Approved!” Credit scoring wasn’t too sophisticated yet, so Liemandt and his team applied for stacks of them. One would come in with a $5,000 limit, they’d take $4,000 as a cash advance, then get another one to cover the monthly payment on the other. Each new credit card form asked how many other credit cards you already had and provided a little box with enough space for a single digit. Liemandt had about 50, which he pyramided to keep funding Trilogy

Liemandt understood that big companies do not buy from startups unless they are desperate. One advice I give to a small company selling to a big one is to charge an outrageously high price for the offering. I call this “standing up to the bear.”

“Great,” he told Watson. “The price has tripled. It’s $300,000.”

“Done,” she told him.

Three months later, HP called. “It’s $3.5 million,” Liemandt said, understanding they’d abandoned their own configurator pursuit. They accepted.

Here is an interesting connection:

One early project he and Price collaborated on was funding a randomized control trial by the Harvard economist Roland Fryer, who experimented with paying fifth-graders at bottom-ranked schools in Houston to complete their homework and get good grades, and parents to attend parent-teacher conferences. Fryer found that the payments affected effort but not results. “It’s like telling a random person you’ll pay them $10 million to win Wimbledon,” Price explained. “It’s hard to incentivize outcomes, but you can incentivize the work itself.”

A key point:

If the consensus of learning science could be summed up in a sentence, it would be that “a teacher in front of a classroom” is one of the least effective ways to teach kids anything.

Pointer from Tyler Cowen. Long as the article is, I enjoyed the whole thing.