A Rocketship lands in Barnet: ARK Schools and 'blended learning'
I do think we’re at a time, as a country, where this stuff really fits pretty well with economic crises … when you’re in a situation when you’re gonna be laying people off anyway, how do you make that as good as possible, and one of the ways is to have more of the work happening somewhere other than a classroom, through technology
John Danner, co-founder of Rocketship Education, 2010
ARK Schools 2.0: from ‘no excuses’ to ‘personalised learning’
Two years ago, the DfE approved a proposal by ARK Schools to open four new schools. One of these was an ‘all-through blended learning school with an emphasis on technology’. The ARK Pioneer Academy will be built on an old football ground in Barnet. It is due to open in September 2018.
In ARK’s original application to the DfE, the new school was called the Blended Learning Academy. The job ad for the school’s principal defines blended learning as ‘a mix of traditional teaching and online learning, in an IT-rich environment’. Blended learning is already in use at ARK’s King Solomon Academy, in Marylebone. The academy, which serves an area with very high levels of child poverty, has been hailed as the ‘best non-selective school in England’ on the basis of its exam results. According to the school website:
From Year 4 and up, children receive a 1h 45 minute literacy lesson based on the Blended Learning model. This ‘blends’ children’s learning through teacher lead activities [sic] and software activities.
Like most aspects of ARK’s ‘school improvement model’, blended learning comes from the USA. It was developed and tested in charter schools – the publicly funded but privately controlled schools which were the model for academies. Charter schools – specifically, the KIPP chain and its imitators – were also the source of so-called ‘no excuses’ schooling, which involves strict control of students’ behaviour, and a relentless focus on improving test and exam results.
ARK Schools played a key role in introducing the 'no excuses' model into English state schools. As Paul Marshall, chair of the ARK Schools board, acknowledges: ‘We model ourselves on the American KIPP schools’. Now, however, the ‘no excuses’ charter school chains, including KIPP, are rapidly switching to a new, technology-driven model: blended learning. KIPP’s first blended learning school, the KIPP Empower Academy in Los Angeles, opened in 2010. And ARK Schools is following suit. The plan for the ARK Pioneer Academy ‘involves leveraging IT to afford each student a more personalised learning experience’.
‘Personalised’ is a key word in the marketing of blended learning. At the King Solomon Academy, for example, where all students are issued with Google Chromebooks in Year 5, the hope is ‘to have a device for every child at primary so that we can further drive innovation and personalised learning’. In planning to issue all primary school children with laptops, King Solomon is following the lead of US charter schools, which have extended computer-based ‘personalised’ instruction not only to kindergartners, but also – as we will see – to pre-kindergartners (i.e. four-year-olds).
Academisation and ‘ed tech’
ARK is a powerful organisation, with significant influence on the government’s agenda for schools. A key part of that agenda, since 2010, has been the promotion of education technology. As Tamasin Cave points out, the policy of mass academisation has made state-funded education in England ‘a magnet for tech interests’.
To give just a few examples:
- between 2010 and 2015, Rupert Murdoch’s News Corp was seeking to develop an education technology business, Amplify; the Leveson inquiry revealed that, during his time as education secretary, Michael Gove had a number of meetings with Murdoch and other News Corp executives
- American firms like K12 Inc. and Mosaica (now Pansophic Learning), whose business model is based on online instruction, have set up their own multi-academy trusts
- the Swedish company Kunskapsskolan, another pioneer of computer-based ‘personalised learning’, has also sponsored an academy chain
- David Ross, the co-founder of Carphone Warehouse, sponsors the David Ross Education Trust, which is currently pursuing ‘a highly ambitious eLearning strategy’ (Carphone Warehouse was involved in the government’s ‘Tablets for Schools’ initiative)
The push to open up English state schools to technology firms and tech-based edu-businesses – Murdoch famously described schools as ‘the last holdout from the digital revolution’ – has not slackened since David Cameron was replaced by Theresa May. September’s Green Paper, with its vision of a new generation of virtual grammar schools – online ‘centres of excellence’ created within academy chains – shows that there has been no change of direction.
The hedge fund managers and ‘venture philanthropists’ who run ARK share the enthusiasm of the wider investment community for education technology (for more detail, see here). Two of them, Ron Beller and Jennifer Moses, even have their own ‘ed tech investment fund’, Ed-Mentor LLC. The couple recently set up the Caliber Beta Academy, a blended learning school, in the impoverished Bay Area of San Francisco.
Beller and Moses were also the founders of the King Solomon Academy, the laboratory where the next ‘iteration’ of the KIPP-ARK model is being tested. In 2012, the school began a trial of blended learning, called Project 24 (‘Delivering personalised learning in a Year 7 maths classroom’). A number of staff travelled to the US to see blended learning schools in California and Louisiana. Bruno Reddy, the head of maths at King Solomon and the leader of Project 24, described the trip in his blog.
A highlight of the trip was a visit to Rocketship Si Se Puede Academy, one of the ‘reknowned [sic] Rocketship schools’. Rocketship Education is a ‘not-for-profit corporation’ which runs a number of K—5 charter schools (K—5: ages five to eleven) in California, Wisconsin, Tennessee and Washington D.C., ‘serving primarily low-income students’. The Rocketship chain is the main pioneer of blended learning. KIPP, and the other ‘next-gen’ charter chains which are switching to blended learning, are basically using versions of the Rocketship model.
The aim of this article is to describe that model. The main sources are: two reports from the Dell Foundation, one of the philanthropic foundations funding the growth of charter schools and blended learning (the reports can be found here and here); Richard Whitmire’s book On the Rocketship, an admiring account by a strong supporter of charter schools; a more critical report by the Economic Policy Institute, a US think tank; and some youtube clips of John Danner, the architect of Rocketship, explaining his ideas (they can be found here and here).
Rocketship: the ‘stripped-down efficiency model’
Danner, the brains behind Rocketship, is a Silicon Valley entrepreneur. He became a multi-millionaire in his early thirties when his internet advertising company was bought up by DoubleClick. He then moved into the charter school business. After a brief spell heading a KIPP school in Nashville, he set up Rocketship Education with Preston Smith, a fellow Teach for America trainee, in 2006.
The first school, Mateo Sheedy Elementary, opened in the Bay Area in 2007. The pupils were mainly Latino, the children of first-generation immigrants (in 2012, 85% of Rocketship students were entitled to free or reduced lunches, and 70% were English Language Learners). The children were packed into a rented Methodist church, where, as Richard Whitmire notes, ‘the “playground” was just a blacktop parking lot next to the kindergarten room’. But the school was successful in boosting the students’ scores in standardised tests – ‘closing the achievement gap’, to use the rhetoric of the US education reform movement.
Rocketship’s educational model has to be understood in relation to the test-based accountability system imposed on US schools by George W. Bush, more than a decade ago. Bush’s No Child Left Behind Act of 2001 mandated annual testing in reading and maths for all children from the third to the eighth grade, with a target of 100% ‘proficiency’ by 2014. Schools which failed to achieve Adequate Yearly Progress were threatened with closure or ‘charterization’.
No Child Left Behind, consolidated by Obama’s ‘Race to the Top’ initiative in 2009, gave rise to the delusory idea that schools and teachers alone could close the ‘achievement gap’ – the disparities in test performance – between ‘low-income children’ and their more privileged peers. To use Roger Titcombe’s phrase, it led to the ‘test-driven marketization’ of the US public schools system, with scores in standardised tests defining the ‘market position’ of a school. The KIPP model, so closely imitated by other charter school chains – and by ARK Schools – is a pure product of No Child Left Behind.
John Danner’s plan for Rocketship was to ‘make something that’s KIPP-like in terms of results’. The chain copied many features of the KIPP model: an extended school day and year, a narrowing of the curriculum in order to focus on literacy and maths, and a ‘data-driven’ approach geared to driving up students’ test scores. In 2012, the young ‘Rocketeers’ had an 8.00 a.m. to 4.00 p.m. day. This was divided into two 100-minute blocks of literacy instruction, one 100-minute block of maths, and a further 100 minutes in the ‘learning lab’. There was no provision for art, music or PE. The current curriculum seems to be slightly broader, including science, social studies and art as well as maths and reading. But science is ‘embedded’ in maths, and social studies and art are ‘embedded’ in literacy instruction. 50 per cent of teachers’ pay is tied to their students’ scores in maths and reading tests.
On the basis of its initial success in ‘accelerating’ the test performance of underprivileged students, Rocketship grew rapidly. By 2012, the company had seven schools in the Bay Area. Expansion was fuelled by a $5 million grant from Reed Hastings, the founder and CEO of Netflix, and the Charter School Growth Fund, a ‘non-profit venture capital firm’. Danner’s growth plan was very ambitious: a million students by 2030. And he aimed to achieve ‘scalability’ – to use the Silicon Valley jargon – without regular injections of philanthropic cash. The plan was to generate sufficient savings from the running of each school to make the growth of the chain self-propelling. Given that state funding of education in California is the fourth-lowest in the USA, this called for a highly cost-efficient operation – what Rocketship’s former chief schools officer describes as the ‘stripped-down efficiency model’.
On average, US elementary schools have six- to eight-acre sites. As Whitmire notes, Rocketship schools ‘got built on tiny parcels of land, using inexpensive modular construction’. At one point, Danner challenged his contractors to come up with a prefabricated school building that would occupy no more than a quarter-acre. But the real savings had to be made elsewhere – in the wage costs that typically make up 75 to 80 per cent of a school’s expenditure. The Rocketship model had to combine continuously-improving test results with a drastic reduction in staffing costs.
‘Kids right out of college’
To staff its schools, Rocketship relies on Teach for America recruits. These are recent graduates who have committed themselves to two years of teaching; they form a significant part of the workforce of charter schools. Their training consists of a five-week summer camp. Danner and Preston Smith, who had both come to teaching via TFA, regarded the organisation as ‘an incredible recruiting and selection company’. In 2012, 75 per cent of Rocketship teachers were either TFA trainees or recent ‘alumni’.
This means that a high staff turnover is built into the model, since few TFA teachers stay in the classroom for more than two years. According to Gordon Lafer’s report for the Economic Policy Institute, the average rate of turnover at Rocketship schools in 2012-13 was 29%. Rocketship’s Si Se Puede Academy lost 37% of its teachers in the year that Bruno Reddy and other ARK staff visited.
Very unusually for elementary school teachers, Rocketship staff are ‘subject specialists’. They teach either maths or literacy: two blocks of literacy instruction per day, or four blocks of maths (by 2013, instruction in both subjects had been totally standardised). This is one way of managing a young, inexperienced, and high-turnover workforce. But, as a Rocketship teacher points out, ‘building relationships with 60 or 120 elementary students and their families, as well as maintaining classroom culture throughout the day, is difficult’.
Teachers are given ‘rigorous support’ by managers known as ‘academic deans’. These are ‘master teachers’ – in other words, people with two or three years’ experience of the Rocketship model – who work with new staff to ‘improve their instructional practice’. Coaching can take place in real time, with an academic dean instructing a new teacher from the back of the class via a wireless headset (for a first-hand account of this type of ‘professional development’, which is surprisingly common in charter schools, see here).
Young, inexperienced trainees – ‘a lot of kids right out of college’, in Danner’s words – are not likely to question the model imposed by Rocketship’s management. The result is that teaching is degraded into training for standardised tests. A Rocketship teacher explains to Richard Whitmire how she cultivates ‘critical thinking’ in her pupils:
In order to connect testing to students’ lives, we think about critical thinking within test taking: this question is trying to trick me in this way….This answer is wrong because of this.
A journalist’s description of her visit to Rocketship’s Fuerza Community Prep, in 2015, shows how far Danner’s and Smith’s start-up schools ‘disrupt’ the 20th century model of primary education:
its brightly-lit classrooms [are] almost entirely devoid of the low-tech educational toys of other elementary schools. On a recent visit, there were no pretend kitchens, boxes of wooden blocks, or easels to be seen […] Students were often spoken to using language more common in corporate offices than elementary schools. A kindergartner whose uniform pants were falling down was told to ‘dress for success’, and administrator boasted that a first-grade teacher ‘was maniacal about not wasting time’ with her young charges
Kids right out of college can also identify with ‘KIPP-like’ systems of behaviour control which – to an older generation of teachers, at least – seem close to child abuse. Rocketship’s ‘Zone Zero’ policy requires total silence in the learning labs (see below), during transitions, and even during parts of the lunch break. Whitmire quotes a Rocketship teacher’s description of how a ‘difficult’ kindergartner is managed: ‘We restrain him for a bit, and offer him the “right choice” of sitting in a chair’. This is called a ‘behaviour plan’.
But none of this – schools without adequate space or facilities, teachers without experience or proper training, the relentless focus on test scores, the demand for ‘100% compliance’ from both pupils and staff – was particularly new or innovative, even in 2007. This culture is common to ‘no excuses’ charter schools. In these chain schools, as in English ‘system leader trusts’ like ARK or Harris, educational aims and principles are replaced by the single goal of ‘improving student achievement’ – in other words, higher test scores. To quote the Director of Pedagogy at another ‘high-performing’ charter chain, Success Academy, the aim is to turn children into ‘little test-taking machines’.
What was new about the Rocketship model was something else: the use of information technology.
In the learning lab
‘this is an industry where technology and the use of technology has been possibly the worst of any industry in the world’
All Rocketeers – including kindergartners and, at some schools, pre-kindergartners – spend 100 minutes each day in the learning lab. The children ‘rotate’ from their classrooms into something like a miniature call centre. There they work on computers, individually and in silence, using programs supplied by commercial vendors – principally DreamBox and ST Math, for maths, and i-Ready for reading. According to John Danner, this was ‘to allow kids to practice and do things in an individualised manner, without the need for a teacher’.
The Dell Foundation reports give a detailed picture of the learning labs. Each one accommodates around 100 pupils. In order to maintain a ‘productive learning culture’, the children sit in separate ‘carrels’, with ‘cardboard or wood dividers on either side to prevent interactions’. They are expected to maintain ‘college library quiet’, using ‘different coloured popsicle sticks’ to signal silently to adult supervisors.
A lot of time is spent training the children to ‘rotate’ from the classroom to the learning lab and back again. As one of the Dell reports notes: ‘students were expected to transition in and out of the lab quietly and begin work immediately on the online software programs, without interacting with other students’. A clip of Rocketship staff managing the ‘transition’ can be seen here; the children are told to keep ‘eyes forward’ and to ‘bubble’ their mouths to enforce silence (a common technique in ‘no excuses’ charter schools).
When Danner and Smith began to look for ‘online learning programs’ to use in the learning lab, they ran into a problem: most educational software is still designed on the assumption that a teacher is present to help students in difficulties. This was no use to Rocketship. As Danner told Whitmire:
We were looking for a very specific thing, a piece of software that would work for fifty or one hundred kids in a computer lab, so that when they get confused they didn’t have to raise their hands and ask for help from an adult, because with that many kids having an adult remediate all the problems just didn’t work
The solution, at least as far as maths was concerned, seemed to be a firm called DreamBox Learning. Their software is ‘adaptive’ – another key word in the marketing of computer-based instruction. The company’s chief programmer explains:
We have deeply integrated assessment into learning. We’re constantly asking, ‘Does the kid really know the stuff we’re teaching?’ If the machine doubts that, it loops back and gives a refresher…
To quote from the DreamBox website: ‘the product’s patent-pending “engine” constantly assesses each student’s mathematical understanding’. In other words, students are continuously tested. Every task ends in a test – usually a series of multiple choice questions – which must be completed before the student can move on to the next task (and test). If the student passes, she is automatically presented with a more difficult task; if she fails, she re-takes the original test after automated ‘hints and encouragement’ – or she is given an easier task. The target market is five- to eight-year-olds.
Soon after Danner and Smith opted to go with DreamBox, Reed Hastings – CEO of Netflix, and Danner’s mentor – bought the company, in partnership with the ‘non-profit’ Charter Schools Growth Fund. Hastings apparently saw no conflict of interest in simultaneously sitting on the boards of both DreamBox, a for-profit company, and Rocketship, a ‘not-for-profit corporation’ providing a testbed for DreamBox’s product development.
A stack of printouts and a script: the Individualised Learning Specialists
While in the learning lab, the pupils are supervised by Individualised Learning Specialists. These are hourly-paid ‘paraprofessionals’ with no teaching qualifications (Danner: ‘a high school degree but not a lot more than that’). Their job is to monitor the children, and to ‘assist struggling students on computer programs’. They are also responsible for tutoring students, individually or in small groups. In their role as tutors, the Individualised Learning Specialists literally work from a script – a ‘scripted tutoring curriculum’ prepared by Rocketship’s central office. As Danner explains:
you’ve got $10 an hour, $15 an hour people working there, who are overjoyed to have the job […] they’re not offended by the idea that they’ve got a script – and they’re a heck of a lot less expensive than a teacher
Another part of an Individualised Learning Specialist’s job description is to ‘interpret and manage online student data generated by multiple educational software platforms’. Computer-based instruction produces a lot of data. As we have seen, ‘adaptive’ software is designed to test students more or less continuously (this is what the promoters of blended learning – and the companies selling the software – call ‘formative assessment’). The test results are relayed, in real time, to ‘data dashboards’ on the adults’ computers.
As Bruno Reddy notes, the stream of data can be ‘overwhelming’, and the ‘coaches’ in the learning lab may themselves need ‘coaching and support for data interpretation’. At Rocketship, the Individualised Learning Specialists are overseen by assistant principals, who are also monitoring ‘student productivity data’ on their own computers, and giving instructions via headsets.
Ideally, the data from the children’s computers allows underperformers to be rapidly ‘targeted for intervention’, as a Washington Post journalist observed:
Ebony-Princess Cutts thumbed through a stack of printouts as first graders clicked away at their computer stations. Cutts […] earns about $14 an hour with benefits. “This tells me that he’s struggling,” she said, referring to a chubby boy sitting at the end of a row of computers, his small ears swallowed by big blue headphones. “I wouldn’t ordinarily notice because he’s quiet and he looks like he’s engaged.”
In the same year, Reddy and the other ARK staff found the work of the Individualised Learning Specialists at Rocketship’s Si Se Puede Academy to be ‘very impressive’.
‘You basically save 25% of your staffing costs’
It was this ‘hybrid school model’ that allowed Danner and Preston Smith to achieve the savings which were essential to their growth plan. As Danner told an audience of potential donors in Denver in 2010: ‘You basically save 25% of your staffing costs’. A typical US elementary school has around 600 pupils, and 21 teachers. The learning labs and the ‘rotation model’ allowed Rocketship to cut this number to 16, while fudging the issue of increased student-teacher ratios. As even Richard Whitmire, a big fan, admits: ‘class sizes were “small” only because so many students were rotating out of the classroom and into the learning lab’. He notes that one school, Rocketship Mosaic Elementary, was serving 630 students with only 16 teachers, plus the hourly-paid assistants.
The money saved – around $500,000 per school per year – is used to drive expansion. As Danner puts it: ‘We can buy land, build buildings, and start schools without raising additional philanthropy’. In a nutshell: ‘The fewer teachers per school, the more schools you can start’.
And Danner is bullish about the future possibilities of online instruction. In 2012, he told the Washington Post that, as the software improved, students could spend as much as 50 per cent of the school day engaged in individual work on computers. He had the same message for the donors in Denver:
This is the real pie-in-the-sky – five, ten years from now. You can imagine moving to a world where you’re splitting your day between basic skills and thinking skills, and you’re spending half your time online and half your time classroom – four hours of each, for us. This gets very compelling … You save about a million dollars a year, you only need 10 teachers
The ‘flex model’
By 2012, Rocketship were ready to expand the Bay Area chain to twenty schools – and also to open their first out-of-state school, in Milwaukee. Then they hit a bump. The chief financial officer had been looking at the figures, and he had bad news for Danner: ‘The formula we used to finance schools one through eight won’t work for schools eight through twenty’. A significantly higher return on investment was needed.
So, just a few months into the school year, Danner initiated a ‘model change’. The learning labs were scrapped. Rotation from classroom to lab was replaced by the ‘flex model’. This involved knocking down walls, and putting 100 children into a single room full of computers – a ‘flexible space’ – where they were supervised by only two qualified teachers. According to Rocketship board documents, the new student-to-teacher ratio would generate an additional $230,000 of net income from each school.
But the experiment was not a success. Rocketship’s own report on the ‘model change’ notes that ‘relationships and individual student accountability’ suffered. To make the flex model work, it would be necessary to hire more ‘hourly “flex” staff’ – when the point of the exercise had been to cut costs even further. Rocketship reverted to the ‘rotation model’ and reinstated the learning labs, while introducing computers – in the form of Google Chromebooks – into every classroom. The company’s growth plan was scaled back.
Early in 2013, Danner left Rocketship. He is now CEO of a for-profit startup called Zeal Learning. In 2014, the company launched ‘an adaptive Common Core practice app with game-like features’. The first charter chain to partner with Zeal was Rocketship Education. Oddly, the Zeal website no longer refers to the app; it offers ‘live, on-demand’ virtual tutoring.
The ‘flex model’ has been taken over and refined by another California-based charter chain, Summit Public Schools. Summit, founded in 2003, is the new market leader in blended learning. The chain has ‘partnered’ with Facebook, and will surely benefit from the multi-billion dollar Chan Zuckerberg Initiative, one focus of which is ‘personalised learning’.
Students at Summit schools have sixteen hours per week of ‘personalised learning time’ – in other words, totally standardised, computer-based instruction. Rocketship’s hourly-paid auxiliaries have been replaced by ‘peer-to-peer coaching’. Summit Denali, the chain’s flagship blended learning school, has no science labs, no library, no art or music rooms, no workshops, and no sports facilities. But every student has a laptop.
Improving cost efficiency: ARK gets on the Rocketship
The question is, how efficient should you be when you’re dealing with little human beings?
Kate Mehr, former chief schools officer at Rocketship
The ARK Pioneer Academy will use a version of Rocketship’s ‘classroom rotation model’ for English, maths and science. The application to the DfE suggests that ARK are sticking closely to the blueprint: students will learn from ‘online content / software’, which ‘frequently adapts to [their] level’, while receiving ‘instruction and input from the teaching assistant’. Apparently, this will help them to become ‘independent thinkers who take ownership for [sic] their learning’.
The job ad for the Pioneer Academy’s principal notes that the rotation model is ‘one of four models: Rotation, Flex, A La Carte, and Enriched Virtual’. The rotation model itself includes ‘four sub-models: Station Rotation, Lab Rotation, Flipped Classroom, and Individual Rotation’. All this stuff is literally cut and pasted from a paper put out in 2013 by a San Francisco-based think tank, the Clayton Christensen Institute for Disruptive Innovation. The authors of the paper were Michael Horn, a Harvard MBA who co-founded the Institute, and Heather Staker, a former management consultant. Neither have ever been teachers, or have any background in education.
Clayton Christensen himself is a professor of business administration at Harvard, and ‘the World’s Most Influential Business Management Thinker’. The author of Disrupting Class: How Disruptive Innovation Will Change the Way the World Learns, he has emerged as the guru of blended learning, which is heavily promoted by his think tank.
Christensen is upfront about the ‘disruptive’ potential of blended learning. He believes, for example, that it will play a big part in ‘eradicating rules that restrict class size and student-teacher ratios’. ARK told the TES, rather more coyly, that blended learning offers ‘an opportunity for revised teacher roles’. (The Daily Mail was less cautious; its story about the Pioneer Academy was gleefully headlined ‘Computers Replace Teachers’.)
Christensen argues that ‘computer-based learning on a large scale is less expensive than the current labour-intensive system, and could solve the financial dilemmas facing public schools’. Similarly, ARK hopes that blended learning will ‘improve cost efficiency through both staffing and school design efficiencies’. It will also, of course, open a huge new market for tech companies, and for investment firms like Ron Beller’s and Jennifer Moses’s Ed-Mentor.
For Bruno Reddy, one key ‘takeaway’ from his visit to a blended learning school in New Orleans – run, incidentally, by Jay Altman, who was once ARK’s Director of Education – was that the new model needs to be marketed carefully: ‘All communication to stakeholders should be about personalisation, not the technology’. Reddy, a Teach First trainee and ‘Google certified teacher’, has now left the classroom to start his own consultancy business.