69传媒

Imagine classrooms outfitted with cameras that run constantly, capturing each child鈥檚 every facial expression, fidget, and social interaction, every day, all year long.

Then imagine on the ceilings of those rooms infrared cameras, documenting the objects that every student touches throughout the day, and microphones, recording every word that each person utters.

Picture now the children themselves wearing Fitbit-like devices that track everything from their heart rates to their time between meals. For about a quarter of the day, the students use Chromebooks and learning software that track their every click and keystroke.

What you鈥檙e seeing is the future of K-12 education through the eyes of Max Ventilla, the CEO of AltSchool, a Bay Area startup that represents the most aggressive, far-reaching foray into the world of big data and analytics that the K-12 education sector has seen to date.

Eventually, Ventilla envisions AltSchool technology facilitating an exponential increase in the amount of information collected on students in school, all in service of expanding the hands-on, project-based model of learning in place at the six private school campuses the company currently operates in Silicon Valley and New York City.

He sees all those torrents of data flowing from the classroom into the cloud, where AltSchool engineers will have built systems for merging the disparate streams into a single river of information. AltSchool software and algorithms created by Silicon Valley鈥檚 top developers and data scientists would then search the waters for patterns in each student鈥檚 engagement level, moods, use of classroom resources, social habits, language and vocabulary use, attention span, academic performance, and more.

The resulting insights鈥攕ay, that 6th graders perform better in math after exercising, or that the girls in a particular science class are bored because boys use the lab equipment more frequently, or that Johnny is using new vocabulary words in conversations with his friends鈥攚ould be fed to teachers, parents, and students via AltSchool鈥檚 digital learning platform and mobile app, which are currently being tested. The information would be accompanied by scheduling tips, recommendations for more gender-neutral science activities, and a playlist of assignments customized to each student.

How those suggestions are used, and whether they make a difference in how well each student learns, would also be tracked, creating a never-ending feedback loop of insights, experiments, recommendations, and product tweaks.

鈥淲e don鈥檛 want to improve some aspects of what schools do. We want a different kind of universe in which schools can exist 30 years from now,鈥� said Ventilla, a 35-year-old Yale University graduate who previously worked as the head of personalization at online-services-giant Google.

For better or worse, it鈥檚 not just pie in the sky talk.

Over the past decade, big data and analytics have slowly crept into the world of public education. Versions of the unobtrusive, real-time, embedded-in-everyday-activities collection of student-learning data now being pioneered by AltSchool and others are touted in the federal government鈥檚 new National Education Technology Plan. Many observers hope the next step is the type of systemwide change that has already transformed the financial sector, health care, consumer technology, retail sales, and professional sports, among other industries.

And while Ventilla鈥檚 plans may seem grandiose, there are some good reasons to pay attention to what the company is doing.

For one, Ventilla has attracted top talent from his old employer, as well as leading companies in consumer technology and some of the top independent schools in the region. The AltSchool team has already prototyped and deployed some of the systems inside its own schools. And fueled by $133 million in venture capital from Facebook CEO Mark Zuckerberg and others, AltSchool鈥檚 50-plus engineers, data scientists, and developers are designing tools that could be available to other schools by the 2018-19 school year.

Still, lots could go wrong. Among other barriers, AltSchool is almost certain to provoke a backlash from parents and privacy advocates who see in its plans the potential for an Orwellian surveillance nightmare, as well as potentially unethical experimentation on children.

But even if the company crashes and burns, some key observers hope its efforts will better illuminate the possibilities and pitfalls confronting a sector still wrestling with the questions of whether and how to embrace analytics.

鈥淭here are a multitude of entrepreneurs who are building learning experiences based on how Google thinks about the world, that you can leverage data in ways that produce beneficial outcomes,鈥� said Robert J. Hutter, a managing partner at Learn Capital, a venture-capital firm that has invested in AltSchool and other companies. 鈥淭he hope is that those experiments will be understood and appropriated by leaders in K-12.鈥�

What Are Big Data and Analytics?

The term 鈥渂ig data鈥� is generally used to describe data sets so large they must be analyzed by computers. Usually, the purpose is to find patterns and connections relating to human behavior and how complex systems function.

Analytics generally refers to the process of collecting such data, conducting those analyses, generating corresponding insights, and using that new information to make (what proponents hope will be) smarter decisions.

For years, public schools and ed-tech companies have experimented with both, usually with two goals in mind: to better personalize instruction, by customizing the learning experience to each student鈥檚 individual skills, abilities, and preferences; and to facilitate more data-driven operational decisions.

Inside school systems, advances have been made. It鈥檚 now common, for example, for classrooms to use learning software and digital games that generate extensive data that can be mined for evidence of student learning. At the macro level, districts routinely analyze large data sets containing information on students鈥� academic performance, attendance patterns, and even involvement with other public agencies. The results are used to predict which students are likely to become disengaged or drop out of school, then to intervene accordingly, among other purposes.

In the ed-tech industry, meanwhile, big data and analytics are everywhere. Companies ranging from Khan Academy to Pearson collect and analyze reams of information on how millions of students interact with digital content. Other companies promise to help district administrators use big data to predict everything from which candidates for teaching jobs are likely to have the biggest impact on student-test scores to where population growth will require that new school buildings be built in the future.

But experts say such initiatives have mostly resulted in small pockets of innovation or incremental shifts to existing practices, rather than systemic transformation.

One big reason: Big chunks of the data currently in use are either stored on paper or in teachers鈥� heads. And much of the digital information in use is generated via students鈥� on-screen and online activity, which even those in the ed-tech world acknowledge can capture only a limited slice of what constitutes real learning.

Other barriers exist, too. Even when new technologies have been introduced into classrooms, teachers have been slow to change the ways they teach. Districts have struggled for years to integrate data housed in separate silos. The education sector is embroiled in debates over how student information should be appropriately collected, shared, and used.

The net result is that school officials often settle for using technology to meet basic compliance requirements, said Jeff Wayman, a former educational researcher at Johns Hopkins University who now consults with districts on effective data systems.

鈥淭echnology got schools to a point where they can get done the things that have to get done,鈥� Wayman said. 鈥淏ut in their heart of hearts, I think a lot of developers would say the technology is so much more powerful than how it鈥檚 being used.鈥�

Analytics in Professional Sports, Other Industries

Other industries have found themselves in analogous positions.

Take, for example, professional sports, including the National Basketball Association.

Until recently, the types of data that even forward-thinking NBA teams collected were fairly limited, said Benjamin Alamar, a former executive with the Seattle SuperSonics (now the Oklahoma City Thunder) and the current director of sports analytics for the giant sports-entertainment company ESPN Inc.

Only a handful of franchises were doing any kind of complex statistical analysis of the information that was collected, Alamar said. Even those teams often stored their information in scattered spreadsheets and databases used by different people for different purposes. Information-related turf wars between departments were common. Many teams suffered from a lack of consensus around which metrics of success really mattered. Inconsistent ways of formatting and handling data exacerbated all those other problems.

The end result was that harried decisionmakers鈥攊n the NBA鈥檚 case, coaches and general managers鈥攐ften struggled to access new forms of information, leading them to rely instead on established habits.

鈥淭he challenges are the same from industry to industry,鈥� Alamar said.

In a short period of time, however, the NBA has been transformed.

To staff their new analytics departments, pro basketball teams now compete with tech giants and blue-chip management-consulting firms for data scientists with computer-programming skills and deep knowledge of advanced statistics.

NBA arenas are now outfitted with sophisticated camera systems that capture the location, movement, and actions of every player on the court 25 times per second. The system has produced a mountain of new data on such minutiae as the optimal number of times each player should dribble before shooting.

And teams are building new information-management systems that integrate such data with quantitative game-performance results, qualitative scouting notes, leaguewide salary information, biometric data on everything from players鈥� sleep habits to their exertion levels during practice, and multimedia files (such as video clips of an individual player鈥檚 tendencies.)

The game itself has changed as a result. One of the original insights of the basketball-analytics movement was that 3-point shots (taken from further out on the court) and layups (taken close to the basket) produce points far more efficiently than mid-range shots.

Changing the Game

The analytics revolution in professional basketball has led to the rapid disappearance of long-range two-point shots.

Source: statemuse.com

In response, teams have made a series of personnel, training, and tactical shifts, resulting in a league-wide trend of players taking inefficient, long 2-point shots far less frequently this year than they did just four years ago, as can be seen in the following charts.

Now, Alamar said, the challenge for NBA decisionmakers is how to stay ahead of the curve. In an analytics-driven culture, 鈥測ou鈥檙e always growing, and you鈥檙e never there,鈥� he said.

From Retail to Education Analytics

As vice president of research and development for retail-analytics firm RetailNext, George Shaw helped large stores usher in similar changes.

Mostly, that meant using video cameras to monitor shopping floors, then applying software and algorithms to the resulting footage to categorize people into shoppers versus employees, for example; track their locations and movements; document each item they touched; and connect all that information with what they ended up buying.

Now AltSchool鈥檚 head of technical research and development, Shaw is at the forefront of developing similar 鈥減assive observation鈥� techniques for education.

鈥淭he idea is to lay down the path for understanding everything that happens in the classroom without the need for any sort of intrusive sensing at all,鈥� he said.

The first big step toward that goal is the company鈥檚 AltVideo camera system, now installed in every classroom across all six AltSchool campuses. For now, the footage is mostly used by teachers and administrators on an ad hoc basis鈥攊f someone wants to review a particularly fruitful interaction with a student, for example.

But Shaw is helping lead the company鈥檚 efforts to begin what he calls 鈥渁utomated metadata production鈥� on that footage. To begin that process, AltSchool is, now testing motion-tracking algorithms similar to those used in both the NBA and retail analytics. Eventually, that could expand to include applying advanced facial recognition, affect detection, and computer-vision algorithms to the footage to generate digital data on students鈥� engagement levels, emotional states, and more.

AltSchool leaders are still imagining the potential uses of such information, said Bharat Mediratta, the company鈥檚 co-founder and chief technology officer.

鈥淔irst, we need to generate the big data,鈥� he said. 鈥淭hen, we start figuring out how to use it to transform education.鈥�

Reasons for Skepticism

Before dismissing such a plan as hopelessly audacious, consider Mediratta鈥檚 pedigree.

For a decade, the 45-year-old Colgate University graduate was a top engineer at Google, where he was responsible for the technological and analytics infrastructure behind the company鈥檚 home page, through which users enter billions of Internet searches every day.

At AltSchool, Mediratta said, he鈥檚 trying to apply similar processes to another 鈥渉umanity-sized challenge": replacing the top-down, slow-moving bureaucratic structures that currently shape public education with a 鈥渘etworked model鈥� in which students, teachers, and schools are connected directly by information and thus capable of learning and adapting more quickly.

鈥淲e will get to the point where we have the same kind of big-data opportunities that Google has,鈥� Mediratta said, 鈥渁nd we鈥檒l be able to take advantage of them.鈥�

Even so, it鈥檚 difficult to envision the country鈥檚 100,000 or so public schools鈥攊n many places still struggling just to get adequate Internet and Wifi connections鈥攁dopting such radically new methods.

鈥淚t鈥檚 not difficult to find many others who have been very accomplished, and brought a lot of money to the table, and still fell short of their own lofty goals,鈥� said Douglas A. Levin, the president of EdTech Strategies, a consulting group on school-technology issues.

For their part, AltSchool鈥檚 founders say that鈥檚 why they are playing the long game.

The company isn鈥檛 trying to create technology tools that have to be quickly sold and shoehorned into traditional public schools. It intentionally avoided the charter-management route taken by other presumptive innovators with big plans to change American public education.

Instead, AltSchool鈥檚 strategy is to invest heavily in research and product development, while also continuing to open more private 鈥渕icro-schools鈥� that serve as labs for those tools to be deployed, tested, and used to generate the data that is ultimately the company鈥檚 lifeblood.

The idea is to first establish a network of schools delivering high-quality education; then develop the technologies that can support the expansion of that network; and ultimately leverage the ever-growing amount of data the network produces to continually make the whole system, schools and technology alike, smarter and more effective.

鈥楳ontessori 2.0'

And what does a 鈥渉igh-quality education鈥� look like at AltSchool?

Ventilla describes it as 鈥楳ontessori 2.0': a kind of supercharged version of the progressive, project-based learning often found in elite private schools and privileged enclaves within traditional school systems.

Play and hands-on learning are stressed鈥攖he less time students spend in front of screens, the better, Ventilla said. Standardized tests are viewed as an inadequate, outdated mode of measuring student learning. Involving parents in the details of their children鈥檚 education is core to the company鈥檚 mission.

That philosophy can be seen in action at AltSchool-Fort Mason, housed in a converted 24-hour fitness center a block from the San Francisco Bay, across the water from Alcatraz Island.

The theme for Christie Seyfert鈥檚 middle-grades class this year is exploring how systems function. 69传媒 are engaged in a months-long, classwide simulation of how different economic systems work. In English/language arts, the students investigated the forces contributing to the city鈥檚 problem with homelessness, conducting real-world interviews with experts as part of an assignment to rewrite an opinion piece on the issue that appeared in the San Francisco Examiner.

Each student received a customized version of the assignment, based in part on insights Seyfert gleaned from her students鈥� daily use of Newsela, a Web tool that customizes newspaper articles to each child鈥檚 reading level, then tracks their progress in developing specific skills.

鈥淗ere, we鈥檙e using data, but in ways that are so personalized to kids鈥� needs, learning styles, passions, and goals,鈥� said Seyfert, who previously worked both as a Teach For America corps member in a tightly regimented public school environment and as an adult 鈥渃ollaborator鈥� in an independent, project-based K-12 school.

鈥淣ow, I have an understanding of where [students] are with core skills, but there鈥檚 also that depth of engagement in what they鈥檙e learning,鈥� she said.

The trade-off, though, is being subjected to AltSchool鈥檚 extensive monitoring. The company believes that scaling personalized, project-based learning is only possible with the new forms of data collection and analytics it is developing.

A great teacher might be able to gather and process and use information about the emotional states, social interactions, and classroom reactions of a handful of students at any given time, Ventilla said. But getting thousands of teachers to consistently do that for 20-plus students apiece requires a big technological boost, he believes.

So far, at least, it鈥檚 a deal that dozens of AltSchool-Fort Mason families have proved willing to pay $26,250 a year to make.

鈥淭hey use the data they get to improve their own performance, for research, and for feedback to teachers and students,鈥� said Mark Eisner, the vice president of a Silicon Valley biotech firm who has a 13-year-old daughter at the school. 鈥淚 honestly have a hard time understanding the resistance to that.鈥�

Despite the intense privacy-related concerns that others hold, AltSchool is betting that Eisner鈥檚 attitude will eventually carry the day. Three decades from now, the company hopes, schools won鈥檛 be able to imagine operating without the types of big-data-driven technologies that it is developing鈥攋ust as most schools now couldn鈥檛 imagine functioning without light bulbs, personal computers, or Internet search engines.

Such a transformation is already well underway in other sectors of the economy. Even if AltSchool isn鈥檛 the one to usher such sweeping shifts into the world of public education, others are amassing on the horizon.

鈥淔olks that are coming from the Internet and more digital domains, we tend to want to be able to constantly measure ourselves, because we need to be constantly improving,鈥� Ventilla said. 鈥淭he model for education right now is not very susceptible to change. But give us time.鈥�