Middleware is useful in this context for the same reason it is useful anywhere -- because applications can integrate to the middleware, rather than integrating point-to-point with every other application.
The sort of effective personalization that drives advertising and marketing systems at Amazon or Google requires the accumulation of large amounts of data pegged to an individual. Suppose a teacher introduces some wonderful math learning software into the classroom, and Johnny works with it an hour a week until it starts to detect patterns in his learning style and help him learn better. Ideally, this information would not be locked up in one application but follow Johnny from class to class and school to school and year to year. When he gets to tackling the equations of high school physics, some of the knowledge about how well he performed in algebra and trigonometry comes along with him.
Of course, I'm not helping inBloom's case by making comparisons to the personalization on Amazon or Google, since plenty of people are uncomfortable with the data those organizations collect about us, also. But, dammit, those "if you liked this book, you might like that book" recommendations can be helpful sometimes. Might not it be a good thing if our education software systems could be at least that smart? As in: "If you had trouble with quadratic equations, you might also struggle with this exercise on the motion of projectiles."
For these benefits to materialize, the inBloom service will have to work as advertised, which shouldn't be taken as a given. Also, the data integration service is only one part of the picture I'm painting of individualized education -- the tutoring software and student information systems at the school level would also have to make intelligent use of the data.
The data-broker role inBloom aims to fill could be filled by one of the software and services vendors in the education market, except that no competitor wants to be subordinate to any other, explained Shawn Bay, the founder and CEO of eScholar, which is working closely with inBloom on the New York State project. Better to have that role filled by a nonprofit, he said. "Theoretically, they're not going to be biased toward any of us." Also, if inBloom wants to take on the complexities of single sign-on authentication, "we're perfectly happy to have it taken off our plate," he said.
Having learned the power of consumer data analytics as an employee of Proctor & Gamble in the 1980s, where the currency was retail scanner data, Bay later led a consulting firm that first became involved in data-analytics work for a New York school district in 1997. While he sees the potential of inBloom, and appeared on an SXSWEdu panel promoting it, he said many of the practical details of working with the master database are still in flux. For example, the architects of the New York State project are finding that having applications interactively fetch data directly from inBloom via Web services protocols doesn't work for performance reasons, so they're having to layer in data caching.
When we sat down over breakfast before a scheduled Bill Gates keynote speech, Bay wondered whether Gates would acknowledge that today's goals for inBloom sounded a lot like what was promised for the Schools Interoperability Framework Microsoft proposed in 1999, which was supposed to ensure "that software applications in K-12 schools can share information seamlessly." The closest Gates came to that was mentioning that the late 1990s was the last time there was a bubble of optimism about the potential of technology to improve education, but it had proven harder than people like him expected. To be fair, some of the technical underpinnings of today's data-integration initiatives can trace their beginnings in part to what Microsoft proposed back then, now with the blessing of federal education officials.
InBloom isn't the only organization trying to translate these ideas into action. Also at SXSWEdu, the Michael and Susan Dell Foundation announced that it was spinning off its education data integration initiative, the Ed-Fi Alliance, to operate as a separate nonprofit. One of the distinctions between Ed-Fi and inBloom is that Ed-Fi has produced an education data analysis dashboard, which it has released as open source software, in an effort to make the data collected about students more useful. In contrast, inBloom is concentrating on the data backend and leaving the user interface to others. The two organizations say they're not rivals, but rather complementary services. Some school systems have partnered with both, using Ed-Fi as an intermediary service to organize data and load it into inBloom.
Yet Bay expressed some frustration with the overlap, saying, "let's just merge these things and solve the problem."
Which brings me back to the problem this is supposed to solve.
"As a teacher, I used technology a lot, but I always looked at all those different toolsets and said, 'I wish they all worked together,'" said Jim Peterson, technology director for the public schools in Bloomington, Ill. "It's even harder when your school district comes back with a great new product that adds to your 36 passwords." For him, the promise of inBloom is to "get rid of that major pain point."
"This inBloom plumbing is boring stuff, and if it works right we shouldn't have to pay attention to it," said Ken Wagner, a New York state education department associate commissioner responsible for curriculum assessment and education technology. "The interesting stuff is the tools on top."
By freeing data from being locked inside proprietary tools, we "get an ecosystem where the tools have to compete with each other," Wagner said.
That's the dream. In some ways, the nightmare of privacy breaches and misuse of private student data is easier to understand, maybe even easier to believe in. Yet I would like to believe we can address those concerns with sensible safeguards, rather than letting fear rule.