The ubiquitous algorithm to improve children’s reading is a blunt tool that could be doing more harm than good, writes Shivan Davis

Allow me to lay my cards on the table: I don’t like Accelerated Reader. I don’t trust the accuracy of its Star Reader test. I don’t think their quizzes prove a student has actually read a book. Nor do I like withdrawing my students from lessons every half term to track their literacy rates. I resent having to trudge through endless sheets of figures, and I am convinced the money thrown at the programme could be better spent.

Of course, I am all in favour of encouraging independent reading and cultivating a culture in which the school library is seen as the beating heart of a school. But I question our blind trust in programmes like AR to recommend texts, track student progress and guide us in setting homework.

At my previous school, AR became a cornerstone of the English department. Almost immediately after its introduction, one lesson a week was given over to independent reading in the library and a policy of setting fortnightly reading homework was instituted.

Unsurprisingly, library lessons were an instant hit with students. Even more unsurprisingly, it was often not for the right reasons. Students used the lessons as an opportunity to have discrete conversations, play games under the tables or simply stare into space. While AR probably benefitted a minority who already read fluently and for pleasure, it did nothing for the vast majority who didn’t.

Library lessons were an instant hit, but not for the right reasons

AR is a blunt tool. Like all blunt tools, connecting it to sanctions only leads to perverse incentives. If students are expected to read a book a fortnight and take a test to prove they have read it, what happens? Students desperately take quizzes on books they haven’t read. And teachers, believing in the columns of data output by AR, go on to shrug off the continuing appeal of Diary of a Wimpy Kid and Gangsta Granny. After all, the algorithm has determined these books to be within the students zone of proximal development.

So over time behaviour in reading ‘lessons’ improved. But the quality of the literature did not. Year 9s still read the works of Jeff Kinney and David Walliams and, lo and behold, their reading levels remained relatively low. Given how much curriculum time we allocate to independent reading, is it not incumbent upon us to search for a better solution?

Interviewed in these pages, the new president of the School Library Association, Richard Gerver worried that a consequence of budget restraints could be that “one of the first things to go might be the trained librarian”. It is a grim prediction, and one all too likely to come to fruition in a sector that invariably looks to the latest digital product for solutions.

As a profession, we need to think more deeply about trade-offs and opportunity cost. In the case of reading programmes like AR, the opportunity cost is huge. For every minute spent on it, a minute of curriculum time spent on a challenging text with a subject specialist teacher is lost. So too is the whole-school influence of a passionate librarian.

The personalisation on offer from such data-driven systems is a sham. Not only can an algorithm not replace qualified professionals, but reading can’t be reduced to an individual practice. We should read more books as a whole class, guide our students in grappling with complicated narrative structures, idiosyncratic narrative perspectives, enigmatic plots and multi-faceted characters.

David Didau argues schools would be better off spending their budget on a collection of set texts for each year group – “books we decide all our students have an entitlement to”. To narrow the reading gap, he encourages us to model reading aloud instead of opting for independent reading. My experience of AR convinces me he is right.

Many schools halt any effort to foster reading at key stage 4, in part because of the dent AR leaves in their budgets. But what improvement might we see across the curriculum if we truly valued collective reading over data harvesting?

When it comes to teaching, knowledge is power, but data is not intelligence.