The OECD reported this morning that average reading among 15-year-olds across its member countries has fallen about 28 points since 2015 — on its own reckoning, roughly a year and a half of learning. One in five is now a low performer in reading, mathematics and science together. The schools story is being written elsewhere.
This is about everyone downstream, who have spent a decade describing their difficulty as a competition for attention and may have been describing something else.
Competition and capacity are different problems
The standard account of why fewer people read books runs on substitution. There is more to do. Attention is contested by services engineered to hold it. Reading loses on convenience.
If that is the whole story, the remedy is competitive: better books, better recommendation, shorter forms, meeting readers where they are. Publishing has spent fifteen years on exactly that, with some success.
A capacity problem is different. If a fifth of a cohort finds sustained prose genuinely effortful — not less appealing than a video, but harder — then no amount of better marketing addresses it, because the constraint is not preference. Reading for pleasure requires fluency sufficient that the decoding is invisible and the mind is free to be somewhere else. Below that threshold reading is work, and nobody does optional work for recreation.
What depends on this pipeline
More than book sales, which is why this is a culture story rather than a trade one.
Long-form journalism assumes a reader who will stay for three thousand words. Criticism assumes someone who has read the thing being criticised. Theatre and literary adaptation assume an audience with the reference. Even the recent good news in this sector runs on it: public libraries have been reinventing themselves as the last free indoor space in town, and regional magazines have been revived by owners treating them as civic institutions. Both depend on a population that reads comfortably.
So does the archive. A back catalogue is only an asset while someone can read it.
The honest caveats
Two, and they matter.
PISA measures a specific construct — comprehension and retrieval under test conditions at 15 — and it is not the same thing as reading for pleasure. A student who scores poorly at 15 may read fluently at 30; adult literacy is not fixed at that age, which is the premise of every adult education programme and one this desk has covered as night school enrolment has climbed.
And the causes are contested. Attributing this to phones is the available story and the evidence is not clean, not least because the decline was visible in data collected before the devices reached saturation among this age group.
What survives both caveats is the direction and its breadth. It is happening across systems that share nothing else, and it has continued through a decade in which many countries tried many remedies.
What a serious response would not be
It would not be shorter books, which treats the symptom as the preference.
The institutions with the strongest position here are the ones that build fluency rather than assume it — libraries running programmes for the age group in question, publishers investing in the school and young-adult market as pipeline rather than as a segment, and anybody willing to treat a decade-long capacity problem as their own rather than the education system's.
That is an unrewarding investment with a fifteen-year payback and no way to attribute the return. It is also the only one that addresses what the numbers actually say.
What to watch
Not book sales, which are dominated by a small number of titles and will not register this for years.
Watch reading-for-pleasure surveys among 15- to 24-year-olds, which several countries run annually. Those measure the behaviour rather than the capability, and if the two diverge — capability falling while pleasure reading holds — then the substitution story was right after all and this is a smaller problem than it looks. If they fall together, publishing's audience is not being competed away. It is not arriving.
The PISA 2025 findings of an approximately 28-point decline in average OECD reading performance since 2015, described by the OECD as roughly a year and a half of learning; the rise in the share of low performers across all three subjects from 16 percent in 2022 to one in five; and the 14-point fall in US reading scores with 25.7 percent low performers are as published by the OECD on 9 September 2026 and reported the same day. No causal claim about screen use, curriculum or any other single factor is made. The consequences described for cultural institutions are our own extrapolation and are not part of the OECD's findings.
Topics culturepublishingeducationlibraries
