Why Phatiks

The reading crisis

2026 has been designated the National Year of Reading but it arrives at a time of crisis for literacy in the UK and North America. In 2025 the National Literacy Trust recorded the lowest levels of reading enjoyment it had ever measured: only about a third of 8-to-18-year-olds said they enjoyed reading in their free time, and fewer than one in five (18.7%) read something daily, roughly half the daily rate of two decades ago. Even among children aged 5 to 8, daily reading fell by 3.4 percentage points in a single year. The 2026 survey, published during the National Year of Reading itself, shows the first green shoots of recovery — enjoyment up to 36.1%, daily reading to 20.3% — but both remain far below even the levels of a decade ago, and the gaps by gender and by disadvantage are widening.

The consequences do not stay inside reading. Poor comprehension travels: it depresses performance across the curriculum, and above all it erodes the capacity to think deeply about writing, what Maryanne Wolf calls deep reading: connecting background knowledge, making inferences, taking other perspectives, reading critically (Reader, Come Home, 2018). In English itself, top grades remain elusive: only around 15.0% of A Level English pupils achieve A*–A grades against 28.3% across all subjects (Nuffield Foundation / FFT Education Datalab), and English Language and Literature entries have declined by roughly 11.3% over seven years, outpacing broader A Level entry changes. The OECD's Survey of Adult Skills (2024) found adult literacy scores falling on average across most member countries over the past decade. In the latest PISA reading assessments the UK scored above the OECD average - but with a recorded decline and significant disparities: 29% of Welsh students fell below the baseline literacy required to participate fully in society (PISA for Schools, OECD, 2026).

As part of the response to this 'crisis' in reading (The Guardian) - or 'the end of reading', as The Atlantic put it this year - the Education Select Committee opened an inquiry into reading for pleasure, publishing its report in July 2026 and warning that the decline is not only an educational concern but a challenge for the whole of society.

The need is clear, and not only for grades. Reading for pleasure is one of the strongest indicators of future life chances, more significant even than socio-economic background (OECD). It is one of the fundamental drivers of our ability to grow, as individuals and as members of wider, thriving communities. As the philosopher Martha Nussbaum has put it, stories are the birthplace of 'compassionate imagination'. Or as Marcel Proust argued, deep reading allows us 'to leave the wisdom of the author behind to discover our own'.

The probable cause

Although the reasons behind this crisis are many and their interaction complex, there is one abiding suspect: the ubiquity of the digital screen. The Select Committee's report identifies the rise of screen use as a major factor in reducing the time children spend reading for pleasure.

The same report is careful, though, and we should be too: the decline cannot be laid solely at the door of screens. That caution suits us. Our argument was never that screens are the enemy, it is that the behaviour screens currently train is.

A 2024 meta-analysis of dozens of studies confirmed a robust 'screen inferiority effect': readers comprehend and retain less from screens than from the same text on paper. Maryanne Wolf's argument is that deep reading - connecting background knowledge, making inferences, taking others' perspectives, critical analysis - is precisely what skimming crowds out, because when we skim we physiologically don't leave time to think.

A 2025 review of two decades of reading research (Norman, Taylor and Rodd) reveals how this works: a word becomes easy to read not because we see it again and again, but because we encounter it many different contexts - the mind holds on to the words it judges most likely to be needed again. Every encounter with a word in a slightly new setting does a little building work. Skimming is corrosive precisely because it thins each encounter to nothing: the eye passes over the word without ever really placing it anywhere.

TikTok, YouTube Shorts and Instagram Reels are all seen as ushering in the post-literate society. The irony is that, however saturated these forms are with competing visual stimulation, most remain heavily text-driven: young people 'multi-screening' run subtitles to keep tabs on films they are half-watching, as an ambient rather than focused narrative experience. The words are everywhere; the reading has thinned. Research from the Haskins Global Literacy Hub points the same way: poor ability to maintain focus, low tolerance for mental exertion, scanning that produces surface understanding.

Ban it, use it, or change the behaviour

Three responses are on offer. The first is to ban: remove the phones, fence the classroom, and hope the habit does not follow pupils back through the school gate. Restriction has its place - but the skimming habit is trained for hours every evening, and no bell protects the exam hall from behaviours learned elsewhere.

The second is to use: put the curriculum on the screen as it stands - worksheets as PDFs, textbooks as apps - and inherit every behaviour the screen has already taught. This is where most educational technology sits, and it is one reason the screen inferiority effect keeps being replicated.

The third response is the one the research quietly licenses. If comprehension collapses on screens because of trained behaviour rather than the technology itself, then behaviour can be trained the other way. A screen can be designed so that scanning, skipping and guessing are losing strategies — so that the only way to score is to read closely, hold evidence, and think before answering. That is the opportunity Phatiks exists to take.

What Phatiks builds

Phatiks builds reading and language games for secondary English (KS3–KS5) in which close, evidence-grounded reading is the winning strategy. You cannot guess your way to a high score in a Phatiks game. Every engine is built so that scoring rewards precision, evidence and re-reading - and punishes the skim. Highlight the wrong words, lose the streak. Infer without evidence, drop a ring. Miss the stylistic fingerprint, lose the case.

Gamification usually means points bolted onto a worksheet. Here the game is the reading discipline. The mechanics are the pedagogy: the chunked reveal that prevents skimming ahead; the evidence budget that makes pupils choose their words rather than sweep the page; the second inference that only pays if you go back and re-read.

And every game can be teacher-authored. A teacher pastes in the text their class is studying; questions, tags and tasks are drafted, checked and edited on the teacher's side before any pupil plays. The teacher stays the expert in the room.

Two doors on one foundation

Every Phatiks game belongs to one of three layers, so the suite grows as a system rather than a grab-bag.

The Toolkit is the foundation: the parts of language and the names for them. Texterminators drills grammatical terminology. MegaMorph teaches how English words are built and taken apart: free and bound morphemes, roots, prefixes, suffixes. Speed Reader trains fast, accurate recognition of what is actually on the page. INFER-RED teaches pupils to close in on an unfamiliar word's meaning by elimination.

The science under the foundation is specific. Recent work on reading acquisition (Taylor and colleagues, 2026) taught adults two invented writing systems - one where symbols mapped systematically onto sounds, one where they did not - and found the brain learns whichever system it is handed by finding its regularities, choosing a different neural solution for each. English is the systematic kind (mostly). Its spellings map onto sounds often enough to be worth learning as a system, and where the sounds mislead, its morphemes hold the pattern steady: 'unaccountable' is irregular as a run of sounds and perfectly regular as a stack of meaningful parts. Fluent reading is built by learning those regularities, not by memorising words one at a time - and that is precisely what the Toolkit trains. This is not decoration on top of comprehension; it is the machinery comprehension runs on.

Rising from the foundation are two doors, because there are really only two things an analytical reader does with a text: read into it, or read across it.

Depth is vertical reading: one text, read to the bottom. Its flagship is Subtext - chunked reveal, heat-scanning for expressive language, chained evidence tasks, scored highlighting for precision and coverage, and a second inference that only pays if you re-read.

Forensics is horizontal reading: texts against each other. Its flagship is Whodunnit? - stylistic fingerprints, register, idiolect and authorship, investigated case by case, with the evidence assembled before the accusation is made.

The research divides reading almost exactly where this architecture divides it. The 2025 review carries a second finding that is easy to miss: meeting a word in many contexts makes it easy to recognise - but it does not make its precise meaning easy to pin down. For the harder task of choosing one exact meaning in one exact place, words that have been met everywhere are actually harder, not easier: the many contexts compete. Breadth of reading and precision of reading are not the same skill. One does not deliver the other.

This is the sentence a parent already knows without the citation: my child reads all the time and still can't tell me what the chapter was about.

Wide reading builds recognition and fluency. It does not, on its own, build the precise, evidence-grounded reading a GCSE inference question demands. That has to be trained directly. The Toolkit builds breadth; Depth trains the precision breadth never delivers on its own. We are not guessing that close reading is a distinct skill. Two decades of lexical research say it is.

Why not a worksheet, why not a chatbot

Why not a worksheet? Because precision is a trained skill with a real cognitive cost, and a worksheet cannot enforce the cost. It cannot stop a pupil skimming straight to the questions; it cannot score the difference between an answer grounded in the text and a plausible guess; and it cannot hand the teacher usable data without also handing them marking. Phatiks scoring rewards evidence and punishes the skim, and the data flows back to the teacher automatically.

Why not a chatbot? Because a chatbot rewards fluent asking, not close reading - and because putting a generative model in front of children raises data questions we refuse to raise. Phatiks operates one hard rule: all AI generation happens on the teacher's side, at authoring time, and is cached. No pupil free text ever reaches a model at runtime. This is not a configuration option; it is the architecture — and it keeps the product squarely within the ICO's Age Appropriate Design Code and the DfE's expectations on pupil data.

Written by a sceptic

A year ago, Phatiks's founder stood in front of a school and delivered an hour's argument against screens in classrooms.

The speech began with Socrates - who distrusted writing itself, fearing it would move knowledge out of the head and onto the page — and traced a line through the alarms that greeted print, television and the smartphone, conceding along the way that every generation's alarm sounds the same, and that the people raising it are usually men in their fifties. Then it made the case that this time the alarm is justified: a House of Commons report finding that nearly a quarter of young people use their smartphones in a way consistent with behavioural addiction; the neuroscience of screens promoting sensation at the expense of thought; the meta-analysis behind the screen inferiority effect; and EdTech's habit of marking its own homework - glossy impact statistics whose source, on close inspection, turns out to be the company selling the device.

A business proposition from the same author might look like a contradiction. It is the opposite. That speech proposed a three-part test for any technology asking to be let into a classroom - on principle, in practice, and against the whole point of education - and Phatiks was built to pass it.

On principle, it matters what is being sold, and by whom. Phatiks is not a technology company looking for a way into education; it is a teacher's answer to a problem watched at close range for nineteen years. Nothing in a Phatiks game is designed to harvest attention: no infinite scroll, no engagement loops, no advertising - and no pupil data flowing to anyone, with the no-runtime-AI rule above as the architecture of that promise. The scoring rewards patience, evidence and re-reading: the exact behaviours the attention economy trains against, deliberately engineered back in.

In practice, the speech attacked evidence that came from the vendor's own brochure - so we hold ourselves to the standard we demanded of others. We will not present our own classroom's enthusiasm as proof: internal use is ring-fenced as usability testing, and the evidence that counts will come from external schools and arm's-length validation such as the DfE EdTech Testbed. Until that evidence exists, we will say so.

And on the point of education: the speech argued that education is learning to think deeply, that deep thinking requires close reading, and that close reading requires practice and patience. Phatiks does not dispute a word of that. It is the founding premise. The games are not a substitute for the book; they are training for it - deliberate practice in the disciplines of close reading, on the very device that currently trains their opposite, designed to send pupils back to the page stronger readers than they left it.

There is a precedent for this move, and it is the oldest one in the story. Socrates distrusted writing; his student Plato used writing to preserve the dialogues - and it is only because he did that Socrates has a voice at all. The technology that threatened the older value was made to serve it. That is the whole of Phatiks's ambition for the screen.

The sceptical English teacher, in other words, is not our hardest audience. They are our founding demographic. Every objection they might raise - screens erode attention, EdTech oversells, games trivialise - is an objection this company was built inside of, by someone who has made each of them out loud.

Who we aim to support

This is how we think our games can help three different groups of people.

The mission

Sources. National Literacy Trust, Annual Literacy Survey 2025 and 2026 · Education Select Committee, Reading for Pleasure (report, 17 July 2026) · House of Commons Education Committee, Screen Time: Impacts on Education and Wellbeing (2024) · Delgado et al. (2018) and subsequent meta-analytic work on the screen inferiority effect · Maryanne Wolf, Reader, Come Home: The Reading Brain in a Digital World (2018) · Norman, R., Taylor, J. S. H., & Rodd, J. M. (2025), 'The effects of contextual diversity on lexical processing: a scoping review', Psychonomic Bulletin & Review, 32, 2763–2806 · Taylor, J. S. H., Jowett, A., Auer, T., Hsieh, C.-Y., Lingnau, A., & Rastle, K. (2026), 'The nature of a writing system shapes the cognitive and neural mechanisms for reading acquisition', Neurobiology of Language, 7 · OECD, Survey of Adult Skills (2024); PISA for Schools (2026) · Nuffield Foundation / FFT Education Datalab, A Level English entry and grade analysis. See also our links & further reading.