The Shadow of the Torturer
It tells a story. Hardcover files this book as fiction. A fact, not a factor: it plays no part in our call.
Read it
Thin evidence · 152 ratings of this reading, short of the 250 we trust. So the call is made without the recording’s half of the case. Why we hold that line
Audio is doing this one no favours: 21% of the reviews we read called it demanding. That is a book that wants a page you can turn back to. Only 152 people have rated the reading, so it is not part of this call either way, and 250 is where we start letting a judgement about a reading move a verdict.
Length does not decide this
11h 38m to listen, at 1×9h 18m to listen, at 1.25×7h 45m to listen, at 1.5×6h 39m to listen, at 1.75×5h 49m to listen, at 2×about 9 hours to readabout 7 hours to readabout 5 hours to readabout 4 hours to readabout 3½ hours to read
The runtime is Audible’s own, shown at the speed beside it. The reading time is our estimate: 262 pages at 30 pages an hour, the pace you told us. Ours is 50, about 300 words a page, read at 250 a minute. Treat the reading estimate as a floor on this one: this one runs 2.7 minutes of narration a page against a shelf median of 1.9, so there is more on each page than our 300-word average assumes and it will take you longer than we say.The runtime is Audible’s own, shown at the speed beside it. The reading time is our estimate: 262 pages at 40 pages an hour, the pace you told us. Ours is 50, about 300 words a page, read at 250 a minute. Treat the reading estimate as a floor on this one: this one runs 2.7 minutes of narration a page against a shelf median of 1.9, so there is more on each page than our 300-word average assumes and it will take you longer than we say.The runtime is Audible’s own, shown at the speed beside it. The reading time is our estimate: 262 pages at 50 pages an hour, which is about 300 words a page, read at 250 a minute. Treat the reading estimate as a floor on this one: this one runs 2.7 minutes of narration a page against a shelf median of 1.9, so there is more on each page than our 300-word average assumes and it will take you longer than we say.The runtime is Audible’s own, shown at the speed beside it. The reading time is our estimate: 262 pages at 65 pages an hour, the pace you told us. Ours is 50, about 300 words a page, read at 250 a minute. Treat the reading estimate as a floor on this one: this one runs 2.7 minutes of narration a page against a shelf median of 1.9, so there is more on each page than our 300-word average assumes and it will take you longer than we say.The runtime is Audible’s own, shown at the speed beside it. The reading time is our estimate: 262 pages at 80 pages an hour, the pace you told us. Ours is 50, about 300 words a page, read at 250 a minute. Treat the reading estimate as a floor on this one: this one runs 2.7 minutes of narration a page against a shelf median of 1.9, so there is more on each page than our 300-word average assumes and it will take you longer than we say.
If you must listen → dedicated listening only
Harder work in audio than The Handmaid's Tale.
Harder work in audio than 95% of the books we hold.
Harder work in audio than 94% of the fiction we hold.
Harder work in audio than 96% of the middling-length books we hold.
Reading load against listening load
Where that puts this one among the books we hold: in the top 4% of our shelf of books for effort on the page and in the top 4% of our shelf of books for effort in your ears. The middle of the scale is not the middle of the shelf (most books sit below the average on both), so a book at 0.00 is already in the top 43% for effort on the page. Each of the three is measured against the whole shelf separately, so the bottom figure is not the top two subtracted. It is this book’s own distance from the middle.
What the audio disadvantage is made of. Two things and no others: how fast readers say this one moves, and how demanding the reviews we read found it. There is no third: the structural one has no data source, so it contributes nothing rather than being guessed at. Both directions are declared priors we have never fitted against a real verdict, which is the weakest-evidenced part of the whole model.
How the call was made
Where a bar carries ticks, the grey one is the middle of our shelf and a coloured one is a line we publish, cool where we start backing the page, warm where we start backing the recording. The evidence squares at the foot are a different thing: how far each of our four sources went, hatched where there was nothing to measure at all.
- The audio case
- −1.73With the reading left out, this figure is the format alone: it is harder work in audio than most books we hold (in the top 4% of our shelf of books for extra work in audio). Past the line where we start backing the page.
- Narration payoff
- 0.00The reading took no part in this call. It scored 1.79 above the middle of our shelf of books (the second mark on the bar), but only 152 people have rated it, short of the 250 at which we let a judgement about a reading move a verdict, so it is recorded at 0.00 and the call rests on the book and the format. No line is drawn on this row, because none was applied.
- What is at stake
- −1.15It sits in the bottom 10% of our shelf of books for how well the print crowds think of a book: not a book the crowds are precious about, so there is less to protect and we can be relaxed about it. Blended across 38,100 Goodreads ratings and 381 on Hardcover, each scored inside its own population first, and never added together.
For the page
- A hard text to follow by ear
- The crowds’ regard did not reach the line where we protect a first read, this reason did not apply
For the recording
- Too few listeners have rated this reading, so we left the voice out of the call, this reason did not apply
- Only 152 listeners have rated this reading, under the 250 we ask for, this reason did not apply
So the case for the page won it clearly.
A bigger dot counted for more: the voice and the ear are sized by their share of the sum above, the other two by how much stands behind them. A hatched square is a reason that did not apply here, which is not the same as one that counted for little.
- Evidence
- 0.93How much of the picture we actually have, out of 1.00: four questions, each answered as far as the source could answer it, less anything taken off for reading fewer reviews than we aim to.The reading152 people have rated this reading: enough to lean onThe second sourceHardcover knows this book, with 381 readers there: as good as it getsThe matchHow sure we are that the recording we scored is this book: as good as it getsThe reviewsWe read 300 reviews to judge how demanding it is: as good as it gets
What each crowd complains about
We would need 40 reviews from each crowd before setting the two side by side, and only 35 of the reviews we read said they listened to it. So we are not drawing this one.
What the two crowds said
Showing the stars as each shop prints them. The two crowds are not measured from the same place: on our shelf Audible’s listeners average 0.35 of a star above Goodreads’ readers, so a recording that is ordinary for Audible can still out-star a book that is genuinely well liked for Goodreads. Switch the toggle to take that difference off.Showing each figure as its distance from its own crowd’s middle on our shelf, in standard deviations, so 0 is a typical book for that crowd. Listeners start 0.35 of a star up, partly because anyone who finishes a twenty-five-hour recording chose it deliberately, and this takes that off. It is not a claim that audio is better, the story figure is still Audible’s own (people rating a story they heard), and none of it moves the verdict above.
152 ratings of this readingstar chartstoo few to draw beside the book’s, and 250 is where we start trusting the shape
overallAudible
overall, pooled by the shop, so we hold no ruler for it
174 ratings · performance and story, as the shop pools them
the performanceAudible
the performance, against Audible’s own crowd
152 ratings
the story, as heardAudible
the story, as heard, against Audible’s own crowd
152 ratings
the book itselfGoodreads
the book itself, against Goodreads’ own crowd
38,100 ratings
The series
The Book of the New Sun
Finished
5 books numbered on Goodreads’ list. The last one came out in 1987 and Goodreads lists nothing further. Five quiet years is where we start using that word.
From Goodreads’ own series index, checked on 11 September 2026.
Every reading we hold
One complete reading of this book, and it is the reading every audio figure above is measured on. If there is another, we have not found it. The shelf grows in reviewed batches, never because somebody searched for a book.
- James Laileythe one we measured11h 38m9h 18m7h 45m6h 39m5h 49m · One voice · Audible UK and US
Only 152 people have rated this reading, short of our published bar of 250, so we name it and print no scores for it.
Hear it · Audible’s clip, its length read from them on load
The recording these numbers came from
- Runtime
- 11h 38m at 1×
- Read by
- James Lailey
- Kind
- One voice
- ASIN
- 1473233070 · Audible UK
- Rated in
- Audible UK and US · 149 + 3 ratings of the performance
- Judged on
- 152 ratings of the narration
- Published
- Gateway · 2021
Too few people have rated this reading for us to call it our pick (our published bar is 250), so these are its details rather than a recommendation, and Audible sorts a search by newest, so check before you buy. The runtime is the check: if the one in front of you says 11h 38m, that is the reading these numbers describe. “Judged on” is the crowd behind every judgement we make about this reading: the smaller of the performance and story counts, which is why it is a little under the two shop counts added. We do not link out to a shop, and we take nothing if you buy there.