December 2021 – January 2022

Vox

A homemade algorithm chooses every book I read

By Russell Marks
How a monumental to-read list turned into a spreadsheet that randomly selects books

For a while now, I’ve listed every book I read. I imagine many people do at one time or another. My list began in April 1997, when I was 15. It was handwritten at first, new sheets of paper stapled onto existing sheets. Frequent use made the paper soft, and the crease across the middle, where the fold was, threatened to erase the title halfway down each page. So, I backed it up, first on Microsoft Word but then on Excel, which has more data-sifting possibilities. With Excel, not only could I record details about each book – title, author, year of publication, number of pages, date begun, date finished, a star rating – but I could also generate aggregates. How many books was I reading each year? What proportion were by women? By people of colour? By Australians? By non-Western authors?

Once my electronic list of books I had read was functioning, I began a second list of books I wanted to read. It began with books I already owned, friends’ recommendations and those featured on ABC Radio’s Late Night Live. It soon included all the Booker longlists. Then the Miles Franklin and Stella longlists. The premiers’ prizes, and Kevin07’s prime minister’s prize. Then all the other prizes. Orange. Hugo. Aurealis. Ned Kelly. The Children’s Book Council of Australia. Amazon’s Best Translated Book Award. And dozens of others. Then, of course, I had to add all those “best of all time” lists. Because it was an electronic list, there was no limit to the number of titles I could include. There was therefore no reason not to add everything in 1001 Books You Must Read Before You Die and Harold Bloom’s The Western Canon, and all of Virago’s Modern Classics.

I now had a problem. There were thousands and thousands of titles on my “to read” list. At the rate I read (26.97 pages per day since 1997, or 44.54 since COVID-19) and given my current life expectancy (the Bureau of Statistics gives me until April 27, 2063), I won’t be reading them all. I had to choose from among them. But experts say that a choice becomes debilitating when the available alternatives number greater than about six. At the time of writing there are 27,512 titles on my list of “Books I Want to Read”. I should add “But Probably Never Will”. I’m reminded of my mortality every time I open the spreadsheet, which, given the size of the file, now takes nearly a minute.

How to decide among this many choices? For the uninitiated: when one types “= RAND ( ) * 27512” into Excel, it returns a random number between 1 and 27,512. That number corresponds with a book title. I realised I could simply outsource my decision to a spreadsheet function. A problem outsourced was a problem solved. I briefly considered a career in government.

There soon arose another problem, though. A random number is truly random. There are some books I actually do want to read more than others. I don’t particularly want to read three 18th-century philosophical treatises in succession, which would be a possible random result. So, I assigned each title a category (Australian Fiction, World Fiction, Australian History, Children’s Fiction, Philosophy & Science et cetera), as well as a weighting (just “high” and “low”, nothing overboard). The random number function now operates according to a cycle: each category (rather than each individual title) has an equal chance of being selected, but additional books from the same category have a much-reduced chance of selection until the cycle resets, which happens once I’ve read at least one book from each of the 20 categories. The spreadsheet now also selects essays, short stories, journal articles, significant reports and court cases. And audiobooks.

I’d been beavering away on this Excel file quite happily until I began spending significant amounts of time with my first spreadsheet-era (SE) girlfriend. Suddenly I could see how weird it looked. The words “Excel” and “incel” aren’t different enough. So, I consciously began keeping it a secret. The idea that anyone might find out about it became a source of shame. Inevitably she caught me. She looked at me like I’d just told her I lash myself whenever I feel happy. Latterly there’s been more self-acceptance. When I worked up the courage to show my new partner, she demanded to come onto the spreadsheet’s management committee. We’re still together and going strong.

A few years ago, I was asked to sit on a Newcastle Writers Festival panel titled “The Writer as Reader”. I barely considered myself either. On the same panel were Geordie Williamson, The Australian’s chief literary critic, and Kate Holden, a bona fide author. (Each of their books was already on my list.) Somehow, I had to find something to talk about with people who’d have something meaningful to say in front of a large and very live audience. The only ace up my sleeve was my secret spreadsheet, by then a decade old. I played it. It seemed to work. People said they wanted to follow me on Twitter.

Now I talk about it all the time. “But what if you’re not in the mood to read a book the spreadsheet selects?” I get asked by friends, who, not unreasonably I suppose, consume culture on the basis of how they feel at the time. They see only severe restrictions on my agency. But I see the document as a kind of literary welfare state or spreadsheet university, expanding the sphere of my literary freedom. I’m naturally bad at making choices. I say I want to read loads of things, but in the moment a choice needs to be made I find it debilitating. If my choice of reading material were entirely my own, I’d probably just pick up another John Grisham novel. (I have much the same problem with Netflix and Stan, and end up watching Blue Heelers.) Yet in the 15 years Bill Gates’s creation has done the choosing for me, I’ve read wonderful books I never would otherwise have opened. The spreadsheet has also allowed me to begin to address my startling lack of reading diversity. The proportion of books I read by women has increased 6 percentage points since mid 2013, and those by authors of colour has increased fourfold.

The spreadsheet has also permitted me to indulge in a category I used to call “Crap” (because much of it is) and now call “Favourites”. It’s full of movie and TV novelisations, superhero prose, cricket biographies and kids’ books. Before the spreadsheet era (BSE), I didn’t read these books. It was a slippery slope thing: if I let myself read one Goosebumps story or Steve Waugh tour diary, there’d be no reason not to read another 10. The spreadsheet throws up a “Favourite” item as every fifth selection. The balance seems right. (The category also now includes genuine favourites I’ve already read, only allowed back into the running after a gap of five years, which allows me sufficient time to forget enough of their plot.)

After a while, the other person with voting rights – my partner – noticed another problem. She’d suggest a book to me and I’d say, “I’ll add it to my list,” which meant there was only a one-in-five chance I’d actually read it before I die (assuming I stopped adding more titles to the list, which I won’t). This led to arguments. So, there’s now a mini-list of my partner’s recommendations, and I’ve tweaked the random number generator in a way that means her recommendations have at least a realistic chance of being selected: she’s provided three out of the spreadsheet’s last 56 selections, which she seems relatively pleased about.

In this age of the quirk and bespoke, people seem to appreciate my spreadsheet’s nerdy uniqueness, even if they struggle to comprehend the discipline it appears to require. “Don’t you ever just ignore the spreadsheet and read a book you really want to read?” is a frequent query. I know I must have been tempted in the early days, but I can’t really remember it happening. Anyway, “Really Want to Read” is a category. My spreadsheet is now a habit. I always accept its selection: to contemplate not doing so feels disturbingly cataclysmic. But like a kind of electronic gym, it produces its own rewards. I get a rush of anticipation when it’s time to discover the next title it has in store for me. (Sometimes I run practice random generations just to see what the spreadsheet would have chosen had it been the time to do so.) Occasionally I’m temporarily disappointed, especially given my self-imposed requirement to complete each book I start (though sometimes there’s quite a lot of skimming). But like the Hogwarts Sorting Hat, the spreadsheet knows best. After all, it was me who put each title on the list in the first place.

I was something of a compulsive book-buyer before the spreadsheet. I think my bookshelf used to function as a “to-read” list, which was a terrible idea. I’ve moved house 20 times. I don’t know what I could have done with all the time I’ve spent packing books into boxes and then unpacking them: learn a musical instrument, maybe. (Or, more likely, simply add more titles to my “to-read” list.) But soon after starting the spreadsheet I noticed I didn’t feel the need to buy every book I walked past. I got the same hit merely by adding its title to my list. Now I only buy books I really want to keep, and I simply take photos of books on shelves in shops and libraries so that I can remember to add them to my list later.


There was no sound reason to stop just at books. Some kindred spirit called Mike has created lists of every major-publisher comic ever published, and makes them downloadable for free from his website, mikesamazingworld.com. I downloaded them, added some bespoke categories and weightings, imported my cycle concept and now read a few pages of a Phantom or Spiderman comic before falling asleep. I used to run a variation on this concept for Australian films: I made a list of every one ever made and used a random number function to tell me which one to watch next. And last year I noticed that the collection of songs I carry around as MP3s on my phone hadn’t changed much since 2004 (with the exception of Paul Kelly’s releases). So, I pulled a bunch of “best albums” lists off the net, created some categories and a cycle, and suddenly I’m making musical discoveries for the first time since uni and wondering how on earth I’d missed out on Los Lobos. Spotify is great, but I only use it to listen to albums my spreadsheet has already selected. There’s a lot that’s not on Spotify, and its algorithms tend to herd our choices rather than free them.

Meanwhile, I’ve loaded a 512-gigabyte memory card with podcasts, songs, albums, audiobooks and MP3 versions of my favourite sitcoms (most of them work really well without pictures), and play them at random on my phone using a remarkably geeky little app called MusicFolderPlayer.

What problem does all this randomness solve? Many small ones, I suppose, all connected to the abundance of cultural choice for the 21st-century middle-class Australian. Which is to say, no problem at all.

Russell Marks

Russell Marks is a lawyer and an adjunct research fellow at La Trobe University. He is the author of Crime and Punishment: Offenders and Victims in a Broken Justice System (Black Inc., 2015). 

From the front page

Illustration by Jeff Fisher

OnlyFans and the adults in the room

The emerging OnlyFans community offering training and support to adult-content creators

Image of US President Joe Biden meeting virtually with Chinese President Xi Jinping from the Roosevelt Room of the White House, November 15, 2021. Image © Susan Walsh / AP Photo

The avoidable war

Kevin Rudd on China, the US and the forces of history

cartoon:In light of recent events

In light of recent events

Who’s preferencing whom?

Detail of cover of Simon Tedeschi’s ‘Fugitive’

Ghost notes: Simon Tedeschi’s ‘Fugitive’

A virtuoso memoir of music and trauma, and his experiences as a child prodigy, from the acclaimed Australian pianist

In This Issue

Image of The Sea of Hands, representing support for reconciliation and the rights of Indigenous Australians

The truth about truth-telling

Revisiting trauma is not the road to justice for Aboriginal people

Image of Gerald Murnane

Final sentence: Gerald Murnane’s ‘Last Letter to a Reader’

The essay anthology that will be the final book from one of Australia’s most idiosyncratic authors

Illustration by Jeff Fisher

Did Federation compromise our democracy?

How the advent of Australia’s national government separated power from people and place

Image of The Kid Laroi

New kid on the block: The Kid Laroi

How Australia has overlooked its biggest global music star, an Indigenous hip-hop prodigy


Online exclusives

Image of US President Joe Biden meeting virtually with Chinese President Xi Jinping from the Roosevelt Room of the White House, November 15, 2021. Image © Susan Walsh / AP Photo

The avoidable war

Kevin Rudd on China, the US and the forces of history

Composite image of Prime Minister Scott Morrison and Opposition Leader Anthony Albanese speaking during the first leaders’ debate on April 20, 2022. Image © Jason Edwards / AAP Images

Election special: Who should you vote for?

Undecided about who to vote for in the upcoming federal election? Take our quiz to find out your least-worst option!

Image of the Stone of Remembrance at the Australian War Memorial, Canberra. Image © Lukas Coch / AAP Images

Remembrance or forgetting?

The Australian War Memorial and the Great Australian Silence

Image of Opposition Leader Anthony Albanese, Labor MP Emma McBride and shadow housing minister Jason Clare after meeting with young renter Lydia Pulley during a visit to her home in Gosford on May 3, 2022. Image © Lukas Coch / AAP Images

Property damage

What will it take for Australia to fix the affordable housing crisis?