Editing by the numbers

The most prolific editor in the world is an unknown genius

Source: Metal Letterpress from Etsy

The publishing industry was built on the legends of great Editors. The best had two things in common. They were laser focused on what their audience wanted, and they had a clear (and often profane) vision of how to commission, curate and package it to keep those readers engaged.

The best editors when I started out were forceful and demanding characters; tough journalists like John Witherow, Rebekah Brooks, Paul Dacre in the UK, Marty Barron, Anna Wintour in the US. They inspired their newsrooms through sheer force of will to get better stories than the opposition.

In the print era success was measured by paid circulation. So the Great Curators were obsessed by a single North Star target, their monthly sales. Advertising revenue was the cream on top, but selling it wasn’t their job.

But a new Editor came through in the early 2000’s, and he would become the preeminent curator of this era, without becoming a household name. He has not been played in the movies by Leib Schreiber or Meryl Streep. Thera are no journalists who tell stories of how they learnt their craft at his knee, even though he curates more stories each day than the others did in their entire careers combined.

His name is Muhammad ibn Mūsā al-Khwārizmī.

His journey to the top has taken a long time. Around 830 AD, Muhammad had written a book on algebra that used Hindu-Arabic numbers running from one to 10 to solve Western mathematical problems. He was the first to use the decimal system to solve equations in a systematic way.

Translated to Latin, al-Khwārizmī’s book was titled ‘Algorithmi de Numero Indorum’, and the word algorithm was born.

We now use it to talk about automated mathematical calculations that take a number of inputs and calculate an action based on a desired outcome it has been trained to produce.

The algorithms used to edit your social media screens are incredibly sophisticated. They take behavioural inputs at an individual level (my previous content consumption, how long since I was on the platform) plus content inputs (what content is queued from things I follow, what is trending amongst my connections) plus content formats (video first) and so on.

Content is scored and the highest ranked content is served in the feed on a rolling and everlasting basis. Meta have been quoted as saying over 10,000 signals are taken into account to select the News Feed, and at any given minute there are thousands of articles clamouring to get selected just for my feed.

So what drives curation in the al-Khwārizmī newsroom?

Success for most digital platforms is advertising revenue, the best leading indicator for which is user engagement; the more time we spend on the platform reading / watching / sharing / liking then the more ads can be served to us and our networks at the highest price.

Muhammad’s models are relentlessly tuned by Growth teams that know that there are very few new customers left to acquire. Growth is now about growing time spent while increasing ad load and pricing.

One way to increase time spent for each user is if you aren’t sharing it with any other platform. Hence we have seen a gradual downgrading of content that contains links out to other sites, as the model is tuned to keep users in. This is a skilful and deliberate application of the editorial commissioning policy.

The coming example is opened up by Generative AI. That has allowed the al-Khwārizmī acolytes at Meta to ask foundational questions about the direction of content commissioning.

If you have built a sophisticated specification of what every individual user will spend time on, why wait for third parties to send in something that scores high enough to match your requirement.. why not just send a prompt requesting your perfect content to your internal generative AI tool and automate the creation process on demand?

This type of content will rank highly because the prompts that create it are born from the real time ranking of what ranks highly - it’s a closed loop system. Meta are already creating AI avatars to front it, essentially building an in-house team that commissions and creates content to order, delivered by users injected into the population.

Their killer advantage is not so much the core content, but the precision of all of the 1% factors that make posts contagious, plus the sheer unmatchable speed with which it can react to user sentiment and cultural moments. Muhammad’s newsroom will create content that surfs trends while third party creators are still scanning the horizon for ripples.

I have heard this dismissed as “AI Slop”, and seen the argument that users will rise above it, that they will seek out ‘better’ content elsewhere.

I’m skeptical. Few will argue that content quality has steadily risen in the past ten years, but time spent on mobiles certainly has. History suggests most users quite happily will take the machine methadone in the absence of something better to do.

What does this mean for professional publishers? Well many of them seem to be planning on the basis that they should;

• Accept that in the medium term they cannot compete in social content on any of cost or speed or customer insight.

• Accept that if platforms are increasingly self serving and customer keeping,

  occasional hits on differentiated content will really be working to generate ad revenue for someone other than them. At that point they are just another thirsty creator hoping for selection, not a publisher using the platforms for traffic acquisition.

• As social closes off, search becomes even more critical for discovery. Publishers must have a plan to take advantage of the biggest disruption in search for twenty years which will play out this year (more on that in previous article here )

• This only works if the owned and operated products are engaging, so would expect more investment to be headed that way. AI will drive efficiency in production, but the winners will be those that combine good tech with a human centred product vision.

• This needs leadership and talent, so investment also is needed in teams that can combine engaging content with profitable business model execution.

The final and most important thing is to ask every day which editor they are following; the one they hired to focus on their audience, or the one al-Khwārizmī designed to sell ads?

Sources

Father of the algorithm Here

Old but Gold WaPo explainer on the FB algorithm process Here