Bullshit Theory

Fellow Lunartiks!

Welcome to the very first Ideas newsletter from Full Moon.

Every half moon — that means twice a month — we’ll send you an email like this: a mixture of news, insight, and provocation. All intended to help you make sense of the world, and do work that matters.

Before we get to the first, a quick message: thanks for joining us on the adventure that lies ahead. We’re just getting started, and your early enthusiasm means a lot.

A reminder: Full Moon is a new research service from us — Mark Curtis and David Mattin — at the intersection of technology, business, and creativity.

On launch last week, we shared our debut essay from Mark: Where is Design Heading? And next week we’ll send you our first podcast, in which we’ll dive into the essay and other matters at hand.

Paid members will be able to read Part II of Mark’s essay on the night of the January full moon. And from there, they’ll receive a deep dive essay and podcast every month. Free readers, meanwhile, will receive essay and podcast previews, along with this Ideas newsletter.

We are, as you’d expect, excited to get going. We have big plans for this place. We’ll tell you more about all that in the podcast next week.

In the meantime, let’s get to the first Idea. This week it’s from Mark, and it’s about a phrase that has gained much currency over the last few years: Bullshit Jobs.

If you find this a useful read, please do forward it to a few friends and colleagues, and let them know about our band of Lunartiks.

Bullshit Theory

Recently I heard someone talking about “BS jobs” on the radio, and realised that the terminology is going somewhat media mainstream. Just in case you don’t know – BS is short for bullshit.

What is a Bullshit Job, exactly?

The term comes from a 2018 book by David Graeber called Bullshit Jobs. The Wikipedia summary says that the book:

“postulates the existence of meaningless jobs and analyzes their societal harm. Graeber contends that over half of societal work is pointless, and becomes psychologically destructive when paired with a work ethic that associates work with self-worth. He describes five types of meaningless jobs, in which workers pretend their role is not as pointless or harmful as they know it to be: flunkies, goons, duct tapers, box tickers, and taskmasters.”

Recently there has been an upsurge in commentary from notable tech figures who seem to regard the elimination of bullshit jobs as a social good, (and thereby a justification for the brave new AI future). Sam Altman has taken this line in an interview.

“The thing about that farmer (from 50 years ago)… [is that] they very likely would look at what you do or I do and say, ‘that’s not real work.’”

He continued, “If you’re… farming… you’re doing something people really need. You're making them food, you're keeping them alive. This is real work. You people of the future, life just got too easy for you.”

AI may take your job, but, Altman says, "I think we'll find plenty of things to do."

Another writer — with a different agenda — talks about The Pandemic of Fake Jobs.

My take on all this? We need to be very careful here. Bullshit is in the eye of the beholder. Graeber's use of the phrase is highly judgemental and subjective — check the derogatory terminology he invented.

At worst, many of us look at other people’s jobs — except maybe for farmers, teachers and firemen — and believe they are probably pointless. It’s maybe a natural but sad human point of view, driven by lack of understanding and context, tribalism, prejudice, sometimes envy. Highly successful technologists who like things in black and white can be prone to it.

Natural, but obnoxious. The BS Job label is too dismissive of other people who are largely doing their best to prosper in a complex economy and society. If a role is pointless, history suggests that over time it will wither and die: think arrowmakers, blacksmiths, pool typists and so on. If a role continues to exist, perhaps this is a signal that it still carries some kind of value? Is taxi driving a bullshit job? As a category, it may disappear over time with robotaxis, but we really should give it the dignity of meaning right now.

The technology writer Cory Doctorow is also critical of the way the idea is being used as ideological cover for the AI elimination of some jobs. He says

“I don’t think AI can do your job (but I do think an AI salesman can convince your boss to fire you and replace you with an AI that can’t do your job)”.

Doctorow also makes a different but related point: that poor implementation of AI leads to “enshittification” of low value roles. Which takes us to the main point about BS Jobs that I want to make here. It is this:

Bullshit Tasks

What do I mean by this? While I am deeply suspicious of the idea of Bullshit Jobs, Bullshit Tasks are everywhere.

They can include meetings we don’t really need to attend, reports that never get read, mails that cc too many people who lose time parsing if they are meaningful or not. We could make a really long list here. It is pretty well documented that the larger an organisation gets, the more bullshit tasks get invented. I’ve definitely seen this in action and often it masks risk avoidance, poorly disguised. For example this happens around how we seek “approvals” – the default is all too often to list everyone who may need to be consulted, which is always a very long list. And it only takes one of those people to suggest change to throw the whole process into spasm and loops. The well known RACI model has a lot to answer for when used to provide aircover rather than focus.

Shifting the focus from jobs to tasks depersonalises the debate, and allows for a really useful reframe when it comes to the role of AI in all this.

We should focus less on the AI-elimination of so-called Bullshit Jobs, and more on tapping AI’s power to automate BS Tasks.

The promise is that it reads mail, sorts the important stuff, does boring things fast, takes meeting notes. As usual with tech, the reality carries benefits and downsides. The productivity benefits of AI are enough to take ones breath away when we stand back and look at what we can now do compared with just four years ago. I'm seeing this daily on our Full Moon work already. Consumer uptake of AI for personal use is the fastest of any tech ever for good reason. The speed with which routine things can be done opens up more time for creativity, thought, the injection of fun (yeah why not?) to daily work.

However the risk is that all this creates more stuff we have to act on, not less. Recently, I chaired a panel at an event. One of the panellists, as an experiment, recorded it and asked ChatGPT to analyse what was said. The result was in some ways good (happily it scored us well), and insightful. I learned something from it. It also wanted more facts from the panellists, rather than illuminating stories and pithy opinions. In this it reminded me of the rigid teacher Gradgrind from Dickens’ Hard Times

"Now, what I want is, Facts. Teach these boys and girls nothing but Facts. Facts alone are wanted in life. Plant nothing else, and root out everything else.”

And its output was soooo very long. There was just too much detail to absorb. It was an experiment. What if someone decides that the evaluation and “lessons learned” have to be reviewed and acted upon after every meeting or event? It is only too easy to imagine that happening, systemically, in an overly bureaucratic organisation.

A piece in Quartz this week about Co-Pilot highlighted enthusiastic insistence on the use of AI in companies citing one worker faced with:

“a constant need to demonstrate AI usage, even when it slowed her down. Copilot made her workload heavier, not lighter. Under pressure to use Copilot to write emails, she used it to generate a first draft, then edited out its most annoying trademarks — passive voice, bullet lists, upbeat platitudes — a rewrite process that consumed more time than a simple, AI-free writing session. Ironically, her manager, intent on having everyone use Copilot, returned her emails rewritten by Copilot, re-adding the hallmarks the trainer laboriously removed and reminding her to please use Copilot.”

That’s the third level of BS — Bullshit AI. It is all too easy now to produce reports, then tasks arising, which together constitute workslop: not just the content but the demand for busyness. Not a surprise that the Economist's Word of the Year is “slop”.

To be fair it is humans, not the technology, who will turn slop into BS tasks. To avoid doing that we just have to be a lot more judgemental about what tasks we use AI for, and its output, and less judgemental about each other. Unlike a human, the AI won’t mind if you call it Bullshit.