Ted Sandyman, AI Booster
I am currently reading Jathan Sadowski's brilliant book The Mechanic and the Luddite. Any summary I offer will be less pithy than the tag line – "A Ruthless Criticism of Technology and Capitalism".
In his chapter on labour, Sadowski touches on the latest hot topic in software – automation through so-called AI – which he calls the latest attempt of capital at creating a perpetual value machine. A perpetual value machine is a machine that creates value without the need of labour, a dream of capitalists. A dream that Sadowski claims is impossible.
To explain why this endeavour is doomed, Sadowski discusses Marx's concept of "living" and "dead" labour. Living labour is human workers, and dead labour is machines and tools. Marx claims that dead labour can only produce value when constantly tended to by living labour. Dead labour must be maintained and monitored by living labour, or it remains inert. Dead labour does not produce value, but acts as a multiplier on the potential value of living labour. The inescapable need for living labour is why a perpetual value machine is not possible.1 Value comes from humans using tools, not the tools. Even in the pipe-dream of AGI (Artificial General Intelligence) a tool must morph into a human before it produces value by itself.
The idea of a worker serving a tool reminded me of Cory Doctorow's metaphor of centaurs. A centaur is a human mind equipped with a machine body, so that a human is empowered to do more. Living labour is at the forefront, choosing to use dead labour as a means to an end. Doctorow also defines a reverse centaur as a machine mind with a human body – a human performing tasks that a machine mind cannot. This chimes well with Marx's view of the living serving the dead.
The distinction between the centaur and its reverse is agency. A centaur has agency to determine how it will use its tools. A reverse centaur has no agency, merely doing what the tool cannot. In either case value is produced by the living labour, but the experience of work is entirely different.
Due to this heady mix of Marxist thinking on industrial labour and mythical creatures, my mind naturally2 turned to The Scouring of the Shire, the penultimate chapter of J. R. R. Tolkien's The Lord of the Rings. While the protagonists have been away on their adventure, the Shire has undergone a radical transformation from a pastoral idyll to an industrialised and spoiled land. It transpires that one rich hobbit, Lotho "Pimple" Sackville-Baggins, has bought much of the Shire and ordered it as he will.
In Lotho we see the driving force of capital accumulation. Out of a desire to "own everything" he hired a paramilitary force of mercenaries, performed a coup on the Shire's (minimal) leadership, and industrialised the area in order to accumulate more wealth. The wellbeing of his neighbours and the environment are externalities that can be ignored, rendered trivial by the overbearing desire to possess more.
Most of the hobbits are discontented with living in an oppressive state, but are too afraid of the mercenary force to offer any organised resistance. There is, however, one hobbit we meet who is completely content with the new regime: the miller, Ted Sandyman. As Farmer Cotton puts it:
Take Sandyman's mill now. Pimple knocked it down almost as soon as he came to Bag End. Then he brought in a lot o' dirty-looking Men to build a bigger one and fill it full o' wheels and outlandish contraptions. Only that fool Ted was pleased by that, and he works there cleaning wheels for the Men, where his dad was the Miller and his own master. Pimple's idea was to grind more and faster, or so he said.
Sandyman has not only transitioned from a centaur to the reverse, he has done so willingly, even joyously. His skills, the dignity of ordering his own work, even the family mill – all can be sacrificed for increased production. Sandyman sees productivity as the ultimate goal: the more flour, the better. He expresses this when he taunts Sam Gamgee over the Shire's despoliation.
Don't 'ee like it Sam? But you always was soft. […] We've work to do in the Shire now.
Sandyman views those who don't share his veneration for production as fantasists. A philosophy of work which has values beyond the pure quantity of output is seen as simply being workshy.
In "reality", Sam Gamgee is far from workshy. Not only is he a diligent gardener, but he is newly returned from a stint as a courier that almost cost him his life! Even in Mordor Sam's escapist fantasy from grim reality is a "day's work in the garden". Sam and Sandyman differ in how they value work. Sam sees value in the work itself, in the process of gardening, and in the joy the product brings him and his fellow hobbits. For Sandyman nought matters but the quantity of grist the mill grinds to flour, regardless of its utility or cost to society.
Sandyman reminds me of AI boosters who happily embrace LLMs and agentic workflows in software (and wider knowledge work). Many have stuffed their mills with these outlandish contraptions, and now tend to the machines rather than running the mill3. The justification is often the same as Sandyman's: productivity.
Boosters see more output made faster as a good in itself, and all can be sacrificed in that name. Just as the Shire had trees chopped down and hobbit holes destroyed to make way for industrialised mills; LLMs require land and homes be given up for data centres. Just as Sandyman surrenders the running of his mill to clean machinery, so too do the boosters give up writing code to supervise an LLM. Sandyman and the boosters both see this as good, despite their work requiring less thought and creativity. A miller may be difficult to replace, but the skills needed to clean machinery can be found more easily.
Sandyman also haunts the rhetoric boosters often use, speaking of how those who don't subscribe to their vision of the future will be "left behind". By not becoming an "AI"-powered 10x engineer, a sceptic is slacking off, unlike the diligent boosters. There's work to do in the Shire now.
There's been a lot of noise lately about whether "AI" has given any return on investment, especially since businesses are no longer paying subsidised rates for LLMs and are staring into the abyss of their immense costs. Increased production alone is not a good. The things being produced need to have a purpose, and tools need raw material to work on. As Farmer Cotton says:
[…] you've got to have grist before you can grind; and there was no more for the new mill to do than for the old.
I think this is (part of) the reason all this busyness has been ineffectual – so much investment and energy used up for a dearth of meaningful results. Without the grist of problems worth solving, what use is the agentic mill?
I side with Sam Gamgee. Sam wouldn't use a contraption that performs all his gardening because he likes the work. I have likewise never used an LLM, or an agent, or any other "AI" tool. Programming is something I enjoy – something that I look forward to even when I'm in Mordor. Why would I give that up to clean Claude's wheels?
Footnotes:
Disclaimer: I have not yet read Marx's Capital, so you're probably better off reading Sadowski directly on this.
Indeed my mind is never far from straying back to Middle Earth.
I refer here to those who "no longer write even a line of code", not those who may occasionally use LLMs as tools for small tasks. I'm sure they can be useful tools for some ends.