Captains of industry

A merchant ship, carrying a mix of cargo on behalf of different business concerns, gets caught in a storm. It’s a bad one, and the ship, and all hands, are at risk. The captain must act. To save the crew, ship, and the bulk of its cargo, the captain orders that some cargo be thrown overboard. The ship becomes more manoeuvrable, and less low in the water, and it is saved. The storm passes, and it continues to its port, off-loading the remaining cargo at profit.

What happens to the merchant whose cargo was thrown into the waters to save the rest and the ship? Its loss has enabled the saving of the rest, and the profits that come from the (at least partially) successful voyage?

Since antiquity, the answer to this question has been the same, so it doesn’t matter if our ship is a Roman corbita, Portuguese carrack, Dutch fluyt, English tea clipper or a vessel today. The so-called ‘rule of general averages’ is applied, and the merchants that benefit from the success of the voyage need to share profits to make the cargo-loser whole. And by ‘whole’ that means the value of the goods at the destination, not on embarkation, i.e. giving the benefit of the successful voyage that their loss enabled.

The core of this rule is equity, or fairness: the voyage is a shared endeavour and the jeopardy is a shared one. Informally, it becomes a partnership where risks need to be shared fairly – though the sense of partnership is only triggered if the captain needs to take such a drastic decision. By granting the captain full discretion, the merchants allow him or her to take the decisions that give the voyage the best chance of overall success, and enable him or her to exercise professional judgement to deliver that result.

A fascinating academic article draws an analogy between this granting of discretion to someone who can exercise professional judgement over the success of an operation to the role of the board of directors and management of a company, and an analogy between the rule of general averages and the governance need to exercise judgement over how the fruits of business success are shared among the stakeholders who share in that endeavour. Called, perhaps inevitably, Captains of industry? Value allocation and the partnering eﬀect of managerial discretion, the article suggests that corporate law could benefit from more of a similar sense of equity (or fairness). Company directors have fiduciary duties to promote the success of the company (the English law formulation is acknowledged in most markets, and is a proper understanding of US law too – or at least how US law was historically understood). This implies a need to seek a fair balance of the interests of stakeholders, as that is what will promote long-term business success. The analogy comes largely in thinking through the equitable sharing of that success. The paper doesn’t argue that the general averages model should be adopted, but rather shows that it is a helpful reframing allowing us to see more clearly how modern corporations are often run. Essentially, it points out that modern management and boards often display a failure to be fair.

For the academics, the first benefit of the analogy is in thinking about any particular sacrifices that stakeholders may make which operate to the benefit of the company. This, they suggest, might give rise to a similar sense of quasi-partnership that the captain’s cargo sacrifice triggers in general averages. The example given is staff undertaking specialist training that is relevant only to that particular business, but it is also relevant to consider any time that workers spend that exceeds the expectations of their job description. Certainly, workers who do work beyond their contract (which is many) tend to feel that they have some right to better reward or treatment over time as a result of such actions. As the academics say: “the rule of general averages makes us realize that some stakeholders can be affected by decisions taken for collective purposes, without being considered as general averages. It thus questions whether these impacts are treated in a fair way.”

The core of the argument is that at least some stakeholders should be seen as corporate partners in the business endeavour, and that their interests need to be fairly considered and balanced alongside shareholder interests in setting strategy and delivering operationally. Thus: “our analogy calls for a solidarity rule to share the impacts of managerial decisions on ‘partnered stakeholders.’ In practice, this rule would require ex ante that any strategy designed by corporate leaders must achieve a balance between employees and shareholders. Management should also be required to report on how they integrate fairness into their strategic thinking.” One specific example that the authors posit is that workers laid off to preserve the company through difficult times might be given a financial instrument that enabled them to participate in the upside should the company not only recover but prosper in the future.

That discussion reads (to this reader at least) as though the authors are going a little further than their own strictures about the limits to the value of the general averages analogy that they themselves set – but it is all the more thought-provoking as a result.

There may well be aspects of this work that we should jettison, but it does seem to me that the overall cargo offered by the article’s thinking could be valuable if safely brought to harbour. Its active consideration of fairness is clearly attractive for this blog, but also the emphasis on businesses as effectively partnerships between stakeholders – certainly when they are successful this is how the best companies feel – and the need for boards to find a fair balance between the interests of different stakeholders are all elements that chime with earlier posts and with a sensible understanding of where long-term business success lies.

Captains of industry? Value allocation and the partnering eﬀect of managerial discretion, Segrestin Blanche, Armand Hatchuel, Ken Starkey, 2020. Management & Organizational History, 15:4

What’s fair pay for a bank?

It’s fair to say that generally banks aren’t well known for the fairness of their approach to pay. So it was interesting to note the emphasis that UBS, now by far Switzerland’s dominant bank following its 2023 rescue of the failing Credit Suisse, places on the issue of fair pay in its latest, recently issued annual report.

But the problem is that it seems to this reader at least that the bank may not necessarily be focusing on the right forms of fairness in pay.

For example, in the introductory section of the 43-page compensation report, the bank answers its own question, How does UBS support pay fairness?, as follows:

“We pay for performance, and we take pay equity seriously. Across all our locations, we apply the same fair pay standards, reinforced by annual reviews of our approach and policies in line with established equal pay methodologies. In 2025, our statistical pay gap analyses reaffirmed that pay differences between male and female employees in similar roles across our core financial hubs remained below 1%, a difference consistent with that for 2024. If we find any gaps not explained by business or by appropriate employee factors, such as role, responsibility, experience, performance or location, we look at the root causes and address them.”

UBS develops this analysis a little further in the Compensation philosophy and governance section, under the title Fair and equitable pay. Key elements of this discussion read:

“Pay equity and equal opportunity are fundamental to support our strategy. Being an employer of choice and inclusive of all experiences, perspectives and backgrounds is critical to our success. Factors such as gender, culture, race, ethnicity, sexual orientation and identity, disability, family, veteran status, generations and part-time status should not impact opportunities available to our employees.

“Fair and consistent pay practices are designed to ensure that employees are appropriately rewarded for their contribution.”

It again emphasises that gender pay gap analysis shows gaps below 1% in pay for male and female employees “in similar roles across our core financial hubs”, an interesting geographical narrowing and also clearly ignoring the general experience that the issues in gender pay arise most often because of differential opportunities for men and women, which tend to lead to a skewing of roles away from strict ‘similarity’. And note that it has a marked gender skew overall in staff, given that only 41% are female.

If we set aside these limitations, this discussion of fairness seems fine as far as it goes, but it barely begins to address the promise of the titles given to it. This is not UBS ‘supporting pay fairness’; this is simply the bank not overtly discriminating in its treatment of different staff – which in many countries and cases will in any case be illegal. The actual question of supporting pay fairness might acknowledge realities outside the bank, rather than this internal focus. The financialisation of our economies has helped fuel the broad inequalities that our world faces, as financial institutions, competing with each other for ‘talent’, bid up the pay of individuals well beyond what is affordable or realistic (even imaginable) for most businesses. The closest UBS comes to a proper understanding of what a broader mindset about fair pay, looking beyond the financial sector, might actually imply is this comment, which at least acknowledges some external benchmark:

“We also aim to ensure that all employees are paid at least a living wage. We regularly assess employees’ salaries against local living wages, using benchmarks defined by the Fair Wage Network. Our analysis in 2025 showed that employees’ salaries were at or above the respective benchmarks.”

A bank that was genuinely seeking to ‘support pay fairness’ might apply similar pay expectations – of that minimum living wage level pay – at all clients and counterparties. A bank that worked with its corporate customers to understand what local living wages might be in their countries of operation and how they might be achieved across relevant workforces while still delivering profitability would be a bank genuinely supporting this ambition.

A bank that was genuinely seeking to ‘support pay fairness’ might consider whether financialisation is leading to pay distortions between the financial sector and almost every other part of the economy.

As ever, the finance industry has a tendency to look inwards at itself and not outward at its role and influence in the real world. An article I reference a couple of times in my forthcoming book invites the industry to think about fairness ‘outside its cocoon’. Only if financial services starts to do that will it genuinely be supporting fairness.

Duncan Mavin, in his (highly recommended) book telling the sad history of UBS’s takeover target, Meltdown: Scandal, Sleaze and the Collapse of Credit Suisse, identifies an unfair approach to pay as being part of the drivers of that bank’s failure: “The behaviour of the bank’s leaders hardly inspired other staff to be the best versions of themselves … Credit Suisse bankers got paid well, whatever happened. When the bank was making a profit, staff made a bundle, regardless of whether the results were driven by strong markets or great management. When the bank made a loss – because of misconduct, fines, bad behaviour or poor strategic decisions – the bonuses were good then too.”

Inward-looking financial institutions are more likely to fail than ones that look outward and measure fairness against external benchmarks. Instead, they need to look beyond their comfortable cocoon.

Annual Report 2025, UBS

Fairness Outside the Cocoon, Meir Statman, Financial Analysts Journal, Vol 60, No 6

Meltdown: Scandal, Sleaze and the Collapse of Credit Suisse, Duncan Mavin, Pan Macmillan 2024

Surprising benefits of labour fairness

Having sent my draft book to my publisher – more news on that when I have it – I can re-devote my fairness writing energies to this blog. One side-effect of the book-writing process, and particularly of the editing process, is that I’ve much material that didn’t make it into the manuscript, at least not in full. This then is likely to be the first of a few (or more!) blogs that are based on materials from the cutting room floor.

Many take it as a given that treating employees well encourages them to deliver more, not least because workers value dignity. There’s been a formal economic theory to that effect since at least 1990, in the form of the Fair Wage-Effort Hypothesis put forward by an illustrious pair, future Nobel prize winner George Akerlof and future Treasury Secretary Janet Yellen. This says that workers treated fairly, and specifically paid fairly, put in greater effort; those treated unfairly don’t bother so much.

Sometimes particularly strong evidence is provided that backs the hypothesis, showing greater productivity of workers who are inspired by being treated fairly. In at least one case, that evidence shows the productivity effect not in terms of fairness facilitating the creation of higher numbers of goods, but in the way that unfair treatment led to the creation of products that were in fact lethal for customers.

That case dates back to the era in which the Fair Wage-Effort Hypothesis was developed: the Bridgestone/Firestone defective tyre issue of the 1990s, and the subsequent product recalls. Some 271 deaths and more than 500 injuries in the US alone have been attributed to the defective tyres. In the end, Bridgestone/Firestone’s costs of the scandal were over $800 million, and those for the other company implicated, Ford, more than $500 million. The detail of the research into the case – particularly a study with the surprising name for an academic paper of Strikes, Scabs, and Tread Separations – is such that it not only succeeds in identifying the implicated plant, but also with some precision the times when most defective tyres were produced. This precision allows us to attribute clearly the defects to failures of fairness in how workers were treated.

Japanese tyre manufacturer Bridgestone had bought US rival Firestone in 1988. Increasingly, it put pressure on Firestone’s loss-making operations. As pressure increased to reduce those losses, the focus came onto staff costs, perhaps inevitably giving rise to industrial unrest. There were three main facilities in North America; the one where the greatest industrial dispute arose was Decatur in Illinois. The Wilson, North Carolina plant wasn’t unionised and saw no strike, while the strike in Joliette, Quebec was brief and no replacement workers were hired there (not least because that’s illegal in Canada). At Decatur, tensions were much greater and longer-lived.

The key trigger to industrial unrest was the process of negotiating a new contract from 1994. The company was seeking to change from 8-hour to 12-hour shift patterns, and wanted to cut pay for newly hired workers by 30%. The prior contract expired in April 1994 and workers continued without a contract for three months before going on strike in July 1994. The company hired replacements, and refused to consider rehiring the strikers. A final contract, under which former strikers were in the end rehired, was agreed only in December 1996.

The issues with the tyres in question were associated with their role on the first Ford Explorer SUVs. In moves that to modern ears have echoes of Boeing’s failed decision-making in the development of the 737 Max, Ford didn’t design the Explorer from scratch but put a new body on an existing pickup chassis. However, the new body was heavier and so the Explorer had a higher centre of gravity and was more prone to roll in an accident. Avoiding a redesign, Ford chose to shorten the suspension and lowered the recommended tyre pressure. Such lower tyre pressure can increase the operating temperature of the tyres, making any faults more likely to cause blowouts or other less catastrophic forms of tyre failure.

No one fault in the tyres has been identified that made them particularly prone to failure, so there was probably a range of possible defects. While the tyres in question were manufactured at all of the three North American facilities, the evidence is that the bulk of the tyres that failed came from Decatur.

In particular, the Strikes, Scabs, and Tread Separations study of the timing and sourcing of the failed tires shows a clear linkage between the labour dispute – and particularly the most contentious moments in the dispute, those that will have generated the greatness sense of unfairness for the workforce – and the production of the substandard tires. Workers that feel themselves to be being treated unfairly are likely to take less care over their work. In some cases, the treatment is seen as so unfair and the lack of care is so significant that there are major quality problems. The study concludes simply that: “the strike and associated labor strife in Decatur was a major contributing factor to the production of defective tires”.

In more detail, it says: “Four years after production, tires that were made in Decatur during the labor dispute were at least 15 times more likely to have resulted in a claim than were tires manufactured in other plants.” As the chart from the research shows, in internal engineering testing to destruction, the top quartile Decatur tires from the relevant period performed at the same level as bottom quartile tires from the other factories.

The evidence doesn’t suggest that the replacement workers were more prone to error (as some commentators alleged), but rather that periods when union members operated alongside replacement workers before the final contract was agreed were more likely to see defective tires produced. Similarly, defective tires were produced in much higher quantities in the months leading up to the old contract coming to an end, when the company was pressurising workers for concessions. These were the moments of greatest industrial tension, the greatest sense of failures of fairness.

Simply, it was the poor industrial relations, the unfairness in the dealings between company and its workforce, that seems to have made production catastrophically less reliable. It turns out that the Fair Wage-Effort hypothesis came to pass in a very direct way in terms of quality and safety outcomes for customers. A Fair Treatment-Production Quality hypothesis perhaps.

I am happy to confirm as ever that the Sense of Fairness blog is a purely personal endeavour

The Fair Wage-Effort Hypothesis and Unemployment, George Akerlof and Janet Yellen, The Quarterly Journal of Economics, Vol. 105, No. 2 (1990)

Strikes, Scabs, and Tread Separations: Labor Strife and the Production of Defective Bridgestone/Firestone Tires, Alan Krueger, Alexandre Mas, Journal of Political Economy, Vol 112, No 2 (2004)

Playing fair in detective fiction

A new novel called Fair Play was of course irresistible for this blog – especially when it turned out that the author, debut novelist Louise Hegarty, was speaking at my local bookshop.

Fair Play is an intriguing, even odd, blend. It starts as a modern story of a group of friends at a New Year’s party in an Airbnb country house. When one of the party dies mysteriously, it fractures into two: a locked-room detective story of the traditional sort, aping Agatha Christie, Arthur Conan Doyle, Dorothy L Sayers and so on, and the story of the grief of the dead man’s sister, Abigail, as she tries to make sense of what makes no sense.

The title of the book comes from what Hegarty has gathered as the ‘Fair Play Rules’ of detective fiction, three sets of guidance from its heyday of the late 1920s. Essentially, these are about being fair to the reader, so that we feel we have a fair chance of finding the truth, just as much as the detective, and at least we are smarter than the detective’s foil, the person Father Knox in one of the three sets of guidance, his 1929 Introduction to The Best Detective Stories of 1928-29, cruelly calls “The stupid friend of the detective, the Watson”.

It’s not hard to think of especially famous examples that breach some of the ‘Twenty Rules for Writing Detective Stories’ by SS Van Dine in The American Magazine in September 1928, for example numbers 12 and 13:

“There must be but one culprit, no matter how many murders are committed”; and
“Secret societies, camorras, mafias et al have no place in a detective story”

But then, some of the best writers have always played with the genre, on occasions bending the rules, only to the greater pleasure of the reader. The sleights of hand to pass quickly over clues after having brought them to our notice is one of the joys of these books (even if we only spot them after the fact!).

It’s not just my admiration for him as a writer that means my favourite of the three sets of guidance is that from TS Eliot, in a 1927 New Criterion piece in ‘Homage to Wilkie Collins’, widely seen to have invented the detective novel. The fifth of these is that “The detective should be highly intelligent but not superhuman. We should be able to follow his inferences and almost, but not quite, make them with him.” While being superhuman is not allowed, odd habits are: Hercule Poirot’s fastidiousness was clearly invented to hide (in plain sight) a decisive clue from the reader. It’s likely that same fastidiousness is what eventually came to annoy Christie so much about her character.

The key point of each of the guides is that the reader of a detective novel must emerge from the story not feeling cheated but feeling we could have got the answer for ourselves if only we had paid more attention, thought a little harder or taken a little more time to mull the clues available to us. We wouldn’t recommend to others a detective novel that was unfair by failing to live up to these expectations, and we probably wouldn’t read more from that writer. In reading, as in other things, humans favour fairness.

It needed no confirmation that Hegarty is a fan of detective stories, and of Christie in particular. Her handling of various of the standards of the genre, and particularly the playful repeated versions of different revelations of alternative murderers, show that very clearly. But in many ways the half of the book that is the detective novel is slight. What elevates Fair Play, and makes it linger in my mind, is the half that is the story of the sister’s grief. This, sparsely told and without easy answers, carries heft without being heavy.

The thing that seemed off limits at the talk was Hegarty’s own experience of grief. But it appears clear from the way she writes of it that she knows whereof she writes. We all have our experiences of grief, and this writing rings true, if anything helped by its sparseness. There are different short vignettes giving a vivid expression. Some of these are jarringly within a workplace setting, memorably a dull business meeting that comes to echo only with the words ‘my brother’s dead’. This feels very real, as does Abigail’s search for the answer, the simple revelation that will give her resolution. But life rarely offers the simple resolutions of a brilliant detective’s summing up.

Though there are no easy answers, I never felt cheated by what is an admirable first novel.

I am happy to confirm as ever that the Sense of Fairness blog is a purely personal endeavour. I am also happy to wish readers the compliments of the season.

Louise Hegarty, 2025. Fair Play, Picador

Father Knox, 1929. Introduction to The Best Detective Stories of 1928-29, Faber

SS Van Dine, 1928. Twenty Rules for Writing Detective Stories, The American Magazine

TS Eliot, 1927. Homage to Wilkie Collins: An omnibus review of nine mystery novels, New Criterion

Bubbles and economic fragilities

With all the talk about a bubble in investment in so-called ‘AI’*, I have taken a moment to reread the classic on the 1929 Wall Street bubble bursting, US economist John Kenneth Galbraith’s The Great Crash 1929, first published in the 1950s.

Speculative market bubbles do come and go. As Galbraith notes, the reason that 1929 is most remembered is not so much that the speculative bubble had grown so large before bursting (though it was unusually large) but that there were such broad real economy impacts from the bursting of the bubble – the lost years known as the Great Depression. A pair of Galbraith datapoints start to capture the scale of the Great Depression in the US and its searing impact on ordinary people: unemployment in 1933 was 13 million, one in four of the labour force; and even in 1938 still one in five were out of work.

So trying to understand why the Great Crash sparked the Great Depression is of real interest and seems like a timely thing to consider. Galbraith has no patience for the Wall Street apologists who argue there was no connection between Crash and Depression, but he does see that there were vulnerabilities in the economy that made it particularly susceptible to the crisis.

The first of these is of most interest to this blog, and Galbraith headlines it ‘The bad distribution of income’, noting that in 1929 the top 5% received around 35% of all personal income, and that interest, dividends and rental income (the almost exclusive preserve of the wealthy) represented fully 22% of total family income.

“This highly unequal income distribution meant that the economy was dependent on a high level of investment or a high level of luxury consumer spending or both. The rich cannot buy great quantities of bread. If they are to dispose of what they receive it must be on luxuries or by way of investment in new plants and new projects. Both investment and luxury spending are subject, inevitably, to more erratic influences and to wider fluctuations than the bread and rent outlays of the $25-week workman. This high-bracket spending and investment was especially susceptible, one may assume, to the crushing news from the stock market in October 1929.”

Readers will recognise some of our current distortions in these reports of the imbalances of pre-crash 1929 (also see Is enough enough? Addressing the problem of the super-rich, and The centre cannot hold). Significant inequalities – especially unfair ones – make economies less robust, more risky and more prone to crisis.

That is the first of Galbraith’s linkages between Great Crash and Great Depression. The others are as follows (deploying my brief characterisations of his comments):

Bad corporate structure. A business sector including many swindlers and fraudsters.
Bad lending. Profligate lending to unsound businesses and investments.
Imbalanced trade positions. Long-term trading imbalances, sometimes exacerbated through the application of tariffs, with the resulting deficits sometimes filled by corrupt, or at least grey, payments.
Poor economic insight. Shonky economic data riddled with holes.

So clearly, we’ve nothing to worry about now.

I will leave Galbraith with the last words of this blogpost, without comment from me. In the last pages of the book, he writes: “during the next boom some newly rediscovered virtuosity of the free enterprise system will be cited. It will be pointed out that people are justified in paying the present prices – indeed, almost any price – to have an equity position in the system. Among the first to accept these rationalizations will be some of those responsible for invoking the controls. The newspapers, some of them, will agree and speak harshly of those who think action might be in order. They will be called men of little faith.”

* Seasoned readers may remember that I am an ‘AI’ sceptic – see A just AI Transition, for example

I am happy to confirm as ever that the Sense of Fairness blog is a purely personal endeavour – and also that I do not give, and am not authorised to give, personal financial advice. This blogpost should not be construed as such advice.

John Kenneth Galbraith, The Great Crash 1929. Hamish Hamilton 1955

The power of powerlessness

From this week’s Economist: how our sense of fairness means that we favour the underdog, and so makes forms of protest that seem powerless, powerful.

[if anyone is missing my blog, don’t worry! My fairness writing energies are largely being taken up by a book commitment; I’ll return re-energised to the blog next year, and you’ll also be able to read a book version of some of my Sense of Fairness thoughts]

Workers value dignity

It shouldn’t really be a surprise, but dignity at work – a combination of things such as the sense of autonomy and relationships with colleagues and bosses, and being treated fairly – matters to people. It’s as true at the bottom of the income scale, where observers might assume concerns about pay outweigh all other considerations, as it is higher up. Dignity matters to people, as I’ve been exploring in recent blogs.

For the book I am writing (on fairness in business and investment) I am currently investigating the literature on monopsony and oligopsony in labour markets. Monopsony is the distorted market situation arising from there being a single buyer of a good or service (a monopoly is where there’s a single seller); oligopsony is where there is a narrow enough group of buyers that they distort the marketplace. Economists are increasingly observing evidence that the labour market suffers inefficiencies that are consistent with oligopsony – employers having excess power in setting pay. Most workers would probably agree that their experiences too are consistent with this.

One part of this literature particularly stood out because it made a link to the issue of dignity, which increasingly seems a key element of people’s innate sense of fairness, and of their inclusion in society and the economy. In particular, a 2022 paper from the US National Bureau of Economic Research, called Power and Dignity in the Low-Wage Labor Market: Theory and Evidence from Wal-Mart Workers, uses evidence from real interactions with US employees of the globe’s largest private sector employer to understand their views and test hypotheses against reality.

The results are striking.

The study included four sentences exploring the degree to which workers had a sense of dignity in their jobs (these sentences were developed based on prior interviews with Wal-Mart workers that sought to understand their experience in the workplace, as well as earlier academic work). The overarching question was Indicate to what extent the sentence describes the workplace of your job at Walmart, and each time respondents were offered four responses (Almost Always; Often; Sometimes; Never). The four sentences were:

You [have/had] the opportunity to express yourself while at work.
You [can/could] rely on your co-workers to help you with work.
Your supervisor [treats/treated] you with respect.
Your supervisor [treats/treated] everyone fairly.

And of these four measures of dignity, it appears to be fairness that matters most. Indeed, a lack of fair treatment by one’s boss is in essence the greatest determinant of likelihood of quitting a job in the study, with the obvious exception of pay (and of the availability of hours of work a week, which is a clear part of the pay equation for those paid on an hourly basis):

Consistently, the study confirms that fairness and dignity are powerful drivers of work satisfaction, and thus in willingness to stay with an employer.

As the study states:

“A natural question is whether firms can adjust the level of dignity at work. While immediate supervisors likely have the most discretion over workplace dignity, supervisors can be incentivized by higher-level managers to treat subordinates fairly and with respect, and workplace rules can be designed to allow opportunities for self-expression and co-worker support. While it may take time to alter workplace experiences, and agency costs might be considerable, the significant cross-store variation we document below suggests that managers have some control over the level of workplace dignity.”

Our bosses, and how they treat us, matter.

As well as enhancing people management, the authors raise the interesting challenge of whether improving the competitive context of the labour market is necessary to increase dignity in the workplace, the experience of fairness for workers:

“any effort to increase workplace amenities (including subjective experiences at low-wage jobs) may require policies that reduce monopsony power in the low-wage labor market. The high levels of labor market competition in the immediate post-COVID labor market may have given workers the opportunity to quit jobs that didn’t provide dignity. Whether this results in firms upgrading the subjective experience of work remains to be seen.”

I’m not sure that we’ve yet seen significant enhancements to workplace dignity and fairness, but perhaps we should continue to live in hope.

I am happy to confirm as ever that the Sense of Fairness blog is a purely personal endeavour

Power and Dignity in the Low-Wage Labor Market: Theory and Evidence from Wal-Mart Workers, Arindrajit Dube, Suresh Naidu, Adam Reich, NBER Working Paper No. 30441, September 2022

Belonging, not belongings

The modern mind profoundly misunderstands our ancient ancestors. Take the mystery of ancient stone circles. One of the oldest is the Ring of Brodgar on Orkney, probably a creation of the culture that later built Stonehenge. Most guidebooks assert that the 21 currently standing stones (8 of them re-erected between 1906 and 1908) are the remnants of a circle of 58 or 60 stones.

The trouble is, this common assertion just isn’t true. There is no sign that there were ever more than 40 standing stones in the Ring – the latest archaeology, using a range of techniques, cannot find evidence for more than 16 sockets for stones other than those still standing. The modern mind has extrapolated from the stones that remain and calculated how many there would be if there had once been a circle of that circumference made up of stones with similar gaps between them (and ignoring the fact that the gaps vary). But that wasn’t the point for the ancient mind. Our ancestors thought differently from us, sometimes in ways that are even odder to our modern eyes: some stones weren’t ever deeply planted in the ground and so may have fallen very quickly.

The point is that the Ring of Brodgar wasn’t a thing, it was a doing: for our ancestors it was the collective action that mattered, its creation, not so much the resulting thing. Brodgar was a community event that brought people together in a shared effort. The stones aren’t matching: they come from different spots around the islands (at least 7), presumably transported by different families the miles from their best quarry, and raised as part of a communal effort. The ditch around the stones was certainly dug in small patches that were then joined together, perhaps further evidence of the bonding process of each family doing its own portion of the work before they were brought together as a unified whole – overall a work requiring an estimated 80,000 person hours.

The Ring of Brodgar, both stones and gaps between them

It’s like Peru’s ancient Nazca lines, about which modern people, flying over them in aeroplanes, puzzle as to the creation of images that can only be seen from above in this way. Most archaeologists now agree that the Nazca images were ceremonial paths, walked by people on special days or for particular reasons. They were not meant to be seen but to be experienced, perhaps to be made anew. The Nazca lines too were a doing, not a thing.

I keep coming back to these thoughts as I read a thoughtful and thought-provoking recent book, Don’t Talk About Politics. Author Sarah Stein Lubrano notes that we fail to persuade each other on politics and are wasting time and energy in talking at – and past – people with whom we disagree. Her core point is that politics is a doing, it is something that we need to participate in rather than see as separate to us, as something about which we can debate while remaining dispassionate observers: “To think effectively about politics…is not only to think about action; it is also, by necessity, to take action. Good political reasoning cannot happen simply by reading and debating.”

She goes on:

“When it comes to politics, we’ve probably placed too much societal emphasis on words and too little on action. Many people would say that democracy is about whether people can make up their own minds and then vote. But this is, in fact, a fairly passive idea of political involvement…our collective common sense of what politics is has, I’d suggest, become hollowed out. It involves too little of what truly changes people, their priorities and their actions.”

Stein Lubrano’s focus is politics, but society too is a doing, and so is the economy. We need to be active participants in both, and dangers lie where many members of a population are not. And at present, too many people do not feel like active participants in either the society or the economy.

This thought process echoed for me across a recent discussion co-hosted by my friends at the Fairness Foundation. They’ll write more fully and directly on what was a discussion under the Chatham House rule, so I will stick with my reflections. The core of the conversation centred on the issues I touched on in An Inequality in Dignity, or the Dignity Deficit; one participant talked about there being a ‘crisis of belonging’ that we all need to seek to address actively.

We discussed the way in which many people currently feel profoundly uninvolved in the economy. It isn’t just that they own little and lack a stake in the economy; more, they feel that the economy is something that occurs elsewhere in which they are not engaged. They lack both belongings and belonging. The unfairness embedded in the current economy, where so many are have-nots, is profoundly alienating and destructive of human dignity. Similarly, a dignity deficit arises because people do not feel they have a place in society, that their voices are not heard and will not be heard, that they have no agency. They withdraw into fury and anger because they feel that they no longer have a place in which they are welcomed as belonging.

This is true in the economy and in society more broadly, and the tensions and disconnects it fuels are plain to be seen. Stein Lubrano devotes a whole chapter to the psychological side of these sentiments. She refers to it as social atrophy, a sense that everyone is becoming isolated, detaching themselves from society and community. She notes the disastrous consequences that arise. As she points out, this is a much more fundamental risk than the way it is usually discussed, as an individual crisis of loneliness; it is a broader communal fracturing. “We may…distrust society more not only because it no longer provides everyone with what they need, but also because, as we become more isolated, the suspicious and paranoia-inducing structures in our brains become more and more activated,” she argues.

If we are to address the dignity deficit, the crisis of belonging, we need to involve people, we need for them to feel that society and the economy (and yes, politics too) is something in which they have a role to play, that these are not just issues that happen elsewhere between other people, but rather that they can act and be engaged. That will require more physical venues for people to meet face to face, for society physically to thrive. It will require a democratisation of finance, so that ordinary people understand that they do literally have a stake in the economy and are economic actors in their own right – an understanding that their pension savings make them part-owners of assets and of companies around the world. This needs to be understood as giving them agency, the scope to be involved and to act, not simply to be onlookers at activity involving others.

It turns out that our ancestors were right: society is a doing, it isn’t about things. We need to get out of the trap of the modern mind, that emphasises the thing rather than the doing. We need less emphasis on belongings and more on belonging.

I am happy to confirm as ever that the Sense of Fairness blog is a purely personal endeavour

Landscapes Revealed: Geophysical survey in the heart of the Neolithic Orkney World Heritage Area 2002-2011, Amanda Brend, Nick Card, Jane Downes, Mark Edmonds, James Moore, Oxbow, 2020

Building the Great Stone Circles of the North, Colin Richards (ed), Windgather, 2013

Don’t Talk About Politics: How to Change 21st Century Minds, Sarah Stein Lubrano, Bloomsbury Continuum, 2025

What’s a fair use?

The media has reported that Meta (the Facebook, Instagram and WhatsApp company) has won a legal case on the use of copyrighted materials in training its AI models, that the use of copyright materials was a ‘fair use’. As often with the law, it’s a bit more complicated than that.

The case in question was Kadrey v Meta, and summary judgement was released last week (the judge, Vince Chhabria, deciding on the basis of arguments that the case did not need to go to jury trial because the plantiffs had not made a convincing case, enabling Meta to succeed in a call to dismiss it). The legal question at issue was whether the accepted abuse of copyrighted works in training AI amounts to a ‘fair use’. As well as considering fairness, the case opens a wider window on AI.

Before delving, I will note that I’ll continue to use the term AI, because it’s used in the case and the term is in general use for these emerging new technologies. But as both recent books The AI Con and AI Snake Oil (the two latest additions to my bookshelf) start off by making clear, there is no such single thing as AI. It is a catch-all term for a range of technologies – some of only very dubious effectiveness – and is really just a brand that is being deployed to raise (enormous amounts of) funding (two headlines from the Financial Times over this weekend cast light on the scale of this financing: Meta seeks $29 billion from private credit giants to fund AI data centres, and Nvidia insiders cash out $1 trillion worth of shares). The best known, and most used, of these new AI technologies are called large language models (LLMs), accurately described as stochastic parrots: models that simply put one word after another according to statistical models developed through their training.

Many legal systems favour the term fairness, and ‘fair use’ is a well-established concept in US law. The country’s Copyright Act (in 17 USC §107) clearly restricts fair use to usage “for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research”. It sets out four factors that should be considered in determining whether a given use is in fact fair:

1. the purpose and character of the use, including whether such use is of a commercial nature or is for non-profit educational purposes;
2. the nature of the copyrighted work;
3. the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
4. the effect of the use upon the potential market for or value of the copyrighted work.

Deciding what uses are fair is both a matter of law and of the specific facts, meaning that there are multiple cases that have considered these factors. The list of four factors is not exhaustive, but are assistants in reaching the overall conclusion. The fourth factor, whether the use risks substituting for the copyright materials in the marketplace, is generally seen to be the most important. Courts need to apply judgment and consideration is deciding on fair use; as ever, assessing fairness requires thought and judgment.

As judge Chhabria explains in his summary judgement:

“What copyright law cares about, above all else, is preserving the incentive for human beings to create artistic and scientific works. Therefore, it is generally illegal to copy protected works without permission. And the doctrine of “fair use,” which provides a defense to certain claims of copyright infringement, typically doesn’t apply to copying that will significantly diminish the ability of copyright holders to make money from their works.”

He is as rude as a judge ever gets about a fellow judge who reached a recent decision on a fair use case in relation to Anthropic, another AI firm (Order on Fair Use at 28, Bartz v Anthropic PBC, No. 24-cv-5417 (N.D. Cal. June 23, 2025), Dkt. No. 231). That judge was convinced by the argument that training AI was no different from – and had no more impact on the market for copyright products – than training schoolchildren to write. Chhabria says: “when it comes to market effects, using books to teach children to write is not remotely like using books to create a product that a single individual could employ to generate countless competing works with a miniscule fraction of the time and creativity it would otherwise take. This inapt analogy is not a basis for blowing off the most important factor in the fair use analysis.”

And surprisingly given his overall ruling, Chhabria is very clear that AI companies are breaching copyright law and are damaging the commercial market for copyrighted works. He seems very sure that AI companies fail at the fourth factor in assessing fair use: “by training generative AI models with copyrighted works, companies are creating something that often will dramatically undermine the market for those works, and thus dramatically undermine the incentive for human beings to create things the old-fashioned way”.

Chhabria also notes a simple flaw in one of the AI companies’ arguments: that applying copyright law will stifle the development of this technology. He notes that any finding that this use of copyrighted materials isn’t fair use does not bar that use, it just requires that AI companies need to reach a commercial agreement with copyright holders to compensate them for the – unfair – use of their materials. As he points out, these businesses project that they will make billions, indeed trillions, of dollars from AI services, so should be able readily to afford such licensing. Indeed, the court saw evidence that Meta initially sought to licence book materials for training purposes, and considered spending up to $100 million on doing so. This never happened because book publishers do not hold rights to this use of book materials – like other novel uses, the rights rest with the authors – so there is no central point or points for such a negotiation. The fact that AI companies are seeking direct commercial benefit from their use of copyright materials makes their burden in demonstrating fair use much harder.

Despite Chhabria’s conclusions that seem to strongly favour the copyright-holders who brought the case, he nonetheless found against them. The copyright holders are 13 authors who argued that their works had been used in training Facebook’s Llama LLM models. In essence they failed in their claim because their lawyers focused their efforts and arguments in the wrong place. They made their arguments predominantly under the first three of the four factors in §107 of the Copyright Act, and failed in those. While the fourth factor – the effect of the use on the potential market for the copyrighted work – is generally seen as the most important, that is not an argument they made strongly. They simply did not argue (or were at best “half-hearted” in those arguments) that their works had been used as the basis for a tool which might flood the market with similar works, undermining the value of their copyright, nor did they provide evidence to support such an argument. This was “the potentially winning argument” according to Chhabria; the (weaker) points actually deployed in argument before the court did not succeed.

Chhabria was clear:

“this ruling does not stand for the proposition that Meta’s use of copyrighted materials to train its language models is lawful. It stands only for the proposition that these plaintiffs made the wrong arguments and failed to develop a record in support of the right one.”

It does seem ludicrous that the most valuable companies in the world should argue that it is fair for them to take stolen copies of books subject to copyright protection (the training materials were taken from so-called ‘shadow libraries’, of illegally scanned books) and make what they predict will be huge commercial profits as a direct result, while providing the copyright holders with no compensation. The fact that Meta explored licensing but found it too difficult and delaying helps support the case that this would be the right thing to do.

The Kadrey case reports one other specific element of the training of Llama models – that they are taught not to produce more than 50 words together that are repeated from any one source (even if provided with highly directive prompts to do so). The fact that this is a deliberate part of the training shows just how prone these technologies are just to leaning on what they have read. In a recent Financial Times interview, Professor Emily Bender, coiner of the term stochastic parrots and co-author of both the academic article that brought the term to prominence and of The AI Con, is quoted as calling LLMs “plagiarism machines”.

I have to admit that, as may be apparent from my recent reading habits, that I am an AI sceptic. I suspect that we will look back on this period with puzzlement, and wonder why we threw colossal amounts of computing power – and colossal levels of energy in our carbon-constrained world – at jobs that human brains are better at. AI is neither artificial nor intelligent: it isn’t artificial because it depends on human creativity in the training, and it also depends on significant, horrible, labour (typically cheap precarious labour in emerging economies) in cleansing the models of the filth that it produces because it has been trained on, among other things, the global sewer that is the Internet. It isn’t intelligent, it’s just reproducing others’ language patterns based on statistics, “haphazardly stitching together sequences of linguistic forms it has observed in its vast training data…without any reference to meaning” as the stochastic parrots paper put it. As Bender told the FT, we are “imagining a mind behind the text…the understanding is all on our end”. There will no doubt be jobs that AI technologies are useful for, but like any human tool it is tailored to its task, and not a general purpose vehicle for all activity. Currently we have a hammer and are making the mistake of seeing everything as a nail.

As a result, I suspect that much of the billions being deployed in AI currently will turn out to have been wasted. I should admit also that my view may be coloured by the fact that I entered the investment world exactly at the time of the dotcom bubble. While I avoided losing money in the dotcom bust, I also missed out on investment gains as that bubble inflated.

But this is a blog on fairness, not AI cynicism. The Kadrey decision did not conclude that Meta’s actions were fair, only that the copyright-holders had failed to deploy the arguments that might have shown how unfair the use of their materials was. This will clearly not be the last such case, and while the AI businesses will continue to deploy some of their investors’ millions into their defence, judge Chhabria’s legal conclusions suggest they will have a challenging time winning cases argued on the right basis.

Rather than finding that Meta’s use was fair, the Kadrey decision is highly suggestive that AI is not fair in its use and abuse of copyright materials. That feels right: fairness should always tend to rebalance power away from those with billions towards those of whom they take uncompensated advantage.

I am happy to confirm as ever that the Sense of Fairness blog is a purely personal endeavour

Kadrey v Meta, Case No. 23-cv-03417-VC, Summary Judgement 25 June 2025 (Docket Nos 482, 501)

The AI Con: How to Fight Big Tech’s Hype and Create the Future We Want, Emily Bender, Alex Hanna, Bodley Head, 2025

AI Snake Oil: What Artificial Intelligence Can Do, What it Can’t, and How to Tell the Difference, Arvind Narayanan, Sayash Kapoor, Princeton University Press, 2024

Meta seeks $29 billion from private credit giants to fund AI data centres, Eric Platt, Oliver Barnes, Hannah Murphy, Financial Times, 27 June 2025

Nvidia insiders cash out $1 trillion worth of shares, Michael Acton, Patrick Templeton-West, Financial Times, 29 June 2025

The Copyright Act, 17 USC

AI sceptic Emily Bender: ‘The emperor has no clothes’, George Hammond, Financial Times, 20 June 2025

On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?, Emily Bender, Timnit Gebru, Angelina McMillan-Major, Shmargaret Shmitchell, Proceedings of FAccT 2021

An archaeology of equality

Reading an old book can sometimes feel like an archaeological dig – you find fragmented artefacts of how people used to think and have to try to piece together an understanding of their world, and their world view. Very often it serves to illuminate our own.

That’s definitely my sense while reading a book called Equality by an old socialist and economic historian, RH (Richard Henry) Tawney. My edition dates from 1964 but the original book was published in 1931, based on lectures given in 1929. This version enjoys a 1964 introduction by founding father of social policy Richard Titmuss, and no fewer than two prefaces by Tawney himself, one from the 1951 revised edition and one from the ‘substantially revised’ 1938 edition. Reading through these in this order is like uncovering historic layers of English inequality, and repeated aspirations for greater equality. What’s more, the first chapter of the book, The Religion of Inequality, starts by referring to a lecture by Matthew Arnold from I think 1878, to which it attributes the coining of that phrase.

I find it impossible to read these archaeological artefacts and not reflect on our own age. This blogpost aims simply to capture a few sentiments from each of these layers of history. Readers will no doubt be conscious of the great ruptures and attempts towards greater equality that provided the context for the writing of each of these layers of commentary, from the heights of the Cold War, the challenges of the Second World War and the creation of the welfare state that followed it, the Great Depression and the rise of fascism – and even, back around the 1870s, the first steps to broad enfranchisement (and while the right to vote did not then extend to women, that decade did see them permitted for the first time to retain their own property rather than simply surrender it on marriage).

Titmuss in his 1964 introduction:

“We…delude ourselves if we think we can equalize the social distribution of life chances by expanding educational opportunities while millions of children live in slums without baths, decent lavatories, leisure facilities, room to explore and the space to dream. Nor do we achieve with any permanency a fairer distribution of rewards and a society less sharply divided by class and status by simply narrowing the differences in cash earnings among men during certain limited periods of their lives.”

“Long years of economic depression, a civilians’ war, rationing and ‘fair shares for all’, so-called ‘penal rates’ of taxation and estate duty, and ‘The Welfare State’ have made little impression on the holdings of great fortunes…Wealth still bestows power, more power than income, though it is probably exercised differently and with more respect for public opinion than in the nineteenth century.”

“These consequences of technology in an age of abundance are more likely to increase than to decrease differentials in income and wealth if no major corrective policies are set to work…Without a major shift in values, an impoverishment in social living for some groups can only result from this new wave of industrialism.”

Tawney in his 1951 preface:

“Like earlier wars of religion, the credal conflicts of our day will find varying issues in different regions; but, if Europe survives, societies convinced that liberty and justice are equally indispensable to civilization will survive as part of her. The experience of a people which regards these great abstractions, not as antagonists, but as allies, and which has endeavoured, during six not too easy years, to serve the cause of both, is not barren of lessons which may profitably be pondered.”

And he quotes The Times from 1 July 1940:

“If we speak of democracy, we do not mean the democracy which maintains the right to vote, but forgets the right to live and work. If we speak of freedom, we do not mean a rugged individualism which excludes social organization and economic planning. If we speak of equality, we do not mean a political equality nullified by social and economic privilege. If we speak of economic reconstruction, we think less of maximum production…than of equitable distribution.”

Tawney in his 1938 preface:

“It is still sometimes suggested that what Professor Pigou, in his latest work, calls ‘the glaring inequalities of fortune and opportunity which deface our present civilization’ are beneficial, irremediable, or both together. Innocent laymen are disposed to believe that these monstrosities, though morally repulsive, are economically advantageous, and that, even were they not, the practical difficulties of abolishing them are too great to be overcome. Both opinions, it may be said with some confidence, are mere superstitions.”

“Institutions which enable a tiny class, amounting to less than two per cent of the population of Great Britain, to take year by year nearly one quarter of the nation’s annual output of wealth…are an economic liability of alarming dimensions. They involve…a perpetual misdirection of limited resources to the production or upkeep of costly futilities, when what the nation requires for its welfare is more and better food, more and better houses, more and better schools.”

“Today, when three-quarters or more of the nation leave less than £100 at death, and nearly two-thirds of the aggregate wealth is owned by about one per cent of it, inheritance is on the way to become little more than a device by which a small minority of rich men bequeath to their heirs a right to free quarters at the expense of their fellow-countrymen. The limitations imposed on that right during the past half-century were greeted, when first introduced, with the usual cries of alarm; and the alarm, as is not less usual, has been proved by experience to be mere hysteria. It is perfectly practicable, by extending those limitations and accelerating their application, to reduce the influence of inheritance – at present a strong poison – to negligible dimensions.”

“To make [democracy] a type of society requires an advance along two lines. It involves, in the first place, the resolute elimination of all forms of special privilege, which favour some groups and depress others, whether their source be differences of environment, of education, or of pecuniary income. It involves, in the second place, the conversion of economic power, now often an irresponsible tyrant, into the servant of society, working within clearly defined limits, and accountable for its action to a public authority.”

Tawney reports that Matthew Arnold said, in c1878:

“Arnold observed that in England inequality is almost a religion. He remarked on the incompatibility of that attitude with the spirit of humanity, and sense of the dignity of man as man, which are the marks of a truly civilized society. ‘On the one side, in fact, inequality harms by pampering; on the other by vulgarizing and depressing. A system founded on it is against nature, and, in the long run, breaks down.’”

As LP Hartley says in another old book, one that deliberately plays with memory and history, “The past is a foreign country; they do things differently there.” But often ‘they’ worried about the same challenges we do, and sought similar solutions.

I am happy to confirm as ever that the Sense of Fairness blog is a purely personal endeavour.

RH Tawney, Equality, 1964, Unwin

Arthur Pigou, Socialism versus Capitalism, 1937

LP Hartley, The Go-Between, 1953