Why “fixing the algorithm” misses the point

Recently, I was at an event where a representative from an organisation advocating for better data stewardship made an impassioned argument: data workers, they said, should have rights over the data they collect. The room nodded along. I felt uneasy. Not because the concern for workers was misplaced, but because the framing felt oddly disconnected from reality. Workers, historically, have rarely owned what they help produce. I couldn’t help wondering: have we ever expected a factory worker to co-own the car rolling off the assembly line because they tightened one bolt? And more importantly, does the average data worker even want ownership of “their” data, or are they more worried about something far more immediate -mounting medical bills, job insecurity, or the next rent payment? That moment stayed with me because it crystallised something I’ve been noticing for a while: how often well-meaning civil society efforts narrow complex realities into hyper-focused problem statements, and then build equally narrow solutions around them.

Over time, advocacy tends to converge on sharply defined issues. I suspect funding pressures play a role -the need to articulate problems succinctly, to package them cleanly, to make them legible to donors and policymakers. But the result is often a kind of reductionism. In this case, the sprawling, messy reality of gig work and platform labour was reduced to a single claim: data ownership for data workers. There is something comforting about such clarity. But there is also something dangerous about it.

Take the gig economy. It is easy – and sometimes justified -to criticise large tech platforms for the power they wield and the profits they extract. And yes, that power absolutely needs to be checked. But it is also undeniable that these platforms create real opportunities. Just yesterday, there was news of a so-called strike by gig workers, and yet many continued to log in and make deliveries. That doesn’t invalidate concerns about unfair treatment, but it does complicate the story.

Why would someone keep working during a strike? Because for many, the platform is not an abstract villain -it is a source of daily income. This is the part that neat problem statements struggle to hold.

I think this is where conversations about fairness in technology often go astray, and also why this paper from my alma mater resonated with me deeply. We start treating technology itself as the site of morality -good or bad, fair or unfair -when in reality, technology is a tool. It expands what is possible. It enables things that would have seemed absurd not too long ago -like ordering paneer butter masala to be delivered to your seat on a moving train between Chennai and Trichy.

That doesn’t make technology neutral in its effects, but it does mean it cannot be judged in isolation. A tool can generate immense value and still produce harm. The question is not whether tech is good or bad, but how power, incentives, and accountability are arranged around it.

This is why simply “fixing the tech” -by making algorithms fairer, more transparent, or even open -is rarely enough. An algorithm can meet every fairness metric and still operate within a system that treats workers unjustly. When that happens, we haven’t solved the problem; we’ve misidentified it.

What actually checks power, historically, are societal responses. Labour law is one such response. Antitrust law is another. Rules around fair working conditions, minimum wages, social security, and collective bargaining exist precisely because markets left to themselves don’t distribute risk or reward equitably.

Similarly, laws that prevent monopolies, duopolies, and oligopolies are essential not because size is inherently evil, but because concentration of power without countervailing force eventually distorts outcomes. The state has a role here -not only as regulator, but also as an enabler. Creating digital public infrastructure and digital public goods can widen access, reduce dependency on a few dominant players, and allow smaller actors to participate meaningfully.

Processes matter too – not just what is done to check power, but how it’s done too – transparency, accountability and governance are all important pieces. But even this is not the whole picture.

At the individual level, people constantly make trade-offs. Gig workers weigh flexibility against precarity. Consumers trade convenience for invisible costs borne elsewhere. Data principals balance privacy concerns against access to services. These decisions are not made in a vacuum; they are shaped by context, necessity, and uneven bargaining power.

Which brings me back to civil society.

If organisations advocating for fairness -whether in data, labour, or technology -focus only on isolated technical fixes or singular rights claims, they risk missing the terrain on which real lives are lived. Worse, they risk advocating solutions that sound principled but fail to resonate with the very people they seek to protect. This is why even laws don’t work in isolation.

This is not an argument against rights, or against critique of powerful tech companies. It is an argument for recognising that fairness is not something we can engineer into a system with one clever intervention. It is something that emerges -imperfectly and continuously -from law, institutions, markets, and social norms interacting with each other.

It also means civil society has to invest more in grounded research: understanding how the various stakeholders actually experience platforms, what trade-offs they are making, and where leverage truly lies. Advocacy that ignores context may be elegant, but it is rarely effective.

Perhaps the more honest question to ask is not “Who should own the data?” but “Who is paying the cost of this convenience, this efficiency, this growth?” And just as importantly: “Who has the power to safeguard those costs from becoming exploitation?” If we can sit with those questions -without rushing to overly neat answers -we might begin to advocate not just loudly, but wisely.

END OF ARTICLE