top of page
  • Writer's pictureKeren Weitzberg

The Mute Calculus of Distant Algorithms

This blog post is based on research for a forthcoming piece in Coda Story.

What kind of work does a credit score do in the world? And what does it mean to give previously “credit invisible” people a digitally generated financial scorecard?

In countries like Kenya where residents have long been excluded from formal banking, people are now receiving credit scores through the culling of mobile data. This has largely been enabled through the advent of digital credit. Millions of cash-strapped Kenyans have turned to the seductive ease of mobile lending, which was first introduced to the country in 2012. Download an app, fill in some basic personal data, agree to the terms and conditions, and money will appear almost instantaneously in your M-Pesa mobile money wallet.

Companies ranging from multinational telecoms, like Safaricom, to Silicon Valley-based startups, like Tala and Branch, are lining up to offer Kenyans loans at the touch of a button. They have also developed novel ways of assessing people’s credit worthiness.

These algorithms prize various kinds of data: Does this person call their mother? How often do they buy phone credit? Do they make frequent payments with their M-Pesa account? How many followers do they have on Twitter? In a feature on Tala, TechCrunch explains that the company "looks at a customer’s texts and calls logs, merchant transactions, overall app usage and other behavioral data....Based on these pieces of information, its machine learning algorithms evaluate the individual risk and provide instant loans in the range of $10 to $500 to customers."

Proponents argue that such methods enable companies to provide non-collateral loans and offer credit to the "unbanked"--those long excluded from financial services. But talk of financial inclusion tends to silence the background noise: the quiet operations that make these digital apps function.

Mobile technology is not only enabling greater and, in many ways, unprecedented state and corporate surveillance of Kenyan consumer habits in the absence of robust data protection laws. It is also opening people up to the mute calculus of distant algorithms.

Financial inclusion, which rests on the larger fetish for choice and freedom, has opened people up to the depersonalized logic of the algorithm.

Digital lending companies claim to "know" a customer through the use of fine-grained personal details gleaned from their mobile phones. But customers are also made legible to fintech through a set of highly depersonalized and statistical assumptions. One assumption is that the lendee is, in fact, the person they claim to be.

Mandatory SIM card registration laws in force throughout Africa have made it easier for mobile credit providers to digitally identify their customers and appear compliant with Know Your Customer (KYC) regulations. But the reliance on digital and biometric IDs is often highly performative, creating avenues for fraud and mimicry within bureaucracies of verification. The speed and ease of mobile transactions have greatly accelerated the problem of identity theft. In Kenya, it remains relatively easy to register a SIM card in another person's name using a forged or stolen ID. Consequently, many Kenyans have found themselves erroneously listed as defaulters with Credit Reference Bureaus with few avenues of redress available to them.

Credit Reference Bureaus (CRBs) are relatively new to Kenya, but they have become increasingly important in the age of digital credit. While CRBs are allowing the previously "credit invisible" to build financial identities, they are also producing new inequalities and barriers to entry by sorting people into the creditworthy and the delinquent. According to one study, 2.7 million Kenyans are now blacklisted by CRBs — 400,000 for loans under $2. Many don't realize that they have been quietly assessed through automated processes and digital traces.

Such digital operations can quickly move from the depersonalizing to the dehumanizing. This was put into sharp relief at the Tatua Center, one of Kenya's only agencies offering alternative CRB dispute mechanisms (a small, under-funded, ostensibly independent operation doing valuable work for consumers). The Tatua Center is also where one can meet those most negatively impacted by the credit-scoring process. There, I was told the story of a man who wanted to send his ill son abroad for treatment. The father discovered he was ineligible for a bank loan, having been mistakenly listed as a defaulter. Despite pursuing multiple avenues to clear his name, he couldn't acquire a loan in time and his son died. “I had to leave the room for thirty minutes so he could cry,” the registrar recalled.

Efforts to remedy one's credit score are often frustrated due to the"inaccessibility of many of these mobile lenders," the registrar explained. One woman had taken out a loan for emergency services for her son. In the end, her son had died and she knew she'd be unable to repay the loan until after his funeral. "This is someone a lender ought to listen to," the registrar lamented. But many companies have no local offices; their headquarters are in distant locations. Some simply say: "We don't see customers".

These cases of desperate parents seeking treatment for their ill children not only highlight the global inequalities in access to health care and the problems of an unregulated fintech sector. They also reveal how algorithmic scoring systems can distance multinational companies from their customers and from the "externalities" of their business models.

Machine learning is transforming routine aspects of people's lives into commodified data points amenable to statistical analysis. This is at once highly personal (culling phone records, payment histories, social media accounts) and deeply depersonalized (assessing individuals based on often biased assumptions and on probabilities gleaned from large sets of aggregate data). Dispassionate, algorithmic decisions about someone's credit-worthiness can have deeply impactful, sometimes heartbreaking effects on people's lives.

Fintech is allowing a statistical, calculative logic to take hold behind the fetish of consumer choice and freedom. By transforming people into probabilities, digital lending algorithms fail to capture the complex, highly personal, often emotional choices that underly people's financial decisions.


Recent Posts

See All


bottom of page