Picture a man walking into a bank in Columbus, Ohio, in 1955. He's thirty-two, a machinist at a local plant, married with two kids and a third on the way. He wants to borrow money to buy a house on Maple Street — a modest place, nothing fancy, but solid. He sits across from the branch manager, a man named Gerald who has worked at that same bank for nineteen years and knows half the town by sight.
Gerald doesn't pull up a credit report. There isn't one. He looks at the man's pay stubs, asks about his employer, asks about his father, and makes a judgment. The loan gets approved — or it doesn't — based on something that doesn't fit neatly into a spreadsheet.
This was how American lending worked for most of the twentieth century. And understanding what replaced it requires understanding what it actually was: a system of remarkable intimacy, genuine community knowledge, and sometimes breathtaking unfairness.
The Character Loan Era
Before credit scoring became standard practice, American banks evaluated borrowers through a framework sometimes called the "Five C's": character, capacity, capital, collateral, and conditions. Of these, character sat at the top of the list — and character was almost entirely subjective.
Did you pay your bills? Did your employer speak well of you? Were you known in the community? Did your family have a history with the bank? These questions mattered enormously. A man who'd banked at the same institution for fifteen years, whose father had banked there before him, whose pastor would vouch for him — that man had something that no income statement could fully capture.
In tight-knit communities, this system worked with a kind of organic efficiency. The banker genuinely knew his borrowers. He knew which local businesses were struggling before their financials showed it. He knew which families had fallen on temporary hard times versus which ones were chronically unreliable. That accumulated knowledge had real value.
The Darker Side of Knowing Your Neighbors
But here's where the story gets uncomfortable, and any honest accounting of this era has to spend time here.
The same subjective judgment that helped a trusted machinist buy a house on Maple Street could — and routinely did — shut out entire categories of Americans based on race, gender, national origin, or religion. A Black family in that same Columbus neighborhood in 1955, regardless of income, savings, or employment record, might find the bank manager suddenly less certain, less generous, less willing to extend the benefit of the doubt.
This wasn't incidental to the character-loan system. It was structural. When lending decisions rested on personal relationships and community standing, they inevitably reflected the biases of whoever held the relationship and defined the community. Redlining — the systematic denial of mortgages in minority neighborhoods — wasn't some rogue behavior. It was policy, sometimes written, sometimes unwritten, but consistently enforced by the same human judgment that the system celebrated as its strength.
Women faced similar barriers. Before the Equal Credit Opportunity Act of 1974, a bank could legally deny a woman credit simply because she was a woman, or require a husband's signature regardless of her own income. The personal touch, in practice, often meant the personal prejudice.
Enter the Algorithm
Bill Fair and Earl Isaac developed the foundational credit scoring model in the late 1950s, but FICO scores didn't become the dominant lending tool until the late 1980s and early '90s, when Fannie Mae and Freddie Mac began requiring them for mortgage underwriting. The shift happened fast once it started.
The logic was compelling: replace subjective human judgment with objective, consistent, mathematically derived assessments of creditworthiness. A three-digit number — built from payment history, debt levels, length of credit history, and a few other factors — would tell any lender, anywhere in the country, how risky a borrower was. No more banker's intuition. No more hometown favoritism. No more discrimination based on who your father was or what your name sounded like.
In theory, the algorithm was blind to race, gender, and religion. In practice, this represented genuine progress. Studies consistently showed that algorithmic lending reached creditworthy borrowers who'd previously been locked out of the system. The expansion of credit access through the 1990s and 2000s — however imperfect and ultimately unstable — reflected real democratization.
What the Number Misses
But the algorithm also stripped something out that turned out to matter.
Context. Narrative. The kind of information that doesn't fit into a database field.
A person who spent three years caring for a dying parent, fell behind on bills during that stretch, and then returned to financial stability — that story is invisible to a credit score. The number sees the late payments. It doesn't see the reason. A recent college graduate with no credit history at all — no late payments, no defaults, no problems of any kind — looks identical to a high-risk borrower under many scoring models, simply because there's no data to evaluate.
Small business lending has been particularly affected. The local banker who once financed a promising young entrepreneur based on a handshake and a business plan has been largely replaced by automated underwriting systems that want two years of tax returns and collateral documentation. The loan that gets made for the right reasons — because someone with real local knowledge believes in a borrower — is harder to arrange than it used to be.
And the algorithmic system has its own embedded biases, just different ones. Because credit scores partly reflect prior access to credit, they can perpetuate historical disadvantages without any human actor making a consciously discriminatory choice. The formula is neutral. Its inputs aren't always.
A Trade-Off, Not a Solution
What the credit score revolution actually delivered was a trade-off, not a clean victory. Consistency replaced discretion — which eliminated some discrimination and introduced new blind spots. Scale replaced local knowledge — which expanded access and eroded context. Objectivity replaced character judgment — which was fairer on average and sometimes brutally indifferent to individual circumstances.
The Americans who benefited most from the shift were those who'd been locked out by the old system's human biases — and that's not a small thing. The Fair Housing Act and Equal Credit Opportunity Act gave the law; algorithmic lending gave the practical mechanism for enforcing it at scale.
But the Americans who lost something were those whose stories didn't fit the data — people whose creditworthiness was real but whose paper trail was thin, or broken, or complicated in ways that a formula couldn't interpret.
Gerald the bank manager, with all his biases and all his genuine local knowledge, has been replaced by a model that doesn't know your name and doesn't care to. Whether that's progress depends almost entirely on which side of his desk you would have been sitting on.
