I originally wrote about the mysteries of algorithms and small business credit after attending two excellent conferences on the subject here in New York City. Here is a good panel discussion on the topic. Since then Cornell Tech has launched the Digital Life Initiative, offering an abundance of excellent research and deeper explorations.
There are two schools of thought on the merits of algorithms or predictive analytics, apart from ethical considerations. One is excitement about the power and scalability of “objective” scoring based on data so, you know, rock and roll.
The other is skepticism about how well they work. How do we know we can trust the results of these machine-driven beasts? We can’t see one or more of the following: 1) the training data used to inform the model used in the algorithm; 2) data against which the algorithms are applied (indeed, we often don’t know how data are collected because of the degree of brokering among data publishers); 3) algorithm source code; 4) underlying logic expressed through source code that may not be obvious; 5) meaning of the outputs; and 6) statistically sound evidence of impact of the outputs.
What does it mean? We all are being profiled in a variety of ways using data we might not expect would be used to profile us. While I can raise the questions about the quality of predictive analytics, I’m more knowledgeable about the ethical and legal implications.
If that profiling takes place in the context of consumer credit decisions, the U.S. provides a measure of regulatory protection and recourse. If it takes place in the context of small business credit decisions, we mostly are left in the dark with no ability to understand much less contest them. It’s a big issue because so many decisions rely on personal data and credit profiling as well as data about the small business.
This current reality as bad news and good news. The bad news is bad decisions and deserving small businesses failing as a result. The good news is there is a “green field” and therefore a very big opportunity to offer technology that fosters accountability, transparency, ethics, and fairness in small business credit.
We do that by following a very simple principle–put small business owners in charge of their credit profile. Give them (you) tools to collect data for your own credit profile that you may share with lenders, creditors, and others, on your terms.
It’s a hard set of problems to solve but we carry on!