from Patrick McKenzie | by Patrick McKenzie

In addition to altruistic motivations, some of the stochasticness of the process includes that you have capacity to do some number, X, of human transaction reviews per day to cover Y transactions. You tell computers to select Z% of heuristically fraudiest to review. t.co/N77PPyjT62

Then the pandemic happens. Unexpectedly, you are hit by two things simultaneously: a) Your capacity for reviews, X, comes out of a relatively large number of people occupying a relatively small number of pockets of air. Many of those close involuntarily for the first time ever.

b) You breathe a sigh of relief, thinking at least the number of transactions declining will save you from this particular business challenge. The number of transactions in economy does not materially decline. The number you process *goes up* as transaction patterns shift.

So what do risk analysts and managers tell the computers? They tell them “OK, relative to pre-pandemic norms, I want you to be less selective with respect to marginal cases. Keep denying the obviously bad. Keep passing through the obviously good. But we cannot review as many.”

And so transactions which were marginal closer to bad were auto declined and transactions which were marginal closer to good were autoaccepted. The bad guys noticed this. Quickly. And they had a field day.

Obligatory disclaimer: not talking about any particular company’s results in particular. Though literally every risk team thinks I somehow got their data.

“Wait everyone can’t have tread action volume go up if economywide numbers are close to flat.” Good catch, second thoughts. I was subconsciously scoping this observation to tech-forward firms. As you know, they took share during the pandemic while e.g. community banks struggled.

*transaction My kingdom for LLMs eating autocorrect.

That should be easily possible within a few years given improvements in inference and a hardware upgrade cycle or two? Though will be challenging for Apple to message. Ah, the future contains wonders.

“You’re being sarcastic?” Nope, minor superficial improvement to my written communication trivially justifies $X,000 per year, assuming it does not break existing workflows or edit out the reasons why my writing works.