Gail's Law and the Confederate Flag

cfo log entry #371.

i met with a dozen bangaloreans to greenlight the product: a confederate 'flag' for sale on amazon.

this was not an obvious call. but i was visiting our indian office to lead investment and budget decisions. and when this product came up for review they asked me to weigh in. it took me a minute to realize i was seeing gail's law in action.

as a boy i watched the confederate flag flying from the dome of the south carolina capitol. i learned that if our southern armies had won, the bus seats occupied by my black friends and classmates would now be empty. i asked my teachers how the losing flag for an egregious historical event was granted such a premium location.

it was and still is legal to sell confederate merchandise. but in the wake of the 2015 church shootings in my home state, amazon and other retailers committed to stop.

but here's the leadership lesson that i almost missed in bangalore.

our ai systems had marked this product for catalogue deletion. but instead we pulled it up for human review. the 'flag' seen by the system was just the cover art for a historical book, pictured alongside a union flag from the opposing northern army.

this book wasn't glorying the southern armies or historical slavery. it was a literary treatment of the complex emotions and conflicting loyalties that led to 700k dead soldiers in the deadliest conflict in u.s. history.

in other words, the type of book that belongs on amazon's shelves. it's just that our automated systems were 'too smart to see it'.

we were observing gail's law

which explains so much of today's confusing world:

as work becomes more automated and digitized, the remaining decisions and efforts left to humans become increasingly complex and difficult.

i coined gail's law at my first job, while programming an optical robot to find weld spatter that would otherwise create issues in nuclear reactor cores. as i succeeded in my job, our inspector gail's job got harder - instead of spending 80% of her time finding issues and 20% fixing them, it became 20% / 80%. meanwhile her hourly pay did not increase.

this type of work is much, if not most, of modern engineering. once you see it, you immediately become nicer to yourself and your employees. you acknowledge our species can't help but routinize last year's innovations so that we can box them into procedures, systems, sub-routines, and prompts.

each time we do this with previous work we are advancing a planet-wide experiment to find out what will be left for us to do, what does it mean to be human and not machine. what is our unique value proposition back to the cosmos.

we could have tried to answer this through plato's philosophy, da vinci's art, paul's theology, or chomsky's language acquisition models, but if we're going to run the human experiment this way, let's at least log the results somewhere.