At Zynga, Facebook and us got into a game of cat and mouse. Facebook would tweak it’s algorithms and rules to limit our reach, and we would work around it.
Facebook never could win. We had 2000 people working to figure out how to work around their algorithms, and they had a few dozen.
They only won when they just put limiters on Zynga and our APIs. And they won when they forced us to sign a contract or go out of business.
This taught me a powerful lesson, Machine Learning and it’s derivatives, are terrible tools when the cost of a mistake is non-zero.
For Facebook, when the cost of a mistake was non-zero, they didn’t rely on machine learning, they relied on regulations and laws.
But when a Facebook post goes viral about the evils of vaccination, when a Facebook post goes viral about how Clinton murdered children, this is to Facebook’s benefit. It drives engagement, it drives advertising revenue, it accrues tremendous benefit to them.
Over the past few years smart folks, friends of mine even, have tried and are trying to do something about this. But at the end of the day, as long as the damage to society is acceptable to society, Facebook has no incentive to do anything.
Because the cost of mistake is zero to Facebook. No one at Facebook or it’s shareholders is foolish enough to not get a vaccination. No one at Facebook is in danger of being physically assaulted on the streets of New York for being LGBTQA+ (oh wait, some of them are, but I digress).
When Facebook exec’s say that nothing can be done, they are lying. When the cost of mistakes was non-zero to them, they discovered the power of regulation. And they embraced it.
Leave a Reply