Thread by @Noahpinion

Maybe M4A is the best health policy. Maybe it isn't. But people who oppose it are not murderers. So stop saying that. It's silly.

bloomberg.com/opinion/articl…

There are actually (at least) three dilemmas associated with statistical consequentialism.

1. Does intent matter?

2. Omission vs. commission (i.e. is a couch potato a murderer for not going out and saving lives?)

3. The epistemology of empirical analysis

The "intent" and "responsibility" questions have been debated as infinitum for centuries. The third dilemma, about how much we should rely on empirical analyses to tell us the consequences of actions, is new. And it interacts with the first two, and makes them tougher.

To illustrate this, consider the case where the empirical consensus on mortality rates from some policy changes. Suppose a new wave of empirical studies concludes that a health policy that earlier studies concluded reduces mortality actually raises mortality...

...If we really don't care about intent, and only consequences, this shift in the empirical consensus means that we now believe that someone we thought was a hero was *always* a murderer!

Now consider how statistical epistemology affects responsibility. Instead of being a murderer for failing to do things that we *know* save lives, now you're a murderer for failing to do anything that we *estimate* to reduce aggregate mortality rates!!

This quickly leads to a particularly bananas version of Effective Altruism.

Now, this might all seem like a straw man. But, a funny thing, in the Twitter age something that seems like a straw man one minute quickly becomes that a whole tribe of internet shouters scream at you for disagreeing with!

Check the replies to the original tweet in this thread!

Earlier this year, a study declared that high crime rates in Native Canadian communities literally constituted a genocide by the Canadian government!

amp.theguardian.com/world/2019/jul…

Statistical consequentialism is alluring because it allows us to vastly expand the set of people we accuse of murder. And in the Twitter age, when dire rhetoric has been devalued by oversupply, accusations of LITERAL MURDER are a way to cut through the noise.

Which is sad, because there is truly dire bad shit happening in much of the world, including in our own society, and yet we're being inundated with B.S. noise about the murderous evil of...Obamacare??

Statistical consequentialism is really just a way to spam our sense of outrage. It's philosophically dubious and pragmatically unworkable.

Reject it.

(end)

bloomberg.com/opinion/articl…