89.0% (.95 probability of being between 84.3% and 92.7%)

Finally, I have exact numbers on how much of a year of abled life a year of my life is worth, complete with 95% confidence intervals! This information will surely be useful to anyone making resource allocation decisions. For example, if a scarce medication has a 95% chance to save an allistic patient’s life, but a 99% chance to save mine, the QALY-maximizing to do is to give it to the allistic patient.

So, where are these numbers from, anyway? Some radical eugenicist or curebie site? No, actually. These are disability weights, as published in The Lancet. Designed to lead to the most beneficial resource distribution when dealing with non-life-threatening conditions, they often suggest inappropriate outcomes when applied in certain cases.

The method used to generate them was a combination of “person tradeoffs”, which effective altruist and blogger Jeff Kaufman describes as asking “people, generally public health professionals, if given the choice between curing 1000 people with disability X and 2000 people with disability Y which one they would choose”, with the more intuitive method of asking people which condition is worse. A full copy of the 2010 results is seen in this PDF.

(Note: Jeff Kaufman does not say ableist things online, and he does not seem to think that my life is less valuable than that of an NT. Do not take any criticisms of disability weights or the QALY as criticisms of him.)

Unfortunately, while this is very useful for cases in which the disability should be cured, it effectively assumes that it should be cured. Using time tradeoffs, another common method, you’d only get me to accept a cure if it increased my lifespan by quite a lot, as I’m not sure whether what I call “me” would survive the process. Clearly, this does not mean that neurotypicality is a horrible nightmare that must be cured now, nor does it mean that my life is more valuable than NT life (although, as a partially selfish being, I naturally value my own life more than that of others, and if you’re reading this website instead of trying to acquire resources for lifesaving organizations, you’re partially selfish too).

So, what course of action am I proposing?

The costs of repeating the research required to generate accurate numbers would be high, because the original research was international. The benefits of having more accurate QALY numbers are probably low, because many groups would keep using the old numbers, and the new numbers are not likely to dictate very different actions to nonprofits. Completely ignoring quality-of-life considerations leads to equally bad ideas, such as considering nonlethal torture of thousands to be permissible if it provides a 5% chance of increasing 1 person’s lifespan by a couple decades. Spending decades trying to calculate humanity’s CEV with human minds and budgets is highly inefficient.

The people looking for some mechanism of arriving at the most moral utility function seem to be people trying to develop a friendly superintelligence, which, if done correctly, would definitely prevent medical ableism, but considering the main impact of a friendly superintelligence to be “an end to medical ableism” would be like considering the main impact of the development of intelligent life to be “extinction of the dodo bird”.

The best action is probably to keep using the same numbers until we actually need better ones, and supporting the search for a most moral utility function by sending money to MIRI or one of the other “safe/Friendly AI” groups, who seem to be trying to create an agent which ends up following the most moral utility function.

Advertisements