[ExI] The post-antibiotic era

[Anders Sandberg]

On 2013-11-20 23:00, BillK wrote:
> I think the UK Chief Medical Officer (and her deputy) writing in the 
> Lancet and the US Centers for Disease Control and Prevention might be 
> a trifle upset at being called 'doomsayers'.

Maybe. But that does not mean the epithet is completely unwarranted. I 
prefer to see them as myopic: they see a major world problem on their 
turf and shout about it. It is just that it is one among several.

> There are different kinds and scales of risks. Since the Cold War 
> ended hardly anyone even considers nuclear annihilation as a 
> significant risk. (Maybe they should, of course). 

My best estimate is around 0.1% chance per year. Over a 70 year 
lifespan, that means a total risk of 6.7%. This means it beats things 
like cerebrovascular disease, accidents, influenza, and liver disease as 
a potential cause of death. It is just that it likely either happens or 
it doesn't, unlike most of the others that have ongoing mortality.

> But people are already dying in hospitals from antibiotic resistant
> infections. That strikes close to home, so people see it as a much
> bigger risk. In my own life I would certainly be far more worried
> about dying in hospital from a resistant infection than a nuclear bomb
> in London.

Exactly. The availability heuristic strikes again.

Note that from an individual perspective antibiotics resistance is 
likely about as big threat as a nuclear war to your health. It is just 
that from a global perspective nuclear wars are much worse, since they 
cause massive correlated death and damage.

> The cure for large scale threats is neither obvious or
> straightforward, so let's solve the easier, more pressing problems
> first.

Which is of course a mistake, since (1) the "easier" problems are often 
rather tough (common cold, antibiotics resistance) and (2) there might 
be rational reasons to focus much more on the big and important problems.

We had a discussion in the office today about this diagram:
http://www.wired.com/wiredscience/2013/11/infoporn-causes-of-death/
Imagine that a genie offered you to either permanently remove kidney 
disease or war from the world. Which one should you choose?

Removing kidney disease will save many more life-years per year than 
removing war. Even a major decrease in the diseases will save more 
lives! So from a consequentialist perspective it looks like one ought to 
do it. A deontologist might argue that wars are more evil because people 
do seriously immoral things in them, but that position essentially means 
that all intentional bad stuff in the diagram (and the vast amount of 
non-lethal intentional evil) should have priority before "mere" bad luck 
like cancer, starvation and malaria. Yep, deontologists are crazy.

Still, there is a very good reason to choose to end war rather than 
kidney disease. Wars have a power law distribution with a very heavy 
tail. The frequency might be going down (c.f. Pinker) but sooner or 
later there will be a Big One, and it has nonzero chance of wiping us 
out. In fact, there is no average war size because of the tail: the 
expectation diverges, so if you measure long enough the war box will 
totally dominate (unless the pandemic box beats it). Meanwhile kidney 
disease remains fluctuating around a mean size: there is no risk of it 
wiping out much of humanity. So in the short run removing it is the best 
choice, but in the long run removing war is the wisest choice.

Same thing for antibiotics resistance: removing one of the nasty heavy 
tail risks is more important in the long run than the resistance. Not 
removing the heavy tails might mean there will not be any long run, even 
if we successfully fix antibiotics resistance.

-- 
Dr Anders Sandberg
Future of Humanity Institute
Oxford Martin School
Oxford University