Comment by brianchu | Hacker News Reader

Comment by brianchu | Hacker News Reader

Parent Full thread

> Furthermore, even if the probabilities were known, what matters is not just the weighted relative costs and benefits, but the potential for absolute disaster. If there is a 1% chance the world will end unless we do x, we shouldn’t do a cost-benefit analysis. Instead, assuming x is feasible, we should simply do it.

This is extraordinarily dangerous thinking. The "One Percent Doctrine" [1] is and was a huge driver behind the rationale for all the national security and counter-terrorism efforts in the post September 11th era - former Vice President Cheney is recounted as saying "If there's a 1% chance that Pakistani scientists are helping al-Qaeda build or develop a nuclear weapon, we have to treat it as a certainty in terms of our response." That very same logic was likely extended to Iraq - if there is a 1% chance Saddam is building WMDs, we must invade (it helps that the Bush Administration believed the chance was far greater than 1%).

Another problem with this 1% kind of thinking is that it's not even clear if many of the kinds of events that are purported to end the world would actually end the world - global warming won't cause human extinction; it will just cause hundreds of millions, if not a billion, deaths (through disease, flooding, displacement); the same applies for nuclear holocast.

[1] http://en.wikipedia.org/wiki/The_One_Percent_Doctrine

EDIT: I realized the irony is that you could call this the "Dangers of Thinking Like a Politician."

People should learn economics, not prance around saying things they think an economist would say. This blog post constructs a straw man of the latter and than attacks it, paradoxically, as the latter.

It's even worse than that, because the example being used to condemn cost/benefit analysis actually reinforces it.

If you have a 1% chance of destroying the world, you're pretty much going to want to avoid taking that chance in almost every case according to a cost/benefit analysis, since 1% times the trillion trillion dollars that the world is worth to us will end up being more valuable than any alternative. The exception is when the next best alternative is even worse, i.e. a much higher probability of something only slightly less catastrophic.

On the other hand, suppose the calculated probability is not 1%. Suppose the alleged world-ending calamity is that the Flying Spaghetti Monster is going to be displeased and set fire to all of reality if you don't pay me a billion dollars. The probability of that actually happening is not 1%. It isn't 1% of 1%. It's close to, but not exactly, 0%. Is the argument that you should still pay me a billion dollars? After all, the risk if you don't is that the world will end. I accept payment as cashier's checks or wire transfer.

"If you have a 1% chance of destroying the world, you're pretty much going to want to avoid taking that chance in almost every case according to a cost/benefit analysis"

No, you aren't. If that 1% chance is balanced by a 99% chance to increase the world's wealth by 2%, the expected gain is .98%, so anyone using pure expected cost/expected benefit analysis would give the go ahead if the cost of this were, say 0.5%.

The problem, I think, is that those statistical calculations do not handle the case where there are outcomes that imply that one cannot repeat the game often enough to filter out statistical variations. In that sense, it is related to http://en.wikipedia.org/wiki/Martingale_(betting_system): Yes, you can beat roulette, if (and that is a big IF) you don't go broke in the process.

>No, you aren't. If that 1% chance is balanced by a 99% chance to increase the world's wealth by 2%, the expected gain is .98%, so anyone using pure expected cost/expected benefit analysis would give the go ahead if the cost of this were, say 0.5%.

If you could actually improve the world by 2% in doing that, maybe you should -- but don't discount what that would mean. You're not talking about a 2% increase in GDP. GDP measures commerce per year, we're talking about everything, forever. You have to account not only for the loss of the economy, but the loss of human life, family, humanity itself and all future endeavors of any nature by anyone. And if the alternative is a gain rather than a competing loss, you have to put a value on our desire to avoid losses as being greater than our desire to make gains. If you still think you can create a gain that would balance that loss, let's hear it.

And the fact is that we do as a factual matter make decisions like that. We built nuclear weapons because the risk of blowing up the world was calculated as being a better choice than the risk of fascism taking over. We grow genetically modified food because the ability to feed the population is deemed more important than the risks it entails. You can dispute decisions like these, but then you're just disputing the value we've placed on each alternative, not the process of comparing relative values in order to make the choice.

The idea that some risks are insufferable doesn't break economics, it just changes valuations. You have to consider both the consequence and the probability. Otherwise the world should pay me the billion dollars, because there is some risk that the alternative will be world-ending.

People object to putting a number on things like life and love, but there is no avoiding it. At the end of the day you have to make decisions. You have to weigh love against safety against money and make a choice.

Cashier's check?? What is this, the dark ages? https://news.ycombinator.com/item?id=5562660

Oh, I thought you were going to mention http://sivers.org/1pct

Sad that Cheney was the most famous and morbid believes in the 1-percent fallacy.

Reply on news.ycombinator.com