At an AMA yesterday, I was asked whether a scientist should publish his results if he knew the result would be used for evil. If you have no social conscience, then the answer is “of course.” For me, the issue is not so simple. It should be clear that what follows is my own reasoning–for reasons that should become apparent in the course of the post, I don’t think there’s an answer that’s both unambiguous and universally applicable.
I believe that people are responsible for the reasonably foreseeable consequences of their actions. That’s why I don’t accept the typical college premise that being drunk excuses whatever you do. You chose to drink, you know what could happen, and so you’re responsible (for the most part) for what does. (See end note, though.)
As an example of what’s “reasonably foreseeable,” if you and I are horsing around on a concrete plaza, and I throw you to the ground and you crack open your skull, I’m partly responsible because I (should) know that when a person’s head impacts concrete, bad things often happen. On the other hand, if we’re horsing around on a grassy lawn (in Houston, anyway), there are unlikely to be large, hidden rocks in the lawn, so if I throw you into what looks like grass and you hit your head on a hidden rock, that’s truly an accident. (Since I don’t go with absolutist positions such as “you chose to horse around”–which is true, but horsing around does not in and of itself lead to trouble.)
Back to the question that prompted the discussion. While it is true that most human discoveries that can be used to harm others will be so used, two things (to me) make the picture less clear cut. The first is our inability to foresee medium-term consequences that involve many other people. Nuclear weapons/accidents have killed a certain number of people. Nuclear energy has helped many more people. I don’t think there’s any rational way to determine the cost/benefit analysis. By no “rational way,” I mean that I think different people will place different values on individual human lives vs the fact that nuclear power (presumably) contributes much less to disastrous climate change than an equivalent amount of burned fossil fuel.
Another cost/benefit analysis hampered by ignorance: the use of atomic weapons on Japan is often justified by saying that the number killed by them is much smaller than the number who would have died in a conventional attack/invasion. Whether you find that argument persuasive or not probably depends on how you view human life. If you view all human life as unique and irreplaceable, you probably won’t find it persuasive since to kill even one person is morally wrong. But to a certain extent, morals really only apply in small-scale interactions anyway (another post on that later).
The problem with all such cost/benefit analyses is that they are based on insufficient data, perhaps unknowable data. So the “analysis” is at best a guess, which tends to reflect the biases/values of those making the decision more than strictly rational decision-making.
Hence, what you believe on this question will be a matter of belief.
A second point is that scientific discoveries are rarely made in a vacuum (and that’s particularly true as science has tackled more complex problems). Cultural considerations play a role in what discoveries are made when. For instance, say you want to argue that Einstein should not have published E=mc^2. He couldn’t have made that discovery in 1600–the scientific culture didn’t know enough even to understand it, much less make use of it. But around 1900, the culture was such that probably someone else would have made the discovery anyway (say, within 50 years or so).
Assume he doesn’t publish, but someone else does later. Atomic weapons are not built in time to end world war 2, but perhaps they become available in time to use in a third world war. By that time, probably more than one side has them, so if they’re used, the results are significantly more catastrophic. How can one rationally predict the cost/benefit results of not publishing?
Again, it’s impossible to predict the results, so whether to publish or not effectively comes down to the beliefs of the individual scientist.
End note: Someone asked me if my position on foreseeable consequences applied to someone who is date raped while drunk. I pointed out that being raped is not an action one takes, it is an action forced upon one, so the situation is entirely different. A drunk rapist is still responsible for his actions…