Facebook Experiments on its Users

http://national.deseretnews.com/article/1791/Why-Facebook-has-been-experimenting-on-you-behind-your-back.html

This article discusses an experiment Facebook recently did on a subset of its users.  It did not bother to tell anyone or even to ask permission.  It justified this by pointing out that its license agreement, which every Facebook user presumably agreed to when signing up for an account, includes some kind of clause, giving them permission to experiment on users.  Many people have become upset over this, and many have accused Facebook of unethical behavior.  Is Facebook really justified by their license agreement?

This goes far deeper than just experimental ethics.  The first question, which has been asked repeatedly but never answered, is, "Does clicking an 'I agree' button hold the same legal weight as a signature?"  Most software, when installed, asks the user to read the license agreement and click a button labeled something like, "I agree."  This is not a signature, and these extremely informal agreements have never been seriously tested legally.  It is reasonable to say then, that Facebook's license agreement does not actually hold any legal weight, and thus, no Facebook users have agreed to allow themselves to be experimented on in a legally binding fashion.  Otherwise stated, depending on the eventual ruling over what counts as a legally binding agreement, Facebook may have actually violated the law by experimenting on its users without legally binding permission.

The second problem is experimental ethics.  This is hairy to begin with.  The first question, which has been asked repeatedly, but never answered satisfactorily, is, "Is it ethical to conduct a deceitful experiment on someone, even with permission?"  This is problematic, because researchers frequently run into problems where they cannot tell the subjects the nature of the experiment without tainting the data.  The most benign form of this is placebo testing.  In this kind of experiment, the subjects typically know that they may be getting the real test drug or a fake drug, but they do not know which.  The point of the placebo (the fake drug) is to eliminate bias introduced when patients improve because they think they should (this is called the placebo effect).  This is widely considered ethical.  Now, the ethics problem comes in when the researchers do not tell the patients that they may be getting a fake drug.  Is it ethical to tell test subjects that they are getting the real thing when some are not?  This does not typically happen in drug tests, but it happens all the time in psychological experiments (the type of experiment conducted by Facebook).  Introducing too much information about the experiment can alter the behavior of the participants.  Because psychological experiments measure behavior, introducing behavior altering knowledge like this will often cause invalid results.  In other words, telling the participants too much information will alter the results of the experiment.  Sometimes, even telling participants that they are test subjects will alter their behavior.  Some researchers consider any deception unethical.  Others consider it unethical to experiment on people without permission, but they do not consider it wrong to withhold information or even lie about the nature of the experiment.  Legally, it is fine to withhold information (within reason; withholding information about risk factors is illegal), but experimenting on subjects without their permission is very shaky ground.  (Observing behavior without asking permission is fine, so long as surveillance laws are adhered to.)  Facebook's situation may technically fit into the ethical-to-ask-but-not-reveal-information area, if the license agreement holds, but there is one more factor that plays into Facebook's potential guilt.

The third problem is what constitutes agreement.  We have already looked at the legal aspects of agreement, but what about the ethical aspects?  Here is Facebook's third problem: Is it ethical to ask general permission to experiment on a person, or should permission be granted on a case-by-case basis?  According to Facebook, their license agreement includes terms allowing them to experiment on users.  While I have not read this license agreement, I can guarantee that it was not referring explicitly to the experiment conducted, and I am likewise certain that if Facebook conducts additional experiments, they will cite the same clause as permission for those experiments as well.  Is this right?  I have never heard of a case where a potential test subject was asked to sign an agreement to allow the researcher to involve him or her in every  experiment that researcher ever conducts (except in this case).  In fact, I strongly suspect most researchers would consider such a long term agreement highly unethical.  Typically, when a study is conducted (any study) involving people, the agreement signed by participants is exclusive to that study.  It does not apply to any other study past, present, or future.  If multiple studies are being conducted at the same time, and some subjects are participating in more than one, individual contracts are signed for each study.  This is both for legal and ethical reasons.  Facebook, on the other hand, has decided that a long term cover-all agreement is sufficient.  Legalities aside, is it ethical to ask people to sign agreements to participate in multiple undefined experimental studies, before the studies have even been planned?

In may opinion, Facebook has made a grave breach of ethics.  While I think that this should be illegal, we will ultimately have to let the law sort that part out.  My biggest problem with Facebook's experiment is not the experiment itself, it is the fact that Facebook required people to agree to participate in an unlimited number of arbitrary studies, as terms of its use.  These are not studies that were already well planned either.  If Facebook came up with some new study tomorrow, it could, according to its own logic, subject its users to that study without any warning.  What if Facebook decides to conduct a study on shifting public opinion (probably in its own favor)?  If you have a Facebook account, you have already consented to participate.  Really, according to the logic of Facebook's lawyers, Facebook could conduct "experiments" designed to change the results of elections, without any accountability.

I suspect a lot of people believe that this is not a problem, because Facebook would never do that.  Maybe this is true.  Consider, however, what happens when current Facebook executives retire or die.  You might trust Facebook now, but can you honestly say that you trust every person who will ever control Facebook (hint: if you say "yes," you are either an idiot or a liar)?  Are you willing to trust your mental health to a for-profit company who's primary goal is to make as much money as possible?  I am certainly not.

I think this incident may be a signal that we need some legal reform in the field of experimental science.  I am not advocating direct government involvement in all science experiments, but we need some serious public discourse on what is ethical and what is not, and then we need some legal reform based on the majority opinions coming out of the discussions.  It is already clear, based on the reaction of many Facebook users that many consider Facebook's actions to be unethical.  If the majority of voting age U.S. citizens agree, then it should also be illegal.  (I believe that public dialog is important though, to ensure that everyone understands all of the costs and benefits, so that they can make well informed decisions.)