Or Simple Miscommunication
We’ve heard about Facebook’s manipulation study of 689,000 users’ home pages to find out if it could make people feel more positive or negative.
Go to Twitter Search and type in Facebook experiment and you can see the many articles appearing and discussing the study.
At first, defenders pointed to its Data Use Policy which allows for research to be conducted on user data. Then this :
Today, Facebook COO, Sheryl Sandberg says,
This was part of ongoing research companies do to test different products, and that was what it was; it was poorly communicated.
I wasn’t going to write anything about this failed experiment until reading Sandberg’s response and realizing that most of the media coverage has also compared this emotion study to the experiments most companies do already.
Where The Comparison Fails
Typically, when companies do such tests, it’s done in a way to measure the reaction to or performance of one version of a product against another (or multiple).
The usual intent is to improve a product or service.
Yes A/B testing is performed all the time. I often disagree with the frequency and live user methodology used by many, but what Facebook did was different.
Facebook wasn’t comparing two product versions live to find out if product version B was an improvement to product A.
It intentially attempted to manipulate user emotion by creating two variations of an idea — Happy product A and Sad product B. The control version I suppose would be the normal algorithm used today to determine what displays in a user’s Facebook feed.
Also, as has been mentioned by other academics, it wasn’t even a good study of emotions since no one actually checked with the subjects to find out how they actually felt.
Then, Facebook intentially exposed a live subset of it’s users to Sad product B, not because it was a real product choice that might be better than the control product (group), but to simply prove or identify the negative affects it has on users.
To be clear, the Facebook experiment to study “emotional contagion ” was not about testing two different product versions and finding out which one resonated better with users. Really, was Facebook considering offering a Sad Algorithm version? No, it wanted to see how users react and obviously didn’t mind or think about the negative consequences it could have on its subjects.
So when Sandberg responds that this is just normal product testing and a matter of miscommunication, she’s trying to spin this failed experiment into something it never was. It’s all spin.
Calling the negative PR storm something that is the result of a miscommunicated study is just a strawman argument meant to deflect the fact that the study was unethical from the get-go. There was no informed consent given by the test subjects. Facebook just chose to deliberately expose some of its users to a negative product experience.
[Update] Here’s a little comparison story.
Let’s say I make a gizmo that can connect to the internet in two ways – with built in Wi-Fi or by using a special antenna that connects to your tongue.
Now, let’s say I want to do a study and decide to ship two different product versions to customers who ordered my gizmo – they have the same price. Sure the box says my gizmo can connect to the internet in two different ways, but I’ll just hope my customers won’t read it – I’ll inform them that some gizmos will only have tongue access for research. Since I already have boxes printed, I’ll just add that little bit later in the fine print.
So my study involves shipping my gizmo with one version having only Wi-Fi access and another version that just comes with my tongue antenna feature.
The rest of my customers receive a gizmo that has both Wi-Fi and tongue access. Everyone, pays the same price.
I do this mostly because I just want to test how long customers will stick out there tongue with my antenna attached and if I can make them sad using my product. I already know it’s an inferior product experience and might even lead to some customers experiencing ridicule and perhaps altercations with others because they are sticking out their tongues everywhere. Oh well, I want to see how long they will do it. Maybe there’s an ad opportunity here as well for periodontists.
This is very different than me testing two different gizmos for price points, feature choice or reliability of tongue antennas. I never plan to offer a tongue-only version because I just know it won’t sell.
But let’s just run this test anyways ….
That’s what Facebook did.
It’s the kind of thing that doesn’t jive with this statement:
Your trust is important to us
Facebook’s Data Use Policy.