3 ms·

There's plenty to pick from when it comes to FB's behavior, but...: > FB running psychological experiments in the newsfeed [...] to see if they could affect yo

by avar 8y ago

There's plenty to pick from when it comes to FB's behavior, but...: > FB running psychological experiments in the newsfeed [...] to see if they could affect your mood (definitely not ok) I don't understand how anyone could have a problem with this in principle. At Facebook's scale any change they make to their website will affect people's mood in one way or the other. Would you prefer they didn't do experiments and just rolled out those changes without testing and hoped for the best? That would also be a sort of experiment, the experiment of closing their eyes to data they know is there. The trolley problem doesn't go away just because you close your eyes as you blindly flip the switch. Now, if we find that they ran such an experiment and found it was more likely to make people sad but also to spend more money we'd have a stand to criticize them.

rhcom2 8y ago
It wasn't A/B testing, or testing a feature, it was psychological study (http://www.pnas.org/content/111/24/8788.full http://www.pnas.org/content/111/24/8788.full) whose goal was to manipulate emotions - not measure the effects of Facebook features on emotions. There was also no informed consent for the users.
- dmurray 8y ago
  How can the goal of a study be to manipulate emotions, not to measure the effect of feature X on emotions? I haven't read the paper, but the abstract barely even claims the latter. Honest question.
dcole2929 8y ago
In principle no, but in execution, yeah sorry no. FB completely dropped the ball. When you have that many users you can't just go let me experiment and see if I can make people sad. The range of people on your service is too high for that. All it takes is one suicide that can be accurately tracked back to someone in that segmentation and FB is fucked, deservedly. For something like that, the name of the game is informed consent. Let people opt into or out of the experiment. That's what people conducting an actual study would do. Not just randomly start experimenting on the public which is essentially what FB did

by avar 8y ago

rhcom2 8y ago

It wasn't A/B testing, or testing a feature, it was psychological study (http://www.pnas.org/content/111/24/8788.full http://www.pnas.org/content/111/24/8788.full) whose goal was to manipulate emotions - not measure the effects of Facebook features on emotions. There was also no informed consent for the users.

dmurray 8y ago
How can the goal of a study be to manipulate emotions, not to measure the effect of feature X on emotions? I haven't read the paper, but the abstract barely even claims the latter. Honest question.

dcole2929 8y ago

In principle no, but in execution, yeah sorry no. FB completely dropped the ball. When you have that many users you can't just go let me experiment and see if I can make people sad. The range of people on your service is too high for that. All it takes is one suicide that can be accurately tracked back to someone in that segmentation and FB is fucked, deservedly. For something like that, the name of the game is informed consent. Let people opt into or out of the experiment. That's what people conducting an actual study would do. Not just randomly start experimenting on the public which is essentially what FB did