Metrics for Facebook’s Autobot

I believe every of Facebook developer, especially the spammers, would be more than happy to know the metrics for Facebook AutoBot that have been going crazy in banning apps that are fully complaint to Facebook’s TOS.

Well, so far, what we know is that the AutoBot will be catching spam apps by looking at the negative user feedback in a relative terms instead of absolute figures.  This is what Douglas has replied to me before.

There is also people asking for that in the dev Fourm.  Don’t expect to get an official answer/feedback there.

FYI, I have raised the issue to Facebook team by drawing Cat Lee’s attention on the subject.  This is what I have commented,

I suppose we need to know a “reference value”, e.g. at what % of feedback is negative will our app be banned. e.g. I am pretty sure that 0.1% of feedback is negative shouldn’t affect our app and 99% of feedback is bad will affect it. If that is the case, I can change my app to provide different user experience. But how about 1%, 5%, 10%, 15%, 25%? 30%? shall we change our app to provide different user exp at these level?

Just in case the team think it is not a good idea to disclose “the expected/accepted level” before an app is banned, I would suggest giving us some relative figure. e.g. “% of post hidden – 5% below average” (the “average” is talking about the average of all apps in the fb world, we can then know how our own apps perform when comparing to other apps in the worlds)

It is good that Cat has replied positively with

Hi Eddie – Good feedback. Will pass back to the team as they’re iterating on the new App Insights.

For the details on the discussion, see the comments at

This entry was posted in Facebook Dev Forum Digest and tagged , . Bookmark the permalink.

One Response to Metrics for Facebook’s Autobot

  1. Pingback: Metrics for Facebook’s Autobot | Wing's Blog on Facebook Development & Virtualization

Leave a Reply

Your email address will not be published. Required fields are marked *