Post author here: this post was written in June 2017, and the subject has become increasingly relevant since the MySpace video advertising bot fraud reported by BuzzFeed on October 27, 2017:
https://www.buzzfeed.com/craigsilverman/remember-tom
The conclusion that Facebook has bots other than Facebot is interesting. I wonder if Scott was able to identify how many problematic user agent strings were on Facebook's network.
Great question! There was a fairly even distribution of slow-performing requests across the many user-agent strings coming from the Facebook ASN.
My hunch is that the Facebook crawlers send user-agent strings that Facebook has seen in requests to their own services. This allows them to crawl content in linked posts masquerading as a device that their users would actually use.
> My hunch is that the Facebook crawlers send user-agent strings that Facebook has seen in requests to their own services. This allows them to crawl content in linked posts masquerading as a device that their users would actually use.
Very interesting theory, I’d love a follow up post if this can eventually be confirmed or denied :). That would be very clever of FB but very, very rude to all of the small fish they crawl.