You make a very good point there. That’s probably the best reasoning to be against federating with them I’ve seen. It also brings up duplication of content. If they have so many users, would it be possible that a smaller instance getting content from them could literally fill up all their storage space, assuming it’s not a gigantic data center but something self-hosted? Text may not be problematic, but images from Instagram can be massive. I suppose if your instance was that small you’d already be picking and choosing what to federate with, though.