When your Explore or Identify search result set includes a lot of observations that have GIFs as their #1 image, the loading can be significantly longer, which is an issue for some that have slower connections. Whenever I show iNatters that they can upload GIFs, I always recommend making GIFs the second image (or just not the first) or at least keeping the GIFs in the #1 slot as small as possible.
Personally, I think the extra effort to make and upload a GIF is a good filter to keep ridiculous volumes of video from being added. It’s not just the data volume that is consumed, it is the observers and identifiers time that can be eaten up by video and audio that hasn’t been suitably trimmed and tidied up. When I encounter an image that has not been well framed or tidied up, it is almost instantaneous to come to that determination and to skip past it, although I do from time to time try and give advice to an observer on how to improve them. But for audio and video, you can’t just make that determination straight away. If it is 3 minutes of near useless audio or video, how are you going to know that is the case until you get to the end of it?