> Which isn't many, but it might have been critical to those 4.

http://xkcd.com/1172/

I understand it as 4 in a million are clicking on video responses. People creating video responses as self-gratification might care a lot about it..

I think the 90-9-1 rule applies.

The reach of those 1% is pretty substantial.

I have never used the feature in question, but I'd certainly think twice about removing any feature that 1% used.

If the click through rate is 0.0004%, I don't think the power users will care all that much. They're probably not getting any significant source of views from video responses. The more powerful tactic on YouTube seems to be linking to other videos at the end of your video.

Content creators on Youtube have a bootstrap problem. Once their videos become popular enough, Youtube's content discovery mechanism will direct potentially interested viewers to said videos. The problem is that if no one has seen the video, then the algorithm has no idea what its quality is, or what type of viewer would be interested. Depending on your target audience, even getting a small number of views by piggybacking a popular channel may be enough to get you onto the algorithms radar, which generates a lot more views that the click through rate would suggest.

Linking within the video works, but it requires that the creator of the original video does more work, and if it was not planned when the video was created then, while doable with annotations, it appears scammy.

I was referring to linking in the video, but you make a good point about wanting to link to videos made after the fact.

While piggybacking may be an option for garnering initial views, using it for this purpose encourages response spam. Using good 'ol search optimization should really be the first option for content creators looking to get off the ground.

'Which isn't many, but it might have been critical to those 4.'

Well yeah. But you can say that for nearly any feature of program. I don't see this as a big deal.

It's such a small number that it makes complete sense to remove that feature. Removing feature that may significantly lower user experience on 0.0004% visits while slightly improving the user experience on 99.9996% visits? Quite obviously good decision.