How have perceptions of ResearchGate changed from 2014 to 2018?
I have a Researchgate profile, but I only see my activity decreasing in the near future. To me, the issue with RG is that it only appeals to junior or unknown scholars. While it is great that they now have a way of disseminating their research, RG doesn't help reaching the "big names" because most don't use the platform. Most fields are sufficiently small that people reasonably know each other or can ask someone else to be introduced. Hence, RG doesn't help narrowing the gaps between academic circles of "big names" and "unknown researchers", although this is something urgently necessary.
Additionally, many of the services that RG provides are subpar compared to competitors. I already have a number of ways to keep updated with the current literature in my field, and most of the recommendations that RG sends me are unsurprising or duplicate. The citation/read count is mostly useless, and many services such as Google Scholar and Scopus are much better at that. The platform for including current projects is also fairly inflexible and generally asks to share more than what the researcher is comfortable with---lab webpages are much better at that. I also don't see RG as a reasonable platform for uploading preprints, and like many uncurated venues, most people who do resemble cranks, and I don't use RG to look into someone's publications because it looks messy and incomplete. I use Google Scholar, Scopus or personal webpages for that.
I have never used the "Questions" section of RG, but the few times I glanced it, I expected to see some sort of a forum for academic discussion over some topic, but it seemed more like a watered-down version of SE minus the efficient curating. And most questions remain unanswered.
I do keep my RG updated, and every week or two I receive a request for a pdf of one of my papers. I suppose easing this request process is an advantage over other means of communication, but I don't think many people who use RG attend those requests anyway. I have also never received a request to collaborate through RG, but I have received such requests through e-mail.
I think the rationale behind RG is interesting (i.e., connecting researchers online), but it just doesn't work well, and doesn't appeal to most of the relevant people in (my) community. Collaborating in science is incredibly difficult and time-consuming. I don't see how an online platform that connects people who don't know each other and have no way of knowing how each other work can reasonably help at that.
I do think one aspect of RG is fine, which is post-publication peer-review. But while it does allow people to publicly comment in publications, the platform is also not built for long comments, only back-patting, and most people don't use it for that anyway.
In general, it seems that maintaining your own webpage, even if with much less functionality, appears to be better than having a ResearchGate profile. But two things that RG attempts to implement are worth pursuing: 1) post-publication peer-review, and 2) some way to connect people from the same field but who belong to different academic circles.