1. A git-aware Google Drive, where I could stick version-controlled code I wanted to sync between computers and possibly share with colleagues (and also create a paper trail if someone beats me to the bunch publishing something, ha)
2. As a CDN serving open-source projects I depended on, but discovered almost exclusively by other means
None of my GitHub-hosted code ever had a README.md, I never starred anything, and I often never even visited the repos I depended on (their documentation would be hosted elsewhere and copying-and-pasting the GH URL from there was all I needed to do to depend on it). If I visited an Issues page, it was because I got there through googling an error message, not because I went there through GH's UI.
This Drive / CDN usage pattern is sort of the baseline, "passive" level of engagement with GH imo.
There are two sides to github consumption - active and passive.
On passive side - we are happy to stick to basics of git on CLI / IDE and get the job done of a version control. And the site is usually to configure CI/CD and so on.
However on active consumption of open source - it is all about discovery of repos relevant to tech you are into and people behind them. It is so amazing to go and find a good committer profile and see what repos he follows.
And the new github trending is simply axing their own feet here.