The past 20 years has seen an explosion in Internet services, but a fundamental (often unspoken) quality of these services is that they must keep individualized customer support costs very low in order to be profitable. I mean, just look at Google, which has billions of end users. You can argue that Google takes in a ton of money, but it's not hard to see that if every Google user just had a single support call requiring 30 minutes of a support rep's time a year, that Google's profitability would tank (never mind the difficulty in hiring that many reps in the first place).
So all these services invest a ton into automated support and systems to ensure the vast majority of users never need support from a real human. That usually works well, except when you get some of these edge cases, and, very importantly, these edge cases are usually the worst when the customer behavior, while totally legit, "looks" pretty similar to malicious usage.
So, in that case, I'm glad that these back-channel avenues still exist when someone gets stuck in the machine. I wish there were a better alternative, but I really am at a loss to think what that could be. The social media channel essentially acts as a filter, as only things that are real problems are likely to get upvotes or lots of visibility. A trade off for users being able to get tons of value for (relatively) very cheap prices is that the "tail end" of support requests is usually pretty nightmarish.