Yeah when I showed the cop the graph of my speed before getting in my car to be 67000mph (speed of the earth around the sun) to 67080mphwhen I was driving it he couldn't see the difference so I didn't get the ticket.
Or sometimes choosing a common-sense reference makes sense.
Which isn't to say THIS one does, it doesn't, but the absolutism of "it's nerf or nothing" is a tad extreme.
What we're interested in is not the number of users, but the trends: whether the number is increasing or decreasing over time. Starting the axis at 0 would not be useful in this regard, as the trend would be almost completely obscured.
If the goal is to visualize growth trends, I don't think raw user counts are the correct value to track on the Y-axis at all. That's where my head was at when I said it doesn't make sense. Abusing the Y-axis to try and coax data out in this case is just a symptom of having the wrong measure.
At some point recognisability is also worth something. I can immediately read this graph, I understand it, it's good.
Occasionally it's used in a confusing way where people assume it starts at zero despite it not being the case, and sometimes intentionally so. But that's just the case here.