Here's what the chart would look like if the data analyst wasn't trying so hard to push their political agenda:
</sarcasm>
But seriously, no, the chart doesn't have to go down to $0 in order to be useful. All you're doing is compressing the data. The original chart makes it very obvious that the manufacturing employment has gone significantly lower than it was back in 1947. Your new chart obscures that, makes it so if I just glance at it, I might come away with the impression that it's close to where it was in 1947, when in fact it's something like 20% lower.
That said, I wouldn't complain if you simply wanted to extend the chart a little downwards, maybe down to $8k or $9k, rather than dropping it to $0.
Your complaint is that the zero'd axis makes it look like 1950 and 2010 aren't actually that much different seems weird to me, because the actual data seems to be that they aren't that different, and showing that is correct (rather than incorrectly hugely magnifying the differences)