Because growth curves will always follow an exponential (although that's actually extremely rare in real life conditions) or sub-exponential (i.e. polynomial - which a quadratic is) trend over a few days - or hours for things that multiply real fast, like bacteria.
There is absolutely nothing surprising about this.
It's incredibly easy to take a handful of data points and find a really really good fit, in particular with a polynomial function, ESPECIALLY when it's overfitted - which you can see in that reddit post given that it does not trend towards 0 but instead curves upwards.
The next few data points will inevitably fall right around the trend curve - again, there's nothing surprising or dodgy about this at all. They're no going to suddenly wildly diverge from the trend established from the prior few days.
It will however invariably fall part really quickly, which is starting to be the case for this example.
But of course, instead of thinking "well maybe it was just really easy to have a really good fit for a few days, but that's not the case anymore because this is how these things fucking work", now we're seeing the classic "THEY MUST HAVE NOTICED US AND ARE CHANGING THE NUMBERS TO MISLEAD US".
It's pretty much exactly like flat earth distinguished mentally-challenged fellows who demand proof, and when you show them pictures of the earth taken from the moon or satellites that are far away enough to capture the whole thing, their answer is "oh but those pics are CGI".