This article contains two misstatements.
The actual length of a solar year is 365.24+ days rather than 365.25, or about 11 minutes less. That tiny difference is critical to the Gregorian algorithm which the author also muddles with the statement that the calendar makes "... century years that were divisible by 400 exempt (such as the year 2000)
That's reversed: end-of-century years in the Gregorian calendar are exempt from the every-4-years rule when they're NOT
divisible by 400. Thus 1700, 1800, and 1900 were not leap years but 2000 was, and our grandchildren will similarly not have a leap year in 2100.
The divisible-by-400 rule adjusts for those little 11-minute snippets that accumulate over the centuries. The adjustment's still not perfect, but close enough that the current calendar won't get out of sync with the solar year for many millennia to come.
As an IT designer who was heavily involved in my then-employer's Y2K preparations, I encountered all sorts of misunderstandings of this end-of-century subtlety, up to and including a major industry publication with a front-page story on how 2000 would only have 365 days. (And it was an actuarial firm so yes, we really did have to worry about the year 2100 too