2013 + 1600 = 3613.
When I did the calculations over thirty years ago (when teaching about magnetism), I had calculated roughly the same dates (3500 plus or minus a few hundred).
The sentence
"At this rate of decrease, the field would reach zero in about 1600 years."
is misleading
Earth's magnetic field is a collection of many fields, and we often model it using two "categories": the dipole field and the random field.
This calculation only deals with the dipole component, not the other one.
The dipole field seems to reset itself after each event, and the magnetic moment seems to reset around 10 units. (Last time it reset was 41,000 years ago, and the entire field reset value was closer to 11).
The moment is presently roughly around 7 (or 7.5, depending who you ask) and the next "event" will likely begin when the moment is down to 4 units (which seems to be what happened on previous occasions). The dipole component represents roughly 60% of the total field, while the "chaotic" component is the other 40%
The discussion (about the decrease of 6.3% per century) seems to deal only with the dipole portion.
The dipole field is easy to "imagine" as a giant bar magnet near the centre of Earth. The chaotic field is a bit tougher to imagine. The best I can do is this: imagine a giant bag full of fridge magnets placed at random in the bag. There will definitely be a magnetic field around the bag, but is will not be well oriented... it will be chaotic.