What's worth pointing out is that the claimed timescales (250,000 years, etc.) are wildly out of whack with the empirical data (approx. an 8% decay in only 170 years). The reversals in the magnetic field are reversals in orientation and, as the linked article points out, are distinct from the overall decay in the field strength. Once the field dies down to nothing there is no reason to believe it will ever start back up again. This is the primary finding of the linked article.
I should also point out the reversals happened quickly, not slowly. Humphreys predicted in 1986 that if reversals were rapid substantial portions of a reversal would be recorded in lava flows within the time it took a single lava flow to solidify. Exactly this evidence was found at Steens Mountain (Oregon) in 1989 and subsequently other locations have been reported in the literature, showing the reversals took place in days or weeks. A mechanism for such reversals was later found in the catastrophic plate tectonics work of Dr. John Baumgardner.
We may have a bit more than 1,000 years, too, since the decay is almost certainly logarythmic not linear. Still, the long-range weather forecast is a bit unsettling. ;-p
I have not (and will not) wade through all that has been posted here in detail. However, I will take the time to point out to you that the data reported is simply a weakening of the field. The reversal, if/when it comes, will be nearly instantaneous, and likely be from a much weaker (+) state to a very weak (-) state.
As far as "reason to believe it will start", the most credible reason to believe that is that we've had a pretty strong magnetic field on this planet for most of our history, apparently. The weakening/reversal/etc. is most likely due to the influence of several factors. Go out to your favorite stream, and watch the whirlpools start up, die down and restart. A priori, by your statement, there is no reason to believe they will restart... but fluid dynamics (which certainly plays a big role in the earth's magnetic field) is far more complex than your statement concedes.