It also says it was triggered by a freak event which happened all at once.
Sudden Arctic sea ice melt could do it if (1) the whole thing went at once, which isn't unlikely for floating ice on water, it's a phase change kinda thang, and (2) the wind was blowing all those bergs into the Atlantic where the had a mighty melty dump - which is much less likely; odds are they'd piffle out en route and in situ.
Greenland ice sheet melt would do it and the current offers on the Copenhagen table are at the 'worst case' end of the scale, anticipating 4-7 degrees of warming, at which you'd get that melt easily.
It's the
suddenness of the change in salinity which is reckoned to have caused the rapidity of the drop in temperature. Without that suddenness, caused by the freak event I referred to, the conveyor would reduce over years, according to the models. I questioned the likelihood of a large enough part of the Greenland ice cap melting in a single summer - and nothing you've said has caused me to re-think that.
Forget the sea ice. The Greenland ice cap contains much more ice, because although it's a smaller area, it's a thousand times (2.3km vs a few metres) thicker than sea ice. Also, it's 100% fresh water, while sea ice is frozen seawater, complete with some of its salt. It starts out at 30% of the salinity of seawater. Old ice has much less salt in it, but the increasing melting of ice each summer in recent years has reduced the proportion of old ice. Only 10% of spring ice cover is now more than 2 years old, down from 30% a decade ago. Allowing for reduced overall ice cover, 75% of the old ice has been lost in the last 20 years without switching off the conveyor. I doubt the last 25% would do it, however quickly it melted.