I think your first line is far closer to the truth of any ROI calculations than the link

Nobody knows how long it'll take for these to ROI (if ever) simply because nobody knows what's going to happen to USD/BTC exchange (if calculating ROI by fiat), nor the difficulty (if calculating ROI by either fiat or BTC).

But with the link in place, which is a highly optimistic estimate, there's specifically the issue of the difficulty.  Right now it suggests break even by 2015/06/13.  However, the next difficulty jump is in just 3 days.. (which is probably about as fast as you could get these sent your way if they even shipped right now), so click the 'next estimate' button, re-calculate, and hey presto... no more break even.

Additionally, if you bought $650 worth of Bitcoin right now at ~$400/BTC, you'd get 1.625BTC.  According to the original calculation, it fizzles out after 2015/12/10, having given 2.44433BTC in revenue, or 0.81933BTC profit.  At that time, the exchange is $833/BTC, so fiat profit would be $682.50.  On the other hand, that 1.625BTC would at that time be worth $1,353, giving a $703.60 profit.
( It's actually a bit worse for the harware, as this excludes operating costs. )

So not only does that scenario seem doomed to fail due to the realities of the impending difficulty increase, it's doomed to fail because you'd be better off buying and holding the BTC.

As much fun as these calculators are for giving lower/upper bounds based on some numbers one plugs in, they're still inferior to a proper crystal ball 

( The above is not a commentary on whether or not these will ROI - I'm with jimmothy on that one, get your own crystal ball. )

Neither - he's just handing out a healthy dose of reality; people asking about ROI need to do their own calculations/guesstimates.  His initial statement holds true, regardless... even if it never ROIs, it never ROIs faster than anything else out there due to the favorable price/GHash and W/Ghash.  Or is that slower... hmm.. brainhurt.