oh yeah, that makes a lot more sense. But then it makes even less sense. It makes about as much sense as those last two sentences :-P Let me explain. For this last period of 2016 blocks, the mining speed appears to have been overall more above the line than below it. So that means 2016 blocks should have gotten done faster. And yet on the 2nd graph on their page, it's obvious that it's been a few extra days compared the last 30 or so difficulty adjustments but that "above the baseline" increase is massive. How can that be?
Which is why difficulty is projected to INCREASE 9% however it won't happen until block 169344 which is projected to be on 03/02/2012.http://dot-bit.org/tools/nextDifficulty.php
Difficulty only adjusts every 2016 blocks. Hashrate could double the block after a difficulty adjustment BUT the next adjustment won't occur for 2016 blocks.
If the hashrate doubles, the 2016 blocks get found twice as quickly at an average rate of 5 minutes per block so the difficulty adjustment comes twice as soon. So since the hashrate has been higher than the difficulty line for the majority of this set of 2016, the adjustment should come soon than others. Which leads me to the next answer for the above post which is, the current difficulty (as rendered) is 39 pixels long thus far. The last 2016 was 18 pixels. The two before that were 19 pixels. Even one of the periods on a massive downward slope was only 23 pixels long. So 39 pixels is a long time! We're almost double overdue actually.
EDIT: oh, except that's not a straight line if you zoom up to full res, lol. Okay, not as inaccurate as I thought. It's just a downsampling problem in the mini-graph.