The simplest way is to derive it from the hash target (difficulty).
If you take the present target value
0000000000800e00000000000000000000000000000000000000000000000000
and divide it by the total search space (all 0xF), you get the odds.
In decimal: 5.26786429665E+64 divided by the total search size (2^256) shows you have a one in 2.19e12 chance of finding a new block with every hash try.
We know that the network, as a whole is guided toward 6/hr, and the difficulty reflects that. If the average generation rate is one every 10 minutes (600 seconds) then we know that the network as a whole is trying
3,663,473,554
hash attempts per second.
So the aggregate network power is around 3,663,473 khash/second. Call it 3.7 gigahash/sec.
Further, if my machine is plugging along at 4,000 khash/sec, then I know I represent about 0.1% of the active network, and have a 1/1000 chance of finding "the next block".
If you use the *actual* recent block find rates (instead of the 10 minute average), which you currently track and graph, then you can see the interesting graph over time of how many total attempts likely produced that result.
I believe the 1/x characteristic would be more interesting to watch for trends than the seconds/block you currently graph, since the number of seconds is forced into a normalized 600 over time. Total network strength should show an always-upward trend, with some notable spikes, dips, and steps.
