Bad news: If the remaining precincts within each California county vote the same way that early-reporting precincts did, Proposition 8 will pass by about 3 million 1.3 million votes.
In fact, the spread would actually get wider. Currently, SF Gate is reporting 52% have voted "yes" and 48% have voted "no." My math suggests that Prop. 8 could pass 55% (yes) to 45% (no).
This is a very basic analysis, so there are lots of reasons to still have hope that Proposition 8 might not pass. Please add more in the comments.
If all registered voters vote, and if unreported precincts within a county follow the same pattern as early precincts, the results would be:
No: 7.9 million
Yes: 9.4 million
You can view the spreadsheet here and edit it (I think) here. (This is my first time publishing Google documents for the world, so hopefully this works.)
I did this at about 10:35-10:50, when about 30-33% of the state had reported, so take the results as a quick snapshot. The percent reporting and the percent voting "yes" and no" come from SF Gate (because the state registrar's website wouldn't load for me), and the number of registered voters come from the state's October 20, 2008 Report of Registration.
There's still hope, since I'm assuming vote-by-mail votes and wealthy precincts probably report earlier and probably skew "yes."
But in the time since I started writing this, the reported gap has widened to 52.5% yes and 47.5% no with 47% reporting. It's not looking good.
(Just before posting this, I saw chrism's diary with the same idea. So, I'm guessing I'll end up deleting this diary, but since we have such different results, I'll post this for now to open the discussion.)
Update (12:30am): I just updated the numbers. Nothing changed much; only two counties changed by more than 3 points (eg, 35% voting "no" earlier and 39% voting "no" now). The new projection overall is 54% yes / 46% no. Here is the 12:30 am version. I'm going to sleep now, but if anyone is still up fretting over the results, feel free to edit the spreadsheet and post an update in the comments.
Also, I mistakenly said 3 million above (I'd fixed a mistake in the spreadsheet but forgotten to change the number in the text). The measure is likely to pass by about 1.3-1.5 million votes, not 3 million.