From ztek thread it is <8.5W.  Roughly 22MH/W.  Make my "efficient" 2.6MH/W GPU rig look like a coal guzzling monster.

After a boatload of money invested in task-specific hw that you could not resell if BTC fails.
And at this price 0.21 USD/24h@100MHash/s http://bitcoinx.com/charts/ means that you produce about 0.33 USD each 24 hours with that, not counting electricity.  That means about 100 USD/year, so maybe 4-5 years before recovering the investment.

Its worse than that. Keep in mind bitcoin generation will have halved twice in that time span.

I think thats a given; mining at a loss or "free" electricity. BTC has lost over 90% of its value since the peak, hash rate and difficulty havent even halved.
It be surprised if more than 10% of that hash power was profitable at actual electricity rates.

Much depends on how many "big" miners there are vs small ones. I suspect most of the big ones will either be profitable or shut down, since any losses would also be far bigger for them; but Im even more certain there are many more miners  who mine with one or two cards, but operate at a loss, stealing electricity, have their parents/employer/landlord pay for it, speculate on future gains, or are just being plain irrational (or terrible at math, like that other dude who could double bitcoins).

The question is, how is hashrate divided between those groups? That I do not know. I would have guesssed something like half the hashrate is provided by casual miners with 2 or less GPUs, but looking at bitminters statistics for the last block (yeah, finally!), that might be mistaken, as the majority of hashrate seems to be provided by at least 3 GPUs.
Not sure if deepbit would give a similar picture though.

Network hashing power tends to lag price action.   Miners tend to convince themselves that prices (and thus profits) will rebound.  The cost of power is also delayed since power bills tend to come once a month.

Also saying "the 30% price drop since Nov 1" isn't a good measure.  The price drop as on 4 days ago was only about 5% and it is hard to call that a "drop" and not just part of the day to day price fluctation.  The significant drop (most of that 30%) happened less than a week ago.  If the price is still <$3.00 lets look at hashing power on Dec 14th (30 days since "the drop").

You are right about the weather though.  Most of hashing power (taking a guess based on node density) is in the Northern Hemisphere and thus the colder temps mean higher profitability.  Miners who would be unprofitable in summer (at current price/difficulty) can still be marginally profitable in the winter.

My expectation is hashing power is roughly 2TH to 3TH lower by next August.

I assume that could also be because its slower?
165 MH is a bit of a bummer though. That will take forever to pay itself back, I hope you bought it more for fun than for financial gain.

All the hardware money spent is gone for me. I look at it as a hobby. I mine because i like the idea of bitcoin. At the moment i burn money but with only 3 GPUs and the board i can stop whenever i want without hard feelings.

@DAT FPGA Board: Accepted 7098, Rejected 60
         5850s: Accepted 309674, Rejected 393

Website updated. Here's what's new:

• My account -> account details: Account info rearranged. Looks better now, I think.
• My account -> banners & avatars: Finally got dynamic banners/avatars!
• Statistics -> luck: See our luck (CDF) from block to block.
• Statistics -> rewards: See how our PPLNS payout compares to the expected average reward.
• Statistics -> API: Started on a simple web service API. But I think some changes are necessary to make it more generic and "future-proof". Consider it a beta that will probably change.
• Statistics -> Live stats, full size: Now shows the 10 latest blocks and shifts. They update whenever an update of round data indicates a change. There is a checkbox to turn on audio notifications of new blocks (off by default whenever you open that page).

If something looks funny, try clearing your browser cache and reloading the page. Some files changed that are typically cached.

Now I can have a look at the client (miner) again. Sorry for the delay, but I had to finish this web update first. Even though I'll be working on the miner next, feel free to make new suggestions for the website as well. They will go on my neverending TODO list.

Update: If you already had live stats open, please reload the page. I made some changes to the data format and the old page can no longer read its data feed correctly.

The live stats full size? What is not working with it?