Far below what i expected.

My prediction is only a few percent off, not bad.

Bitcoin difficulty just hit 148,819,200, and the next bitcoin difficulty is likely to be ~200,000,000! 

Do you know where your miner is?

Difficulty and hashrate and perfectly linear.   If network hash rate doubles, then difficulty doubles (eventually), and the GROSS return on x GH/s is cut perfectly in half.  The net return is reduced by more than 50% because the electrical cost remains fixed.

Based on existing (not future) pre-orders we are looking at 6 PH/s min by end of Dec.  If you want to be more conservative you should consider 8 PH/s.  

1 PH/s ~= 140 mil difficulty
1 billion difficulty ~= 7 PH/s

If you want the exact formulas

hashrate = (2^32 * difficulty) 600
difficulty = (600 * hashrate) 2^32

Remember the network/protocol has no way on "knowing" the hashrate (and neither do any sites which chart it).  We can only estimate the hashrate based on the frequency that blocks are found.

^^Puppet = ActiveMiner fanboi. 
Hopefully that puts his reasoning in perspective

It is more complex than that.  Lets look forward a little bit.

My guess is that by end of Q1 2014 we are probably looking at 16 Ph/s.  That would be difficulty of 2.2 billion.

At this point ROI will be significantly reduced.  Even if ASIC companies slash prices to the bone electricity will be making up a larger and larger portion of gross revenue.  An Avalon miner at $0.10 per kWh will be spending 80% of gross revenue on electricity.  All of these things will weigh into future sales.  The next 16 PH/s isn't going to be "easy money" even optimistically the ROI% (using the term correctly) will be measured in % PER YEAR.

During that slow climb from difficulty 2.5 billion to difficulty 5.0 billion the more efficient devices will allow miners to keep a greater portion of the gross revenue.

For example at difficulty 5 billion, $0.10 per kWh, and current exchange rate:
Avalon & AsicMiner - negative operating revenue.  would need electricity <$0.056 just to break even ($1 in electricity produces $1 in coins)
BFL (65nm) - roughly break even higher difficulty will require going idle or selling unit to someone with lower costs
KNC - power cost is 36% of gross revenue
Bitfury - power cost is 18% of gross revenue
Other 28nm players - power cost is 17% of gross revenue

Note at this point I don't think it makes sense for anyone to place any NEW order (analyzing existing orders is more complex because a lot depends on early growth, timing, etc).  It is very likely every single company will be forced to cut prices 50% to 75% or more over the next three months to move units.  Since new units aren't going to reach break even in the next 90 days it makes more sense to wait for the inevitable price drops.

https://bt.irlbtc.com/view/281279.0

The simple version.  Sure if you think difficulty will grow exponentially (doubling every 2 months) until we hit 50 billion or so and everyone even a Conterra miner with no AC costs and $0.05 per kWh electricity is mining at double digit losses then sure efficiency doesn't matter.  However the more realistic scenario is that as difficulty piles up and the "day 0" returns drop sales will slow.  ASIC providers can cut the costs of the chips but that only means efficiency becomes even MORE important.  There will be an inflection point where the growth curve flattens.   The laws of physics ensure that will happen otherwise in roughly 4 years mining will use more energy than the entire human race uses for all other purposes combined.

I can't argue with your conclusion (and already read the thread you linked & your thread estimating hashrate based on pre-orders -- thanks).  I think we disagree on minor details -- i think the irrational component, and inability to accurately predict the difficulty three months in advance (the average pre-order time -- i might be off), play a greater role in the hashrate increase.  It's not simple math -- it's estimating the ability of others to estimate the ability of others (...) to do simple math.  

So, looking into the future, the best profitability would be to buy when the cost of miners is just a bit more than cost of production and the manufacturers are about to quit.  But, somewhere in the future before that, the cost of buying then presently available miners which can truly see ROI within 12 months might happen?  One might assume that would be within 12 months of one of more major manufacturers permanently halting production.

I think the missing factor in all of these analyses of ROI is how much of the daily volume on all the currency exchanges combined comes from mining output. Above a certain threshold and the price is influenced by the costs of entering the mining arena. In other words, miners can convert their gains to fiat with limit orders and as the cost of mining goes up or the amount of bitcoins received per miner goes down, surely some of those miners will insist on a higher ask price?

Obviously markets are chaotic and unpredictable, but I haven't seen any decent analysis of how much an impact the mining community does have on the exchange rates. Any theories or evidence?