Here is a fun experiment.   Put 3x 7990s in a sealed 3U or 4U rackmount chassis and try to keep the cards from burning up.  There is a reason that people use open rigs.  Trying to remove that kind of heat is next to impossible.  Even if BFL delivers the form factor is next to useless because you are never going to be able to fill a rackmount server with 3 to 5 of them.

Ahh. It is strange that their members are over here talking about their organization's matters.

Avalon has zero warranty.  Did it stop anyone then?  Maybe a few...

Because emotion and greed appear to overpower rational thought and impair basic math skills.

a lot of bitcoiners in the area of wa - cheap electricity in certain areas of washington.
it's like mini silicone valley -

Companies choose to operate on the interruptable schedule in exchange for lower rates. 

There is always a contract involved.  Your choice is interruptable schedule for a lower rate OR non-interruptable schedule for a higher rate.
Why would you pay more for power on Bitcoin mining then you have to?

Of course this is mostly academic it would take a very long Bitcoin operation to even be given the choice.

A datacenters will CHOOSE to pay MORE for power in return not having an interrupt provision.
A aluminum smelter (or Bitcoin farm) will CHOOSE to pay LESS for power in return for having an interrupt provision.

The way that I see it, the amount of energy it takes to perform a hash per second on a general purpose device (phone, pad, netbook, etc.) is more important than the hash rate of the latest miners.  If the first number becomes very low, then perhaps billions of users could make Bitcoin essentially unbreakable and make a modest amount on the side (when billions of people are using Bitcoin, even a few satoshi a week would be nothing to spit at).  I would much rather see mass decentralization than a scenario wherein Bitcoin is kept afloat by a few hundred (or fewer?) mining centers. 

I think you are still misunderstanding. You wouldn't want to be under the cap.  Would you be upset to have cheaper power?  
This is something companies contact for BECAUSE they want very cheap power rates.  Cheaper than the rates offered to customers under non-interruptable contracts.

30 days seems really short for a $4500 computer product - don't most warranties on computer equipment last at least a year??  Even if they offered to sell an extended warranty it seems like a better idea then putting it at 10 days!  Seriously - if I had it for 11 days and it started on fire and they told me they wouldn't cover it (even though I was operating in all the "appropriate" conditions) I'd be seriously pissed off!  Thanks for catching this...

Now imagine you were building a massive hashing farm.  Would you be willing to trade (hypothetically) up to 1% annual outage for 30% lower power rate?  Like the aluminum smelter it doesn't matter if you are producing Bitcoins in any given second just how much you can produce in a year and at what cost.

I would love a contract like that on a smaller scale but power companies generally only make contracts like that with the largest of users because if they need to cut 1% of demand it is 1% of demand not necessarily 1% of customers.  If they need to shed 1MW of demand it is easier to disconnect (under contract) one 1 MW customer then it is to disconnect one hundred 10KW customers.

I see the confusing, he was referring to an OPTIONAL contract not the power company just cutting people off.  There are likely customers in your area who do the same thing.  Just using some made up number (large power delivery if often more complex then a fixed kWh rate).  A big power user (think aluminum smelter) might be offered a guaranteed delivery rate of $0.05 per kWh OR they could agree to no more than 1% annual outage and get $0.035 per kWh.  The customer chooses (isn't forced) to accept a possibility of up to 1% downtime in return for a 30% reduction in electrical rates.  For a business like aluminum smelting where you production cost is mostly electricity this is a beneficial arrangement for BOTH the customer and the power company.

What does hydro have to do with it? The interruptions come at times of unusually high load. Typically in mid winter when there will be high loads for heating and lighting then, during the morning peak demand (cooking breakfast and heating water for coffee and showers) or in the early evening peak (cooking dinner or the whole country turning on the kettle when there is an ad break on TV) the network may simply not have the capacity to deliver the full demand and so the high consumption customers who have agreed to interruption at such periods will be turned off. In hotter places, the peak demand will come on summer afternoons as all the air-con systems work flat out.

I have worked in the electricity supply industry in countries that have hydro power and they have contracts that allow power companies to interrupt the supply to large users.

All power grids have to deal with DEMAND > CAPACITY.
You either
a) use very expensive (in terms of cost per KW/h) "peaking plants" normally natural gas turbines because they can rapidly adjust load
b) you offer your largest customers a deal where they have to idle/scale back (they often do this because b is cheaper than a)
c) you have brownouts and destroy electronic equipment on a masive scale
d) you have rolling blackouts

A hydro-electric dam has a max capacity.  No power grid runs ONLY on hydro-electric dams and even if they did if the total instantaneous demand is greater than the total power grid capacity then something has to give.  Power companies have found that contracting to lower the load on large users is often the cheapest solution.   If 99% of the time your customers consume <10 GW of power and the other 1% of the time it can spike to 11 GW why build out an extra 1GW of capacity which is only used 1% of the time.  The use of hydro electric power doesn't change that dynamic.

Even with other fees, it's well under $0.05/kwh in certain areas.

As for power interruptions, high power consumption sites (industrial loads) can get substantially cheaper power in some areas if they agree to load shed a few times a year (typically for minutes to hours) and basically turn off the main power consumption items to reduce grid load.  You have to be sucking a LOT of power to do this, but in exchange for the power company being able to turn you off with little to no warning, you can get a much reduced power rate.  Bitcoin mining being interrupted doesn't affect it other than the loss of mining capacity during those times, so allowing it to be interrupted and restarted for a reduced power rate year round is likely to be a net win as well.

If you are looking for the cheapest electricity costs you should keep in mind that this is just a percentage of the total cost per kwh. The other costs involved are often much bigger than the electricity itself.

If we look at the Belgian electricity cost, at first glance you could think it is cheap, around 5 to 8 € cents. If you dig a little deeper you will find out this is only about 35% of the total electricity bill. You have also distribution costs, transportation costs, green energy costs, taxes, fixed costs etc This will make the total cost of 1 kwh about 18 to 22 € cent, a world of difference.

So is the 2-3 cents the total cost of 1 kwh?

Power interruptions...big difference between (milli)seconds and hours... Can you explain a little more?

Bitcoin mining went through this sort of thing a while back when the price crashed down to $2.50. Certain GPU rigs were now costing more in electricity to run than they were generating in Bitcoin. However, the price rose again and there was new life (very long life as it turned out) for those GPU mining rigs.

Another rise in the exchange rate could buy more time for obsolete hardware.