Friedcat will you be producing any small scale miners in the future like the BFL Jalapeno?
In business aspect, we will learn from the market what miners want the most.
There are some practical limits. E.g. one target spec is about 1200W-1500W per device, because that's what most US households can deliver per breaker (10A) (probably 1200W to allow for some buffer). Other countries may allow more or less. Dependent on the target customer base (US might not be the majority
) that number may have to be scaled.
Above 1.5kW you're looking at custom power supply installations, where usually an expert has to be involved.
Then there's a sweet spot for the price of course. Not everybody wants to invest >$1000. Maybe $200. But that's easy to determine. Just ask the customer.
However, as far as the small scale miners are concerned, I think the defining feature is mobility and plug&playability. This is deduced via inversion: If the miner would be concerned about efficiency and the economy of mining, he'd opt for a solution which maximizes Hashpower / equipment cost, which tends to reduce for smaller devices... , it's the same reason why desktop computers provide more bang for the buck than laptops. For that very reason a lot of large scale miners don't do any casings at all... it just cuts into their competitive advantage.
Now the interesting question is why would people opt for a small scale mining device, knowing that it provides less MH/s/$ (just look at the BFL pricing, 23MH/s/$ vs. 30 MH/s/$). I just have to assume that people sacrifice 7 MH/s/$ for the mobility factor.