Won't that just lead to people making machines with giant racks of RAM, causing the exact same difficulty race as we're seeing with the giant racks of ASCIs in Bitcoin?
And even if that's not the optimization (I'm not an expert on ProtoShares or memory-constrained algorithms), I am highly skeptical of anyone claiming a computation problem is optimally solved by a laptop, and cannot be optimized with custom engineering.
It's about cost versus yield. Custom-designed ASIC chipsets are orders of magnitude more efficient at their intended task compared to general-purpose CPUs; whereas it's impossible to buy or make "special" RAM, and cheap RAM is something everybody wants for everything anyway.
And even if that's not the optimization (I'm not an expert on ProtoShares or memory-constrained algorithms), I am highly skeptical of anyone claiming a computation problem is optimally solved by a laptop, and cannot be optimized with custom engineering.