If BFL and Avalon can't provide significant hashpower to the community, ASICMINER may be forced to either hold back hashpower for a while or push the date for creating customer ready devices. Anything above a 1/3 target for the hashpower hashrate is thin ice because of the fleeting trust in the decentralized nature of bitcoin.
I could not agree more! ASICMINER could adjust hashrate with a feedback loop to keep it at a constant level of below 50% with some safety margin.
What does that help? Unless someone checks how many of their chips are ready for activation, they'd still de facto have 51% control they can enable at will. Bitcoin is not decentralized as soon as they have the
ability to control the network and thereby do things like undo transactions or require arbitrary fees for block chain inclusion.
The obvious solution would be that if Avalon and BFL have still not shipped in bulk by the time ASICMINER has their additional 50TH/s coming out of the fab, they could keep their existing 12TH/s and begin offering that 50TH/s for public sale as they talked about during their IPO. Their design might not be conducive to selling 60GH/s to individuals, but even if they ship full rack kits it would preserve decentralization without negatively impacting their bottom line.