Hi Zooko,
Thanks for your thoughtful comments. I appreciate the perspective you’re taking; at least, I can identify with it and the process you’re working through. I also appreciate that you aren’t advocating for a specific decision yet! That’s a remarkable sign, IMHO, of nascent leadership.
I would just like to contribute my own perspective to the cacophony of voices.
First, I decidedly prefer PoW security over PoS security because PoW allows one to directly convert work to wealth. I have always gotten better results by acknowledging people’s personal hard work (e.g. Carol Dweck’s research on fixed/incremental intelligence) rather than past success, and I think that PoW has the same net effect. By encouraging hard work, people are encouraged to continue contributing, and contribution breeds more contribution. It’s an enticement to become a contributor!
Second, I believe the economics surrounding ASIC are a problem, rather than the ASICs themselves (though admittedly the two are difficult to separate). ASIC mining equipment is both a blessing and a curse. On one hand, ASIC mining increases the hashrate dramatically, securing the blockchain against bad actors by increasing the work required to rewrite the chain, and that means stability in the form of resistance to rewrites. On the other hand, ASIC manufacturing creates natural monopolies. The first person to develop an ASIC is so rewarded for doing so that they accumulate rewards which allow them to remain at the forefront of development, always controlling the supply and distribution of contribution prerequisites. This is, IMO, an insecurity in the form of a single-point-of-failure. Political pressure can easily be brought to bear on the manufacturer, manufacturers may be motivated to simply accumulate rewards rather than distribute the hardware, etc.
Third, the argument that ASIC manufacturing has an unrecoverable investment cost is, I think, poorly founded. It is eliminated by sufficiently powerful hardware, undistributed, that the manufacturer can recoup the investment quickly through their own private mining efforts. The Monero experience suggests this is the optimal economic strategy for manufacturers.
GPU mining is a unique blend of specialized and generalized manufacturing. The required hardware can easily be redeployed to other ends, which decreases the manufacturer’s reliance on monopolization to recoup investment costs. Barriers to investment in hardware are low but scalable, allowing any size of investment (much like Zcash value itself, divisible down to less than the smallest unit of, say, American dollars). GPU mining uniquely allows numerous non-participants can quickly redeploy their hardware investments for other reasons to mining Zcash. An investment in neural network hardware can be redirected to secure the currency such that economic investment in Zcash security is itself fungible! If something happens such that political instability causes a disruption to the power supply for large miners, non-miners who use Zcash value for commerce could redirect their hardware to temporarily make up for the gap! That alone is a neat idea (to me).
What I would personally like to see is the democratization of photolithographic manufacturing! An open-source ASIC that anyone could build could be a boon for numerous reasons, decentralizing both manufacturing and participation.
I’ve certainly omitted important points for this conversation, but it is my hope that any random person can decide that they want to help secure the Zcash network by mining on their home computer when they’re not playing games, training neural networks, or doing really fast matrix multiplcations. It’s a neat idea that encourages adoption and participation.
Anyway, thanks for writing. I really like the zk-SNARK work Zcash has contributed and nurtured. So, if nothing else, thanks for that. 
c.p.
