Not agree. Idea isn't in just adding memory requirements to the POW but making the POW seriously memory-constrained. Say 1 POW needs 1 million sequential computations and 32M of RAM at a whole for each computation and that 32M can't be shared between POWs because there are different data. So the only way to implement such POW in ASIC is to add 32M of RAM to the chip wasting its area and dramatically increasing its cost and reducing its efficiency. And you can't make 32M of on-ASIC RAM cheaper than 32M of DDR module RAM.
1) OK, it is usefull to make ASIC less effective. But it is not enough. At least, they can use 20,000-40,000 GPUs to make the 51% attack. We need one more solution for this case.
2) What new POW do you suggest? Your scheme with N=32M/32=1M ? IMHO, it is not good because calculation of 2^20 sha256 is too time-consuming. Each node (not only miners) needs to calculate POW each time when recieves a new block. Do you have another one idea? We can remember results of intermediate steps 0,8,16,24,32,40,48,63 of sha256 calculation, for example. It is 8 times more information,
but 2^17 calculations of sha256 is very time-consuming too.
Bad idea. No one can prevent me from making millions of nodes each of which trusts to each other. Newly connected nodes have to trust my malicious sub-network with high probability only because of its size.
If you propose to dedicate one bootstrap node and make it trusted by default (hard-coding certificate into client for example) you just invent PKI in its traditional form and that trusted-by-default node would become a central authority and would perform central-bank like functions. That's not we're all want to happen with Bitcoin.
Yes, if we will make "net of trust", then an attacker with more then 50% of computational power can create "alternative reality" in bitcoin, i mean alternative block chain, that will begin with the same generic block but then diverses with the "real" block chain. In fact, in this case can exist more then two block chains. And for a node that connects to the net, there will be impossible understand wich of them to connect and belief. IMHO, this problem can be solved only by out-of-bitcoin methods. All sites that work with bitcoin may\have to publish in what reality\block chain they work. I belief that they can detect which block chain is real. And usual users will check from time to time that they live in the same reality. It is not excellent, but I don't see any better solution. At least, by this approach it is possible to eliminate other problems of 50%+ attack:
Reverse transactions that the atacker sends while he's in control
Prevent some or all transactions from gaining any confirmations
Prevent some or all other generators from getting any generations