We are looking at block chain that exceeds a Terabyte easily. Might take a year to download it if we are talking about decentralization and accommodating slower connections.
I don't believe you need to download the entire blockchain to a mobile device
I am referring to if you wanted full nodes to be plentiful and have the ability pop in and out of the network spontaneously, in order to defeat for example blacklisting by the authorities.
Perhaps these nodes could download only what had changed since the last time they were online (or they keep another connection always on to stay up to date and only pop in and out for their node presence). But even so, the sizes computed below are intractable for those who don't have a server farm of hard drives.
, and we now have some real world usage data showing that Monero (ring sig plus denominations) vs Bitcoin for typical usage is not orders of magnitude (it might be around 5x).
I don't understand how you computed that metric? Even if your standard denominations are powers-of-2 from $1 to $1024, that is 10 denominations. And then you need to multiply the signature sizes by the number of inputs which determines the level of mixing. Anything less than say 16 inputs is not much statistical anonymity.
But putting those aside (it is certainly true that widespread adoption means much higher transaction volumes), a terabyte is not necessarily a show stopper. I get about 2 megabytes/sec on my mobile devices. That is 6 days to download, not a year. Soon this will be considered slow. I've heard talk about gigabit mobile (50x faster than mine -- download 1 TB in hours).
You mean 2 megaBITS/sec, thus 48 days.
Bitcoin is at 60,000 tx per day and I think Visa + Mastercard is 6,000 per sec. So multiply the 10GB for Bitcoin blockchain by 8000 (= 80 Terabytes) just to get to current commerce. And Visa + MC don't do micro transactions. With micro transactions we would be in the Petabyte range.
So even if your bloat is only a constant factor less than an order-of-magnitude, the real problem is one-time ring signatures defeat the mini blockchain design (because you can't know which address has which balance with ring signatures).
Edit: I think there is some effort underway looking into how to prune the Bitcoin blockchain, but afaics this won't be possible with ring signatures because you don't know which inputs were spent. There are some heuristics that could be applied, but it looked very messy when I thought about it a bit. One of the mixed inputs not spent can cascade and prevent the pruning of all those mixed with.