if we wait just a little longer, it won't be necessary to scale because network growth will enter a reversal.
It would seem that temporary reversals in the network "data rate" growth are necessary steps intended to keep the validation costs sustainable (within a certain corridor) over the decades to come. We will have to accept whatever other effects come from it in order to preserve this one fundamental attribute of the system.
As long as the block size remains such that the blockchain can be passed along on a single and affordable consumer-grade storage device, I'm mostly happy.
I've been obsessed with the data rate and how to mitigate the weaknesses high data rates expose since I first read the whitepaper.
I agree with this, though I would consider the "data rate" to be more vulnerable characteristic of the network than "storage costs", due to the fact that some ISPs might (and do) impose "total bandwidth caps per month" limitations having direct impact on running a full node continuously, while storage devices can be accumulated without any limit.
Regarding the Blockstream and subordinate chains I would give them a benefit of a doubt, simply because I haven't been following the developments there and cannot properly assess the proposals on their technical merit at the moment. On the other hand, if the protocol begins "endorsing" the accumulation of unlimited amounts of bitcoins in third-party services (subordinate chains), while the bandwidth in the main chain remains limited and the costs high, the situation might produce what I call systemic risks and can severely shatter the trust for the whole solution.
I think, Paul Sztorc mentioned in his "Measuring Decentralization" article something about custodial periods of about a week for bitcoins to be locked to a particular side chain, maybe that would work or maybe it will become cost prohibitive, hard to predict at the moment. But regardless of the above, it's pretty clear that the main chain needs to remain sustainable (cost-wise) over the long periods of time and we have to accept whatever other effects this particular attribute brings.
