saying LN is a caching solution is like saying it would be normal if a browser would lock down a picture for the whole internet because it need to use it locally.
[tl;dr ignored]
[tl;dr ignored]
Payment channels only lock down as many Bitcoins as the participants see fit to lock down. The rest of the 21M coins may keep shuffling around without restriction.
A static .gif may be duplicated across the edge of CDNs (caching proxies) as needed; unlike e-cash, there is no double-spending problem for cat memes.
If random small writes are literally the work that kills NAND the most, why do you think random small writes are good for blockchains?
Why not have a distributed layer of ad hoc write caches consolidating and optimizing blockchain commits, especially when the ROI is a TPS increase from ~12tps to basically infinity tps?
If you insist on Bitcoin competing with commercial banking (ie Visa/Paypal/ACH/SEPA) and absolutely must use it for Starbucks lattes, payment channels are the only way to get there from here.
Unlimite_ is vaporware; Core has working code ready to start laying the foundation for scaling Bitcoin to high-powered super-money.
If segwit is implemented and we still have insufficient tps capacity, the Big-blockers will have a much more believable, perhaps compelling, case for an increase to 2MB.
Blocking segwit and LN out of spite while implausibly moaning about how Bitcoin needs to be unlimited is the epitome of cynical hypocrisy.
I admire the cynicism, but abhor the hypocrisy. Looking forward to the Unlimite_ #REKT thread...