i'm not convinced on the whole "infinite scale" mantra here, i mean at some point a payment channel must settle, and the settlement is still limited to (at most) 1500txps, which is, frankly, not much in the global scheme of things (plus one must assume spam/unimportant traffic)
now granted, you might have clever ways around it, i dunno - currency fx hedging techniques and so on, but still, it's a bottleneck -- if you are holding xrp in payment channels, and the price is volatile, you don't really want to hold for too long, else you defeat the entire point of the 3-5 sec settlement time to reduce volatility; but then again, opening and closing channels constantly defeats the point os escaping the 1500txps bottleneck
am i missing something here?! if you settle "on-chain" regularly, even for small amounts, eventually your payments can "get stuck" if we assume global VISA type scales
also, it's worth mentioning here that it's not just XRPL that can do payment channels, it's pretty much any blockchain that wants to... so yeah, i am still not clear exactly how this will play out, i can't visualise it all in my head, but my guess is we'll get newer better systems in the next 3-5 years, for example look at what Larsen-backed Elixxir is working on -- they are claiming WAY higher throughput "out the box" (and this is Chaum, so this isn't more crypto bullsh*t).
for sure, something like Interledger is "infinitely scalable" just like any base communications protocol is... but settlement seems to be a bottleneck currently that needs to be improved and i don't think saying XRP itself is limitless is fair or accurate
it's good - VERY good - at what it does, and hell, yes even better than the competition (so far), but limitless? not so sure...
i still think that XRPL will just be one of many, many settlement engines and you simply choose (via pathfinding) the least "clogged up" or best fees, or liquidity, or whatever, at the time the transaction is made over an interoperable web of ledger networks