I remember reading a while back about how sharding wasn't going to be realistically achievable until proof of stake is implemented. Technically it would be possible, but the complexity of doing it on a proof of work system would be enormously higher, and likely not worth the effort for a crypto that is moving to staking anyway until after the switch. Maybe I was misinformed, but the jist was that splitting up nodes into shards in a way that distributes transactions relatively evenly and efficiently is difficult with the randomness and huge number of miners under proof of work, whereas things are a little more predictable under proof of stake.
Anyway, I just mean to say that I think it's Casper that's holding things back. Even though Casper itself will only increase capacity modestly, it's the last piece of the foundation needed before the true scaling projects can be expected to go live. At least as I understand it. Of course, Casper has been delayed significantly already. I hear you on that. But Ethereum is still way out in pole position with the Lion's share of developers, and I'd rather they take their time to do it right than have another DAO-like incident because that actually might threaten Ethereum's future. Though I do hope it comes pretty quickly.
Edit: I forgot to say, I do think Plasma doesn't rely on PoS, and supposedly OMG is on the verge of releasing their version of it. So the first step in scaling may actually come from that route. It just won't hit Visa+ levels of transaction capacity until sharding is also online.
Facebook
Twitter