I'm enjoying the conversation too! Your point about the scaling properties making LH2 more favourable for larger vehicles is something I hadn't properly appreciated in those terms, but is certainly correct. As far as T/W goes, I think that you do need to be able to pull 3G off the tower, and moreover do it with one propulsion system (or cluster of identical systems). Again from a systems-engineering point of view, the point where a launch vehicle design jumps the shark is where you're strapping together completely disparate types of rockets to get the desired effect. I don't think that an LH2 engine has ever actually done this, although the the X-33 designs perhaps could have worked. (I always thought that the Venturestar was emblematic of everything that can go wrong with an LH2-driven design, since it basically turned into a (non-)flying kitchen sink exposition. But the DC-Y was elegant enough; I would have loved to see somebody make an honest attempt at that.)

So basically, it sounds like there are three things required to make a cryogenic first stage/RLV viable:

1. A market for large payloads -- large enough to allow the scaling laws to work out in LH2's favour -- with sufficient frequency to pay off the development costs. As such, this doesn't exist yet, but it's certainly conceivable that someday it could.

2. New types of LH2 engine designs, which are much more durable than SSMEs (which I guess would be the current benchmark of LOX/LH2 durability? Or would that be RL-10s?) Presumably with some substantial revolutions in materials sciences. This may be well underway already (as you allude), but is not public info yet.

3. New, cheaper ways of generating hydrogen. As long as the lowest-cost method is steam reformation, then hydrocarbons will, by definition, be cheaper. If that's the case, then you'd end up in a curious situation where LH2 might only be economical in the middle of the spectrum. In an ELV which is primarily concerned with capital costs, hydrocarbons clearly are more more cost-effective, since they're much easier to develop and build. On the other hand, in an RLV which achieves true airline-like operations -- where the cost of fuel begins to become a meaningful concern, which it currently is not -- then hydrocarbons win again. That might only leave a thin band in the middle where LH2-based designs are the most cost-effective. On the other hand, a new low-cost hydrogen-generation technique (eg., bacteria that exhale it) could change the game.

So yes, I guess I'd echo your sentiment that this is something that could be relevant in a decade, if all three of those items develop in the right way. Definitely promising enough to merit some R&D, but not yet assured or inevitable.