I was certainly never confronted with the "fact" that SV engineers "certainly" do create proportional value. I was confronted with a bunch of bad math that proves nothing. Did you miss the part about the target ratio not needing to be 1:1? Or the part about developers' contribution during a year not being the only (or even main) determinant of revenue during that same year, which makes division of revenue for all by salary for some utterly ridiculous? Where are these engineers who drive the bottom line by 10x their burdened cost (not just salary BTW)? I've been one myself, I've met others, but for every such person I've also met dozens who contributed 10% or less over their cost. You need to stop assuming, and start measuring.
OK, on to the point about whether SV engineers are worth more than those elsewhere. Let's forget Bengaluru for a moment. Consider Boston or Boulder. Brisbane or Birmingham. Engineers there are every bit as good as those in SV. Sure, there are costs involved in having developers who are elsewhere, but that's balanced by the extra cost of building out offices etc. in one of the most expensive markets in the world. Buy some decent teleconferencing gear, provide a generous travel budget, and you'll still come out ahead even in these relatively high-salary locations. Now consider a place like Bengaluru. I haven't worked with everyone here, nor have I worked with everyone in SV, but I've worked with enough to be skeptical of the assumption that the average SV engineer can deliver enough to be worth several times the salary. Some can, most can't.
That's two strikes against you. Let's go for three. Should the free market determine wages? Absolutely. Do these no-poaching agreements subvert the free market? Again, absolutely. Does that add up to a conclusion that a bidding war would be a good thing? Absolutely not. CDNF. A temperate amount of poaching would be a good thing, but I don't think that's likely to happen. What's more likely to happen comes back to your statement about "a net loss or wash at best" because at some point that does become true. I personally think it has already become true at many companies. Those newly hired engineers are going to count against this year's budget; their contribution to revenue won't show up until next year's. The extra cost of hiring that first developer of the year has to be made up somehow. Part of it will be made up by reducing capital costs and travel budgets and such. Part of it will be made up by hiring nine new developers instead of ten (wherever they are) or by giving smaller raises to current employees (including developers). Maybe that "rock star" you just hired away from Google will pay off next year, but this year (and probably the next couple) their bidding-war premium will exert downward pressure on everyone else. A bidding war is not only bad for developers elsewhere, or for non-developers. It's bad for other developers in SV besides the few who win the lottery.
The problem with the no-poaching agreements is that they've created pent-up demand. Now, like soda that has been shaken, the newly opened bottle is going to fizz all over the place. Those making the agreements should have been prosecuted years ago, because waiting has created an even bigger mess. The absolute least likely outcome is an efficient short-term reallocation of resources. More likely is that some companies without deep pockets will be starved out of the market.