I've read that in practice, the weight of a bicycle in use doesn't vary - the lighter the frame & hardware, the heavier the lock required! :)
The price of carbon fiber has been a major factor, along with the difficulty of using it in manufacturing. Both of those have improved by orders of magnitude, and if one company I'm familiar with succeeds, they'll cut the cost of the fiber by another factor of 10 while greatly improving the quality. These incremental advances are truly changing the equations for many of the things you're pointing out.
As someone peripherally involved in "New Space", I'll say that the advances in technology really are making some of those fanciful ideas possible, even economically feasible. People right now are putting their money into those new ventures, and they're not doing it for entertainment but because the numbers pan out at least as well as many of the dotcom ventures. SpaceX has cut the cost of launch by 50% just by using well-tried industrial cost management methods - faced with exorbitant pricing for turbine pumps, the company built their own at 1/10 the cost. If the reusable first stage pans out, that will cut the cost by about another 1/2 (they think it will be better than that, but I think their long term costs will be higher as they transition their systems to support the more rigorous requirements of manned launches - that will affect all of their systems, even the unmanned launch ones.)
Another example, 3D printing has already proven itself with multiple different entities successfully printing components and even entire rocket engines, with costs and production times reduced by 90%.
Bottom line - much of the extremely high cost of space has been the government-run cost-plus market structure and extremely careful engineering practices. In fairness, I think this was necessary for the early days, but now we can move past that and into a more market-driven economic model with fixed prices on off-the-shelf products. Consider that development costs on a rocket engine are going to be on the order of a billion dollars, regardless of whether 10 or 10,000 of that engine are produced; but if 10,000 are built the amortized cost is $100,000 vs. $100,000,000 per engine.
If reusability pans out, the fuel cost of a launch is less than $500,000.