The "infrastructure" argument asserts that bicyclists need more environmental protection from motorists, and the way to make that happen is with new bike-lane infrastructure. This argument tends to support the intuitive position of people who think they would like to cycle more but are not currently doing so, i.e., "Gee, if only there were more bike lanes like in Europe, then I would get out and cycle on a much more regular, even frequent, basis."
The "bicycling skills development" argument asserts that if bicyclists learn and use the rules of "vehicular bicycling," then additional infrastructure not only isn't necessary, but can have negative effects such as creating confusion (if poorly engineered), creating a false sense of security for unskilled cyclists, and (ultimately) relegating cyclists to "second-class" vehicular status (because of the additional confinement of the bike lane -- e.g., if a bike lane exists but is not used, even if for a good reason, is the cyclist liable?) and ultimately driving (so to speak) the cyclist off the roadways. The "skills" argument seems to be adopted more by people who are actually regular cycle commuters.
I believe it to be the case that bicycle accident statistics tend to bear out the "skills" arguments rather than the "infrastructure" argument. See my earlier post here.
I personally think that bicyclists need more education and skills development to be safer. I commute regularly by bicycle, and the stupid (i.e., dangerous) things I've seen cyclists do is breathtaking. This includes inexperienced cyclists, who are still in the "toy-bike" mode, as well as experienced road riders (who are nonetheless inexperienced vehicular cyclists) who believe, since they are so experienced, the "rules of the road" don't apply to their seasoned selves.
If the problem could be described as a problem of educational infrastructure rather than transportation infrastructure, then what is the best approach? I don't see this as something that can or will be taught in public schools -- they certainly have enough problems, not the least budgetary. But the educational infrastructure needs a curriculum, a venue, teachers, and a funding source.
A curriculum could be developed by interested parties drawing from the LAB curriculum (which the LAB tends to keep under rather tight control, as their training courses serve as a funding source) and other sources, such as Forester's Effective Cycling curriculum (but much abbreviated). Of course, the curriculum should be regularized on a national basis, while being scrupulous to minor variations within the governing state's "rules of the road". The cost to do this would be a few tens of thousands of dollars, I would think.
The venue for this education should be diffuse, locally-based, and trusted. I think a good candidate would be the public library system. Public libraries are available and currently serve a well-recognized educational role. They typically have many outreach programs, which a bicycle-skills program could join. They have large parking lots for basic skills training. They have a good image.
The teachers will of course be drawn from the ranks of current practitioners.
Funding is of course always an issue. How is this to be controlled and paid for? Will the state DOT have any say, will they want control or licensing authority? (I assume driver-training schools are licensed by the state DOTs.) Would this ultimately be linked with (and at least partially paid for by) a cyclist-licensing program? Others have broached the subject of testing and licensing bicyclists, and I think it's perhaps an idea whose time has come.