Great question, and it'd take me a long time to answer.
My first thought about resumes is that they serve two purposes, one legitimate and one not:
(a) social status grading, which is easy for douchebags to game and for extortionists ("do or I'll fire you and give you a bad reference") to abuse.
(b) a list of "ask me about " topics where X ranges over areas of professional expertise and interests, so they can probe you during the interview as to what you actually know.
I'd focus on (b) while throwing (a) to the wind. One thought I had is an "allocate 20 points" system. You don't actually have to prove anything because resumes suck at that, but if you put "Machine Learning: 7" that shows that you view yourself as being "35%" Machine Learning and are fairly comfortable discussing it on an interview.
This leads to the concept of a scarce graph, which is weighted but also imposes granularity (e.g. 20-point limit with 1-point minimum units) to prevent sprawl, and forces people to prioritize. Running graph algorithms against scarce graphs, with data pertaining to peoples' desires for connection (new jobs, new candidates) could be interesting.
So there's that. You get as much legitimate information out of a 20-point allocation of interests and experience as you would out of a resume or a job posting. What you don't get is the social status bullshit (dates and titles).
Ok, onto fixing the labor market in general, I think the best solution I can come up with is to build this: http://michaelochurch.wordpress.com/2013/05/07/fixing-employ...
That's Part I. Part II comes from the fact that, if Part I is build, people are going to want to get better, fast, at marketing themselves so their call options trade at a higher rate. That leads naturally into career coaching (a better model than traditional recruiting) but also into objective evaluation of, for example, source code quality. Now we can actually verify that, yes, John is a top-notch programmer and his $200/hr-struck call options actually aren't out of the money.
Part III would be to use all the professional development data thus gathered and start scoring employers based on how much value they add to peoples' careers. How fast does a typical person grow, as a programmer, after 2 years at Google? What does it do for that person's employment potential 10 years down the road? Those would be great things to know.