What to do about inflated letters of recommendation

In the midst of the job season, Leiter has an interesting discussion on the practice of inflated letters of recommendation. It's no longer sufficient that you have positive recommendations, with no doubt raisers, but importantly, the letters have to be long (i.e., at least 3 pages), full of praise (without raising ridiculously high expectations), comparative (lack of comparative claims make the candidate look weak), and very detailed. 
Also, letter writers are preferably famous, or at the very least established figures in the field. Here are some problems with this emphasis on inflated letters of recommendation: 

1. It makes the candidate dependent on the good will of his/her letter writers. If the candidate's letter writers don't want to go through the trouble of writing an inflated letter, does this say something about the candidate (as is commonly assumed), or about the letter writers? In other words: the personality and eloquence of one's advisor and one's personal rapport (or lack thereof) with him or her play a large role. 
2. It puts people from the non-US market at a disadvantage. In Europe, for example, the tradition is to write very brief letters. They are typically only solicited once the long listing process is completed. There is simply no tradition to write long letters, especially comparative ones.
3. It drains time and resources from letter-writers as well as increasing the volume of material that SC need to go through.
4. It increases inequality, giving people from strong programs/departments an unfair advantage. A few days ago, I had lunch with a professor here at Oxford, who was in the process of shortlisting applicants for PPE (philosophy, politics and economics). Obviously, at Oxford, there are many more applicants than places (acceptance rates at his college for PPE were slightly below 10 %). The overwhelming majority of those shortlisted for interviews came from public schools (i.e., privately funded, expensive) schools. As the professor explained, one reason is that students from public schools get much better letters of reference. The students who came from government-run schools consistently received weaker letters of recommendation no matter what their grades, and this in part contributed to the fact that fewer students from government-run schools got selected. As one of the commentators wrote on Leiter's thread: "When the top two criteria for job placement are pedigree and letters, we are but one small step from the olden days in which departments would call up the chair at, say, Harvard or Princeton and ask who the top job candidate is. That person would then get hired without an interview. Thankfully, we at least have interviews these days, but the fact that some folks inflate letters and others don't makes our current system in some ways even worse than this "good ole boy" club practice. One's entire career can depend on whether one's letter writers are willing to write inflated letters."

Given these disadvantages, what to do? Could we perhaps go to a process similar to many online refereeing systems for journals and ask referees to simply tick boxes (e.g.," I think this candidate is in the top"5 %, 10 %, 20 %.... (tick one); "I would describe this candidate's teaching as " enthusiastic, thorough, ... (select all that apply))? Should we only consider letters of recommendation after the shortlisting has taken place (like in Europe)? Get rid of letters of recommendation altogether? Solicit letters from peers after sending them the applicant's materials once the longlisting or shortlisting has taken place?