Current Wikipedia policy is to grant administrator status to anyone who has been an active Wikipedia contributor for a while and is generally a known and trusted member of the community. Most users seem to agree that the more administrators there are the better.
Contrary to the Wikipedia mantra 'Adminship is not a big deal', it is — because of the very stressful and oft humiliating experience of the inquisition itself. The trials and tribulations of being a sysop come later.
Requests for adminship (RfA) is the process for creating new administrators on Wikipedia. RfA has been around since June 2003, when Camembert created the page after discussion on a mailing list. Reform of the process has been discussed continually since at least 2005, but the process has remained virtually the same as when Camembert created the page in 2003.
The first major attempt to reform the RfA process was with discussions for adminship (DfA). Bureaucrats actually initiated many of the mooted changes before reverting to the old format pending discussion. The idea behind DfA was relatively simple—a week of discussion would precede the voting phase. This would combat voting that organizers felt was little more than "stab[s] in the dark" thanks to limited information about each candidate. They took pains to note that they were not looking to completely modify RfA, but they felt limited tweaks were necessary. Ilyanep emphasized the "ridiculously increasing high standards", "huge amount of instruction creep", and RfA's similarity to polling. Lar noted that while the process was currently working and "good enough", he believed "... 'better' is the enemy of 'good enough'. I think that even if this current process works pretty well, there may nevertheless be better ones out there."
Also in 2006, Adminship renewal, proposing a term of office for administrators, was rejected. This idea is common; it had been proposed before and has been proposed many times since, which is why it is now listed on the perennial proposals page. In 2007, a large survey was conducted to attempt to find what the views of the Wikipedian community were. Around the same time, Wikipedia talk:Requests for adminship/Reform was created to attempt to improve the process. The page quickly ballooned before participation suddenly declined, leading editors to conclude that it was "dead" and trying to draw consensus for RfA reform from it would be flawed. 2008 saw a major request for comment, where many of the perennial proposals were debated, but no consensus was found for any substantive changes. Interestingly, the request spawned from an attempt to delete the main RfA page. Many of the arguments against reform were based in the number of administrators promoted in the previous month (34), which led to the belief that with so many successful RfAs, the process itself could not be broken. As Deskana stated, "Many people desire that RFA be changed, to improve Wikipedia. Change can be good, and it can be bad. What a fair chunk of the people that argue for the reform of RFA don't seem to appreciate is that it isn't 'broken'. That's not to say that it can't be improved. Broken would imply that it's not working at all, which it is."
RfA reform 2011
The most recent and probably most in-depth attempt at reform was in 2011, with the aptly named "RfA reform 2011". The process was immediately spawned by My76Strat's RfA, which was widely viewed as a microcosm of the larger problems at RfA. In its aftermath, Jimmy Wales commented "RfA is a horrible and broken process", and the general feelings led to the creation of RfA2011 by Kudpung. Even this massive effort failed to break the RfA deadlock; although nearly all of the participants desired some sort of reform, they were unable to get their proposals adopted by the main RfA community. Worm That Turned (WTT), one of the coordinators of the initiative, blamed this on the inability to find a root cause of the problems at RfA: "Different people thought it was it was too hard, too easy, got the wrong candidates through, was too uncivil, had too many questions, could give votes without reasons, with [poor] reasons... there was a long list." Kudpung added that the participants were also discouraged by those who vocally voiced their opinion that the page would not accomplish anything.
Why has reform failed?
With these many attempts at reforms, why is there still no consensus that RfA needs to be changed? Perhaps the answer lies in something simple: the natural conservative tendencies of Wikimedians, as most notably illustrated by Ironholdsin the Signpost and on his blog. This is not to say most Wikipedians hold conservative political views, but that they are resistant to most forms of change. As Ironholds stated on both pages, "Wikimedians actually tend to put a fairly small amount of stock in changing things to boost the community or the social aspects of the movement. Whether it's WikiLove, help reform or any other project to ameliorate the less pleasant aspects of the projects, the same refrain comes from an annoyingly large chunk of the community ... people don't like change, and ... existing editors are largely comfortable with the current situation". This mindset is seen in the RfA archives with comments like "maybe it's better we stick with what we've got, and try and tweak it to perfection!" or "It's the worst system except for all the others." 
With the demonstrable issues with RfA's process and culture—its central two tenets—many strategies, some mentioned above, but most not, for addressing them have been raised. Perennial proposals may sum it up best: "While RfA is our most debated process and nearly everybody seems to think there's something wrong with it, literally years of discussion have yielded no consensus on what exactly is wrong with it, nor on what should be done about that." Reformist editors are therefore swimming against a strong current to even stay afloat, much less find concrete proposals that may garner support. Despite acknowledging these difficulties, there are those who still attempt to reshape RfA. The aforementioned RfA2011 was successful in implementing a editor review-style process which gives candidates a chance to catch possible problems before the public process of RfA. It was also able to put an edit notice above the main RfA page warning inexperienced candidates of their RfA's likely conclusion, and contributed large amounts of research into RfA which is still available for other editors to read through. The Signpost asked WereSpielChequers, Worm That Turned (WTT), Dweller, and Kudpung what changes to process and culture it would take to bring RfA to a level where it could help maintain the administrator core while not driving disheartened editors away after a bad experience.
WereSpielChequers' full views on RfA are available in his userspace, but he believes that RfA is 'broken' and that action needs to be taken to halt the decline in active admins—more than 300 since its peak in 2006. His solutions for RfA vary:
Decide on clear-cut criteria for adminship;
Give bureaucrats additional leeway in deciding RfAs and striking !votes, and perhaps the authority to promote at a lower percentage value if the candidate meets certain conditions ("not unless");
Institute a representative democracy, where a dedicated committee would elect administrators;
"Upbundle", or remove certain roles from administrators and give them to the bureaucrats, thereby making the former role less important;
Clerks, similar to the arbitration committee, who would serve to ensure the discussions stay on track (see this and one of the well-developed proposals from RfA2011 for more details);
Pre-vetting, giving editors a chance to discover and resolve issues prior to an RfA.
A counterpoint is provided by WTT, who says that his research conducted during RfA2011 has led him to believe that the process is not 'broken' per se, but "it's keeping the right people out and letting the right ones through. If people stopped considering it as a hell-hole, I'm sure they'd realise it isn't one. ... adminship is "no big deal", even if RfA is." Buttressing WTT's argument, only one editor has passed RfA with less than 3,000 edits since 2009, but he had over a million edits to other Wikimedia projects. Dweller agrees in that he does not think RfA is 'broken'. However, in his view the standards for becoming an administrator have risen—in some cases too high. The impact of this has been limited, thanks to the increase in highly-capable bots, but the "ordeal" of the process itself has a detrimental effect on the current efforts to retain editors: "Even more than good admins, we must value our editors, and bad experiences threaten ongoing participation and RfA is an area where bad experiences can happen." Kudpung takes a middle ground, saying RfAs have "become such a rare event, it's not possible to be able to say whether it has become calmer or not."
Still, even if we are ensuring that only qualified candidates are applying, then there is clear evidence that the number of qualified candidates is falling. The administrator corps is currently in decline through attrition and a lack of new blood (see table, below). Whether RfA is 'broken' or functional, it seems to not be fulfilling its intended purpose of at least maintaining the number of administrators: there are 705 active administrators as of this writing, down from a peak of 1,021 in mid-2008.
Successful requests for adminship on the English Wikipedia