I have picked a recent successful RFA to compare: Icewedge's 2nd, which passed on 24th December 2008. A fairly average RFA - a self nomination, a second try, and nearly 100 participants. Quite normal aspects for any RFA, which is one of the reasons I picked it.
Now let's go back a year, and take a look at Islander's 2nd, which passed a year ago today. A self-nomination again, but with an additional co-nomination by Rudget. It's also his 2nd go, but has only 60 participants this time. Let's compare it with Icewedge's. Islander has a lot of questions: 15 to be precise, and Icewedge has 14. So, the number of questions is still just as high as it was in early 2008. Both contain "stock questions" (i.e. questions that have a right and wrong answer, such as "When should a cool-down block be used"). The rest is all similar, with similar participants as well. A major difference not on the RFA is the amount of RFAs that had passed in January 2008 was much, much higher than January 2009: 17 prior to Islander, but only 3 this month so far.
Back to January 2007 now, with a look at BozMo's RFA. At the top one can immediately see there's a minor formatting change in that there is no table of contents. The headers to produce this must have been added during 2007. Additionally, while today it is common for the bureaucrat to put "Closed by ~~~~ at ~~~~~", in those days, "Ended at ~~~~~" was sufficient. This is another self-nomination, and a fairly short one at that. I'm not sure how today's crowd would take such a candidate. According to one of the opposers, he has 893 mainspace edits, which is a significantly low amount in today's climate. A big difference between this one at Islander's is the amount of questions is halved. While the four extra questions are all stock questions, there are still significantly fewer. You may also notice the use of the word "voters" in question one. The 73 participants in this also differ quite a bit from today's RFA regulars. Another additional difference is the opposition over edit count. While today the focus is more on where one focuses ones edits, today it doesn't matter so much (for most people). The three opposes in Islander's RFA actually suggest why he would make a bad admin. The opposers for Bozmo merely make vague gestures that he hasn't got "enough" edits.
Back to January 2006 now: to Dsmdgold's RFA, from January 2006. An obvious difference here is the lack of background colours. These came into effect in 2006. The headers are still there though. This is another self-nomination. There are only 49 participants, though this is not really reflective of the period (there were at least two RFAs with more than 100 participants in January 2006). One might be wondering "Where are the questions???" Well, scroll down. They were moved up in 2006 so people actually read them. Whether this was a good thing or not I don't know. This user has 9 questions, but they're all stock questions again. The number of questions dipped up and down, but often no optional ones were asked at all. Despite having 8 opposes, that section looks suspiciously short. Normally, that section includes large conversations of some kind or another. It didn't then as much. Most of the opposes are opposing over edit summary usage - something today would probably not be heard of, and usually moves a concerned user to support on the promise the candidate will set the preferences to prompt them to use one. A user of note that appears in this RFA is Masssiveego, who some may today substitute with the name "Kmweber".
Back another year now, to January 2005, and to Jni's RFA. This one lacks headers completely, and is remarkably short, with just 26 participants (or voters, as that word was allowed then). The oppose side is nearly empty, apart from a single oppose that has nothing to do with the candidate. A cursory glance of the RFAs of that date show that most passed with no or very few opposes. An additional point is that most of the supports are votes, and provide no explanation. There is also a lack of link to edit summary usage and edit count. There are only three questions as well.
2004 now. I have not got round to making archive pages for the earliest RFAs, so this one exists only as an old page version: Morven's RFA. This is a very big difference from Jni's RFA. The obvious point is it wasn't on its own subpage, but added manually. While Jni's RFA had a few boldfaced "Supports", this one has none - the trend to bolden the vote must have come in later. There were also only 14 voters. This request has no opposes, and very votey supports. Well it was a vote then, so understandable. There are also no questions, no ending time, no tally, no sections. Also no numbers, but bullets. It's remarkably different from any RFA today.
So what has changed at RFA? In summary:
- The requests are now on subpages (introduced 2004).
- The requests are closed with a header and coloured background depending on the result (headers introduced 2005, colour background introduced 2006).
- The support, oppose, neutral, and all the other headers are now actually sections (introduced in 2008).
- A tally and end date have been added (introduced 2004).
- Standard questions have been added (introduced 2004). They were moved to the top in 2006.
- Links to the edit summary and edit count tools have been added (introduced 2005).
- "Votes" are not votes anymore, but "comments" or "!votes" (changed in 2007, probably following the aftermath of the controversial closure of Carnildo's 3rd RFA at 61% the previous September). The comments are numbered now, instead of bulleted (changed 2004). The actual vote (support, oppose, neutral) is now bolded (became standard in 2005). People nowadays tend to provide reasoning more than before.
- There was very little discussion in RFAs. Discussion is much more common today.
What is established by the list above however, that apart from a few minor formatting and design changes, the only real change on the approach to RFA is that it was very much a "Support ~~~~" thing, and very little discussion occurred. Today, while many people still support without a reason, most actually do, and there's a lot of discussion on most requests. That is probably the only thing that has changed. And it wasn't as a result of RFA talk either.
So, judging by the lack of real change produced by this talk page, I propose we stop wasting our time here and do something else instead.