Here's a proper old-school GFNQ ramble for you. :-)
In the past year and a half I've been playing Warhammer 40,000 again. I say "again" because I've owned Warhammer 40,000 in one edition or another since 1995 - and by then I already owned Man O' War and Warhammer Fantasy - but I had hardly ever played the damn things until I started gaming with 40K Dave, Stuart, Charles and the rest. I've finally got a substantial amount of figures painted up and though I've plenty, plenty, PLENTY more unpainted in my collection it's awesome to finally be able to play a game with an army I assembled & painted myself.
When I first met Dave to play 40K I brought the latest core rules edition I had: 3rd Edition. At the time 5th edition was the game and that was what I learned and played with them. Last year 6th edition came out and we very quickly all transferred over, with me finally acquiring a copy of the new core rules at Christmas time. However, the process was comparatively minor because the editions are all based on the same core engine. Indeed, 3rd Edition codices (army lists) were still being used in 5th edition when they hadn't been updated yet, and the same for 4th ed codices now in 6th edition. Though they do require some FAQing and errataing, the key rules of the codices are basically transferable between editions - though older ones tend to be weaker.
This got me thinking about new editions of projects: be they role-playing games, war games, card games or the like. There's a big balance to be struck between starting from scratch and being just a longer errata, and some people prefer one over the other. (That's before you factor in any fluff changes - I'm going to purely discuss rules here.) Edition shifts can be hugely divisive - anyone with experience in these fields won't have to think long to think of things like D&D 3rd vs 4th Edition where it gets really ugly. When people have heavily invested in a product both in time and money they get very emotional when changes get involved and anyone who disagrees with them is obviously STUPID. To make matters worse there's a huge economic problem at the heart of it: a new core rules is by far the best selling product in the line (since by definition anyone who plays should own it, unlike your many supplements) but if a new edition is a disaster then your whole line is buggered for years. The best way to get more cash in is to release a new edition, but to do so is to walk into a minefield.
If you make the edition close to the previous one, you can mostly preserve backwards compatibility. This is good for longer-term players who don't need to buy new supplements - their existing Dwarf Army Book or Guide To Snowmonsters can be used as is, or perhaps with minor alterations, as with the earlier mentioned codices. This is also better for retailers who don't end up with worthless stock: those same supplements might be years old buy can still be ordered and sold. Similar editions, the theory goes, are also easier to teach to people who played previous editions - they only need to learn a few minor changes, not a huge wad of new rules. Perhaps more than that, people play a game in part because they like the rules and so if people liked Edition X, a similar Edition X+1 is likely to contain they parts of the game they like rather than being a new game with a different name.
By contrast, a serious set of changes is the only way to fix some problems - you can errata & FAQ all you want but sometimes major game engines need to be changed to get rid of problems with grappling, firing into close combat, re-roll priority or whatever game mechanic is That Game Mechanic. Sometimes games need a fresh pair of hands at the wheel who are willing to kill scared cows rather than include everything because "It's always been" - which can make the game play faster, easier to learn or what have you. There's an odd inverse finacnial argument too: if I own Edition X, then I may want Edition Q to be different enough to justify the expense of buying the core rules again because why would I pay for 95% the same material?
Needless to say, many of these positive for one side can be negatives for the other. Other negatives spring to mind to. For games too close to the original, it can sometimes be harder to remember the slight differences between similar editions with old timers - small differences in the stats of a machine gun or how you determine charge distance are easily forgotten when you're used to the old way and everything else is mostly the same, whereas a clean break of a system makes it easier to clear your head. For vastly changed systems, previously valid characters/squads/armies/whatever can be rendered invalid by the nature of the changes and make converting to the new edition awkward for someone who built their game off a now invalid combination. Warhammer 40K contains examples of both: 5th to 6th ed was a small change and the barely changed stat blocks of a Sonic Blaster continually trip me up, but 2nd to 3rd ed was a bigger change and I possess figures which had valid armaments then but are now unusable without some surgery. D&D has similar issues - individual spell differences between 3.0 and 3.5, like the duration on the buff spells, versus the absence of various character races/classes in 4E PHB.
Finally one needs to consider the time frame between edition shifts. Wait too long and people can abandon a game they they think has systemic problems, but usually the problem is making editions too soon - people resent "having" to buy core rules again. (This applies more so to wargames and the like where the social aspect is more likely to require you to transfer editions - a small RPG group can get away easier with using older rules.) Some people prefer smaller tweaks to come through errata but those lists can grow to become too clumsy to use - by the end of D&D 3.5 some trouble-zones like Polymorph can been errata'd numerous times, and there's a host of Magic: The Gathering card where the errata text is vastly different to the original printed text. Digital production helps -PDFs and the like are often auto-updated with errata for free download - but for hard copy it's a challenge.
I'd argue there's strong parallels with other fields here, such as operating systems. Moving from Windows 3.x to Windows 9x was a big change that was popular, and Windows 9x to XP was a smaller one that was also popular, but XP to Vista proved less popular. As with games there's social aspects, monetary aspects and a balance between replacing things and keeping them familiar. Sometimes you can strike the worst of both worlds - Vista managed to break compatibility with substantial changes without actually adding much new and useful to the table. (I would argue this is akin to AD&D 2nd Ed, which is different enough to be annoying when porting from 1E but similar enough to leave all the big problems intact.)
In the case of Warhammer, there's only been small changes since 3rd edition - which was first released back in 1998. I think the game is overdue a serious change, but there's serious financial implications of that: pulping and reprinting all the existing army books, have swollen in size and number since 3rd ed came out, and you need to include interim lists to cover the armies until those books come out - something much harder now there are more armies and they have more different units in them than they used to back in 1998. Still, I think as strong as the 3E base was that it's time for a brand new engine under the hood if we want to see any real change, and with increasing focus on larger games the rules could use tweaks to improve that level of play.
I'm very curious to hear your feelings & experiences with edition changes. Do you prefer any particular level of change between editions? Are there any edition changes your personally found difficult?