You are basically forced. Considering some of the bugs in past versions that were corrected in later ones...
And not like newer versions do not introduce new bugs? I've spent thousands of hours playing on 1.6.4 and have not encountered some pretty serious bugs which are present in newer versions, such as one that is often reported that corrupts chunks for no reason (not due to a game crash or improper shutdown):
I see quite a few people posting crash reports for newer versions as well, yet I've never had any crashes that were not due to mods or doing something silly like setting off a Superflat TNT world, or possible in current versions (e.g. stack overflow error due to water/gravel in certain situations, mainly due to 32 bit Java using a smaller stack by default, fixed with a simple JVM argument to use the 64 bit stack size).
Of course, I've fixed some bugs myself - including ones which are still not fixed as of 1.12.2; for instance, ever noticed how lighting is weird in staircases, among other areas (despite the versions listed this has been around for far longer than just 1.7.4)?
(this bug only started affecting singleplayer in 1.7.4 (the first bug report) due to chunk loading being made dependent on render distance - why not do what Optifine does in 1.6.4 and limit the minimum chunk load distance to 10 but allow it to go higher as needed? Or, you know, actually fix the bug by making mob despawn distance (viewdistance - 2) * 16 blocks, with a maximum of 128 and a minimum view distance of perhaps 5 so mobs are active within a reasonable range on 2-4 chunk render distance; the current minimum used by the server is 3, which only allows entities to be ticked within a 3x3 chunk area around the chunk the player is in)
The first issue may also have to do with the fact that the client does not wait for the internal server to shut down (a longer shutdown time due to deoptimized code could be why it is far more common since 1.8) - which has an extremely simple fix (just make the client wait for the server to fully shut down before letting you do anything else), which also fixes numerous other bugs, such as worlds not deleting properly or worlds simply disappearing from the saves menu (again, without a crash or improper shutdown; if you quit the game normally while the server is still saving it is the same as forcibly closing it).
And, of course, the more features that are added the more bugs pop up.
I decided to update "Old Caves" to 1.12.2, and as a bonus I included an "Old Mineshafts" mod which increases mineshaft frequency to 1.6.4 levels (2.5 times more common); both mods together make the underground nearly identical to 1.6.4 except for mineshafts in mesa biomes, ravines, which are only different in the way they curve from their starting point, and dungeons, which will still be twice as rare unless you increase the count to 16 in Customized and will be in different locations either way (they doubled their y-range from 128 to 256 in 1.7, meaning that only half as many attempts are below sea level).
Here is a comparison of the same seed in 1.6.4 (actually, 1.5.1, which is virtually the same; this is a world I play on, ignore the mine tunnels) and modified 1.12.2:
For comparison, this is what vanilla 1.7+ looks like - there is not much variation in the size and density of cave systems, much less different types of caves:
There are 31 unique biomes here, not counting sub-biomes or edge/river biomes, and representing the full gamut of climates in an area measuring about 2200x2200 blocks (there are many other biomes which are not shown here):
For comparison, this AMIDST map of 1.7+ shows the same area - just look at how bland and boring this seed is! At least it has a jungle, and a reasonably-sized one at that (if I want huge biomes I'll play on Large Biomes). Not much height variation in most biomes either, and while you can use Customized it wrecks biomes that are supposed to be flat:
As far as I know you have to actually hack the game to do any of those, and I doubt that commands or add-ons will ever completely displace mods because in order to do that they would have to basically open-source the game so you can have full access to the game's code and alter anything you want, not whatever Mojang deems worthy of adding a hook or setting for (even Forge suffers from this problem, which is one reason why I do not use it and many of the more complex Forge mods contain "core mods" which directly alter the game).
They'll also be raising the block ID limit in 1.13. And I can guarantee you nobody's complaining about that one. Or would want Mojang to scrap it.
That's not really new though; ever since 1.2 the game has mostly been able to handle 4096 block IDs except for a small issue with how ItemStacks were handled, limiting them to 256 (the same numerical ID is used to represent both blocks and items) - which I thought was removed in 1.8 when they removed numerical item IDs (though since older worlds can be read they must have retained it in some manner), which mods, including Forge itself, also fixed (ever wondered how big modpacks add hundreds of new blocks, many with IDs in the 1000+ range?). It is actually surprisingly easy to fix this and Mojang could have done so years ago (the easiest method, which I've tried before, shifts item IDs up so they start at 4096 instead of 256, breaking compatibility with older versions but you can just convert the IDs and add a version tag to chunks the first time they are loaded).
Also, the way they are planning to make every variant of block its own block ID means that many of those 4096 IDs are going to be used up anyway (even 256 IDs with 16 data values each is 4096 separate states), and others will change, so existing tools and mapping utilities will break, and coding will become more complicated (for example, currently ores and many other world decoration features replace any type of stone by only checking for the block ID; in 1.13 they will have to check for every single variant separately, slowing things down as well). The save format itself will likely change as well (why store IDs as two separate 8 and 4 bit arrays, which requires fancy bit twiddling (Java, as well as most other languages, does not natively support 4 bit datatypes so the game has to split/combine bytes to make efficient use of storage and memory), plus another 4 bits for metadata, instead of a single 16 bit array that stores all of them and allows for 65536 IDs).
Even without all that, TheMasterCaver wants a "super duper" cave mining experience?
That's exactly why I've made many mods that enhance that aspect of the game:
Vanilla 1.6.4; notice how much cave density varies, with a couple very large cave systems complemented by large cave-free areas, areas of more scattered caves and several smaller dense cave systems. Note that there are not that many more caves overall than in 1.7, they are just distributed differently (the difference in underground air volume is even less due to caves not overlapping as much in 1.7+):
TheMasterCaver's World; caves are based on 1.6.4 with more variation, including occasional larger caves and ravines (rarely up to half a million air blocks in volume), new cave structures, dungeons with more types of mobs and faster spawn rates (I mean really, 4 mobs as seldom as once every 40 seconds and able to run up to a spawner and deactivate it before it even spawns anything?), including a special "double dungeon" with two spawners, more variation in the size of mineshafts (which appear to be much more common but are actually less common than in 1.6.4; vanilla makes them less common near the origin, as in the maps above; they are less common since they cannot generate if there are too many caves in the area, which is something I even added to an otherwise vanilla world because it bugs me that much):
If they listened to the community there would be far more underground content, but no, they even made it less interesting in 1.7 (there are suggestions to add more content on Reddit, the only place Mojang actually reads, and all they said is that it is "impossible" to even add customization even though it only took me a few minutes to hack 1.12.2 without using MCP so it almost perfectly replicates 1.6.4, just by changing a few numbers). They sure seem to have forgotten that the game is called "Mine"craft.
You're THAT guy. What? Don't like horses? For Pete's sake!
I don't know why they still play 1.5.2 but I don't care about horses either (not that I hate them) as they do not work well with my playstyle and railways (I just added another 1500 block long section to a network that already had 10 times that in total; the total round-trip distance to my latest base is 7740 blocks) are more convenient when traveling between established bases; I wouldn't mind if I had to play in 1.5.2 (IIRC some people could not update to the 1.6+ launcher due to their computer being too old), which would be pretty much the same as playing 1.6+ except for no coal blocks (I didn't start mining all coal until 1.6 so that wouldn't even matter; if I had to go back right now I'd just make a simple mod that adds them to 1.5.2, just as I added "rail blocks" and "cobweb blocks" to the game so I can compact those as well).
Going back further, I'd miss redstone blocks (storage blocks and a convenient way to power powered rails) and Nether quartz ore (easy XP for enchanting, though I only do it one time anyway) from 1.5, anvils (infinitely repair the same items by just renaming them) from 1.4, XP from mining (I get over 3,000 XP per play session from mining, overall enough to repair my gear several times over) and trading with villagers in 1.3 (though I didn't start this until relatively recently in my history and I don't need to trade for diamond gear anyway), jungle trees from 1.2 (easy source of wood, which I use to make torches), enchantments in 1.0, and ravines, mineshafts, and a cave bug fix from Beta 1.8 (by far the most significant update to the underground in Minecraft's history). There are more but those are the biggest features that I can think of that I would miss.
PCs in general are declining in popularity as a primary computing device - increasingly, more and more people are no longer even using them:
A decade ago, almost all web browsing was happening on desktop and laptop computers. The explosion in smartphone popularity has rapidly changed the dynamic, and older predictions may have underestimated the impact. According to new report from comScore, computer use is dropping off quickly as mobile devices encourage new types of interaction. In fact, many younger people don’t use computers at all, instead relying entirely upon their smartphones.
So it should be of no surprise that the majority of Minecraft sales have been for the Console and Pocket Editions. That said, sales of the PC version have not declined but rather have remained steady - remarkably so since the game was first released, actually; the Wiki has an outdated sales chart that shows 3-4 million PC sales per year between 2010-2015, a trend which has continued since then (around 20 million in June 2015, 23 million in March 2016, and currently 26.8 million). Of course, CE and PE sales have accelerated upwards with PE surpassing PC more than 4 years ago even though it was very simple back then (it didn't even get End content until late last year), as was CE (not just based on a PC Beta version but crippled with limited world size, mob limits, and more).
Note that was as of 2013, when CE was far more limited than it is today (as far as I can tell it was based on Beta 1.6.6 back then, plus the aforementioned limitations) - and even then it had already sold half as many copies as PC in a third of the time (around 2 million of those were in just 1-2 months, while as mentioned before PC has just steadily sold copies since 2010, contrary to most games, which see a big spike followed by a gradual decline, though Minecraft's indie origins make it a bit different from most big title games).
I'd rather see damage variance. The current static damage means that a sharpness V sword is WAY better than a sharpness IV sword in many cases, while the difference between sharpness IV and III often makes zero actual difference. If we got damage variance, then instead of it being a binary matter of either always killing in 3 hits or always killing in 2 hits, it would be that you could kill in 2 hits with a weaker weapon, but each higher damage increment means your chances of killing in 2 hits are increased.
This was actually the case before 1.6, where Sharpness added a random 1 to 2.5 damage per level, meaning that a Sharpness V diamond sword dealt anywhere from 12 to 19 damage (rounded down from 19.5); in 1.6 this was changed to a constant 1.25 damage per level for 14.25 damage (including 1 for your unarmed damage, which was added to weapon damage between 1.6-1.8). If I ever noticed any change I don't remember as in either case they can kill unarmored zombies in two hits and cave spiders in one hit (another advantage that pre-1.9 has) and I did not use unenchanted swords much (the +1 damage in 1.6 meant that spiders could now be killed in 2 hits instead of 3 and silverfish in 1 instead of 2; conversely, unless I do a critical hit Sharpness V is useless on spiders but could sometimes kill them in one non-critical hit before).
Not only that, in 1.9 they changed the way Protection enchantments work - instead of multiplying their protection value by a random 50-100% they made then constant so instead of reducing damage by 40-80% full Protection IV now reduced damage by a constant 64%, close to the old average (which is about 63.6% and not 60% since damage taken decreases non-linearly as damage reduction increases) while the specific forms of protection went from 52-80% to a constant 80%, making them much more effective (for example, 52-80% guarantees survival from a max 44 block fall, 80% is always 102 blocks).
IMO, they did this to remove luck from battles - it isn't fair if your opponent wins because their weapon happened to deal the maximum randomized damage at the same time your armor offered the minimum randomized protection. Also, by making enchantments reduce damage by a constant amount this means that you don't need such powerful armor to ensure that e.g. creeper explosions cannot kill you in one hit (since otherwise you need to use the minimum amount, not average or maximum). In fact, for this reason I changed Protection to reduce damage by a fixed 60% and others by 75% in TMCW.
At least now you can't click the LMB until it becomes sentient and tries to go on strike and both fend off AND quickly kill a crowd.
And what is wrong with being able to attack multiple opponents? This is why my own version of an attack cooldown takes effect on a per-mob basis; you can only attack a single mob once per second or your attacks will be weaker but you can attack multiple mobs in the same time without a penalty (while you have to actually hit a mob for the cooldown to have any effect, letting you spam-click empty air, I've come up with a solution for the next version of TMCW which penalizes you for doing that by adding an additional damage reduction which increases as you click faster without hitting anything - in any I see the cooldown as a punishment, not something that is intended to affect normal gameplay; likewise, I did not fix MC-4732 (see description for original bug) which causes weapons to take damage when hitting a mob while it is damage immune even though it is a one-line fix - either attack properly or spend more XP and/or materials to repair your sword).
How do you take on a dungeon that is spawning up to 1 mob every 1.25 seconds (which was increased from vanilla to make them actually a challenge) without resorting to measures like dumping lava on the spawner. Or even "normal" mob encounters, some of which actually force me to take shelter even with the ability to attack as fast as I can and Knockback II on my sword (I'm still surprised that some people abhor Knockback even after 1.9):
You might say that is not a good comparison because vanilla doesn't have caves so large that the entire mob cap can spawn within them, respawning faster than you can kill them until night falls (more surface spawns = less underground) but I actually did not kill that many more mobs in this world as compared to vanilla (1.6.4) and have encountered similar events in vanilla (such as a ravine I found in a region where I'd explored everything else; I killed around 100 mobs before lighting it up). Or even just getting rid of a bunch of zombies that have collected outside a village overnight (sure, they would just burn and aren't even a threat but I like to kill them, and do not like things that slow me down; that's why I use an Eff V diamond pickaxe to mine the 3,000 ores I mine per play session instead of unenchanted stone/iron tools like some do, and indeed why some people can't believe I can mine that many ores at all without making them more common - I really tear through the underground, as seen on this before/after of my last session. The same goes for weapons; another annoying thing about 1.9 is that Sharpness V diamond sword is just a bit too weak to kill zombies in two hits unless one is critical (they effectively have 20.3 HP when armor penetration is factored in while the sword deals 10 damage. Why not make Sharpness add 0.75 damage per level (instead of 0.5 + level * 0.5) for 10.75 damage? Smite would still be better against armored mobs).
Even Mojang thought that being unable to attack multiple mobs at once was not a good idea - why did they add Sweeping Edge, which can deal significant damage to multiple mobs at once without even having to directly hit most of them (sweep attacks alone do not count since they deal a mere 1 damage, hardly enough to change the outcome of most battles)? I see that as a reduction in skill when compared to quickly moving your mouse over each mob you hit (how fast and precisely you can move determines your attack rate; I know from measurements I frequently attack two mobs less than a quarter of a second apart, with one click of the mouse button per mob hit, which is why I have never considered a general cooldown like 1.9 added, instead one which is tied to individual mobs; indeed, vanilla already had such a system in the form of damage immunity and unlike most players I know that you cannot deal any damage in most circumstances (the exception being if you can deal more damage than the first hit, and even then the game subtracts the first damage dealt so you only deal the second damage overall, and is its hard to do a critical within half a second of a previous hit).
I see people constantly telling others if they don't like the new changes to just use older versions, but the older versions include some bugs I think would 'bug' people out
And not like newer versions have their own bugs? For example, I'd say this is a pretty serious bug, and while it has been reported before 1.8 it has been far more common since that version based on all the threads I've seen about misplaced chunks - and I've never seen this on any of my worlds, even one with more than 116 days of real-time playtime; crashes are also not necessary for this to happen (otherwise the most likely cause, same for "world suddenly disappeared", which I've also seen people report with no prior crashes or improper shutdown):
And what if you have fixed bugs yourself by modding the game (I believe there are also mods, and even Forge itself, for fixing bugs)? Some of which have still not been fixed as of 1.12.1. Indeed, one such bug may be a cause of the aforementioned bugs (swapping chunks and world disappearing); the internal server does not get shut down properly because the client does not wait for it to shut down, which can be fixed with a few lines of code, which also prevents other bugs from occurring, such as the world deletion bug that causes parts of a deleted world to appear in a new world with the same name (the deletion of in-use files fails and/or the server writes new data after the files were deleted).
I'm one of those who has not upgraded to newer versions (since 1.6.4) but mainly for reasons unrelated to additions, and many of the new additions can be ignored or are easy to work around, and as for changes, don't like the new textures? Just extract the textures from an older version and make a resource pack, which is pretty easy to make. Mojang is even making it easier to work around many of the other changes by making the game more customizable; in 1.13 you'll be able to add a crafting recipe for Notch apples or even change the golden apple recipe back to nuggets (both of these have caused a lot of complaints for something I hardly ever use) without using 3rd party mods or commands, and you can already edit loot tables to make Notch apples easier to get if you want.
Some of the complaints that I've seen border on flat-out ridiculous, like what happened when roses were changed to poppies (check out this 17 page thread - a thread on the combat changes in 1.9, which had far more impact on gameplay, has only 27 pages) - another change that is very easy to revert. I have more reason to complain than many people yet I don't go around making threads about how so-and-so update sucks (I'd have to actually mod the game to work around some of the changes in 1.7; this was made worse by the complete lack of customization in 1.8 despite adding so many settings for world customization, and they even claimed that said customization was impossible when it is no harder than customizing ores or such. I have made a thread about why I have not updated past 1.6.4 but it is not a complaint thread and I don't really care what they add in newer updates anymore since I can just mod the game myself if I want something (I've even added some newer features to my mods); even when playing "vanilla" I use a few minor mods).