One decade later, Minecraft world generation is interesting again


The new worldgen is incredible, both above and below the surface. Hats off to everyone at Mojang that worked on this!

One of my favorite game memories of all time was the first time I played Minecraft during the alpha days and immediately got lost in the world. For me, going into a 1.18 world and finding a giant mountain chain and then finding a cave on the peak of the mountain that falls all the way into a giant underground ocean at the bottom of the world in a huge cavern complete with underground vegetation… it brought me right back to that sense of exploration and wonder that I remember from the very first time I played Minecraft.

Also, I found this datapack (sort of like a mod but doesn’t require patching the game since the game natively supports loading datapack files) called Terralith [1] that extends the 1.18 worldgen with even more cool stuff. Wandering around the new 1.18-style mountains is incredible, and then stumbling across something the resembles Yellowstone National Park added by Terralith is even more incredible. It’s worth a try if you like just wandering around in Minecraft marveling at the scenery.



I also recently (today in fact) encountered the Terralith datapack, it's rather impressive and I'm tempted to try it out. From what I know it doesn't add any new blocks, instead it just reuses existing vanilla ones.

There's a trailer they made for a recent update that shows off some of the terrain results:


The one thing that always stuns me is how in the hell did they (both Mojang and terrain pack creators) figure this out from math?

Like, it's perlin noise and a bunch of algorithms and it makes mountains and caves and overhangs that are all decorated and such and a river bleeding into a ravine that leads into a giant cave network.


It is not that difficult if you apply yourself.

Just yesterday I watched this fantastic explanation of how to generate realistically looking ocean surface:

The fantastic part about this video is just how little of actual physics you need to know to create convincing, complex waves. You can skip entire fluid dynamics and condense it to "water molecules move in circles when subjected to a wave".

Forget about Perlin noise, it is useful to getting something but it is IMO dead end if you are for realistic looking world. There isn't a way to fix it because, fundamentally, it is completely disconnected from how land features are created in real life. To get asymptotically to realism you have to start taking into account more and more of physics and passage of time and Perlin noise just completely ignores this.

To get realistic representation you need to build a little model that knows a little bit about how the features are created.

You can even go Dwarf Fortress direction and model processes that create various features of the world over time.


Don't underestimate the role of the human observer focusing and exploring the most interesting parts.


Didn't we all do that as kids with the Mandelbrot set viewers? Just explore it to find some cool visualizations.. then zoom in and repeat %)


I wonder how many if those alpha worlds are preserved. I started when the alpha was free and the beta was pay, and the scope of those alpha worlds was just incredible. There wasn't much in the way of nature in the alpha so all those worlds were more art project than construction.


Minecraft is one of the biggest games of my teenage years, starting in Alpha, and I always was a bit paranoid, making backups and copies of things. As the sibling comment points out, you can go back in versions from the game launcher itself and I recently did a nostalgia trip loading up those worlds again, seeing how the game has evolved, how my playing style evolved... By loading them into a few intermediate versions, you can even get them to load in the latest one ! I really appreciate to be able to do that. (Even if I still have a couple of copies of the older versions I could play with !)


With the current launcher, you can actually pick which version/update to play, including old versions. For Java Edition, it lists versions all the way back to alpha as options.


This is so user-unhostile I wanna cry. Most companies just want the power to force antifeutures on the users nowadays.


I added the Terralith datapack to a new world on my server, but am I not getting it? Everybody gets the same world right? Yo have to download other seeds? I don't see massive changes, any coordinates that are particularly interesting in the default world?

Edit, I think I found something nice (a cave) at -281, -15, 436, that's not something 1.18 generates by default I guess?


Thank you for mentioning this. I've been out of the "MCraftsphere" for a long time now, so jumping from ~1.6.4 to 1.18, alongside the plethora of mods (data packs now?) is absolutely mind blowing. I had no idea the depth to which it has been taken, but it looks like I now have an afternoon project.


Man. I really do want to like it, but this update is the first one that gives me the Microsoft-finally-took-over vibes. Traversal and with that exploration just became so much harder. The landscape is now riddled with holes that you either build scraggly bridges over, or lead to massive detours. The sides of bodies of water are much higher and steeper on average, making it impossible to see the landscape you're passing, and getting up there to take a look is tedious.

So you're basically stuck near spawn unless you want to spend much more time getting around.

A lot of surface iron was replaced with copper, which is useless, meaning progress is now slower than ever, and grinding becomes even more of a necessity. Generally, many of the items that now clutter the inventory are mostly useless, and the inventory needs constant attention.

Minecraft is a fantastic game that I have spent well over a thousand hours on over the last decade, with many more to come very likely. That being said, with the account transition to Microsoft/Xbox and this update that gives me personally the "Microsoft knows whats best for its users" vibes that I still distinctly remember from WLM and Skype, I am definitely less optimistic about the game's future than I was a few years ago. I do hope my caution proves to be unwarranted.


I'm no Microsoft apologist, but I believe you've misunderstood the intention and the effect of the changes.

Previously, the time-optimal strategy to acquire iron and diamonds was to dig a hole deep into the earth and then bore straight paths. This was exceptionally grindy (fans would say zen, detractors would say tedious), but effective.

With 1.18, the optimal strategy is now to enter a cave and find the resources there. The holes in the ground that you dislike are entry points to that system, where many resources are available -- though at the risk of encounters with monsters.

So the transition is from a static grind to the new system based on cave exploration and risk. It's more dynamic and engaging, directly the opposite of your claim here. You mention iron, claiming that progress is slower, but in actuality it's significantly easier to acquire useful amounts of iron now. Jump into any random cave and it's all over the place. Sure, there's a lot of relatively useless copper too, but no one is forcing you to mine it. Just skip what you don't need and you don't need to manage it in your inventory.

Likewise, the new topology with increased heights means that rivers are significantly more important for exploring. You point out that it makes it more difficult to go from one peak to the next, which is accurate (though in my opinion you overstate the difficulty), but regardless it is quite easy to make long journeys if you follow the rivers and low terrain.

So yeah it's harder to go directly in a straight line than it was in before, but now you're paying attention to the terrain in the world and adapting your gameplay as a result. It just requires different navigation techniques. Like it or not, you're absolutely not stuck at spawn.


> I believe you've misunderstood the intention and the effect of the change.

No, and I don't think I have made any comment on the intention. I understand what they were going for. My point is specifically what you are saying: It massively changes the game play. If I wanted different game play I would play a different game. Every update so far, including the ones since MS takeover have felt like natural progression of the game to me. I enjoyed those updates and have lots of good things to say about them and the way they expanded the game.

This update feels to me like it should have been a different game, or a mod. That's the contrast I was trying to highlight with my comment.

It's worth noting I'm not claiming anything as a fact. I was stating my personal opinion and concern for the direction the game is headed in, relative to my own expectations, after a decade of playing Minecraft. I explained why I think that way, because we're on HN.

> The holes in the ground that you dislike are entry points to that system

Those existed before, just that covering them and getting the materials for it if you're not a fan of big craters and abysses in the ground wasn't a project of its own before.

> So the transition is from a static grind to the new system based on cave exploration and risk.

Cave exploration and mob bashing is still grinding. This will get old fast if you just want to get to it and build your base or whatever self assigned goal you have.

> It's more dynamic and engaging, directly the opposite of your claim here.

That is entirely subjective.

> You point out that it makes it more difficult to go from one peak to the next, which is accurate (though in my opinion you overstate the difficulty)

That wasn't my claim, no. I said it is now more difficult to see directly from the water, because the landscape is raised so much relatively the water. Overcoming that is just tedious, not difficult, because you have to get out of the boat every so often to get a view of the landscape.

For me it is comparable to amplified maps no longer being opt-in, basically. It's fun for a change, but not the game experience I am looking for most of the time. Now I no longer get to pick unless I stick with old versions, which of course sucks for online play.


One of the great things about Minecraft is game play can be tailored to a player's particular tastes. In this instance: if you want to dig straight down and build a branch mine to get diamonds and iron, you can; if you want to take the more risky route of exploring caves, you can. Personally, I fall into the former category since my main interests lay in exploration and building. Dropping down to the bottom of the world is a quick way to get the resources required to craft an enchantment table and suit up in armor, leaving more time to explore the surface for things like jungles, mesas, sunken ships, and anything else that will provide the resources for my latest build.

Does 1.18 change things? I'm trying to keep an open mind, trying to convince myself that the new ore distribution adds nuance to my opening moves. When that conviction fails, I look for other things that hold promise. Something that I've toyed around with doing in the past is creating a Zork-esque underground empire, but have always held back on those plans since the old cave networks were bland and cramped. Yet the new underground biomes make such a project more appealing.


> Previously, the time-optimal strategy to acquire iron and diamonds was to dig a hole deep into the earth and then bore straight paths. This was exceptionally grindy (fans would say zen, detractors would say tedious), but effective.

I'm not a big Minecraft player, but I'm pretty sure the time-optimal strategy for getting diamonds consistently was to set up a deliberate mine topology which is mathematically efficient, not just dig forward in straight lines. AFAIK diamond mining was the only thing that couldn't be automated; for iron the time-optimal way is to construct an iron farm, i.e. engineering, not grinding.


Unless you're really working to hit ore in a limited space, you pretty much just want to make a bunch of long parallel tunnels at the right height.


Is the strategy you're talking about (digging straight down upon spawn, as I've seen people do) just applicable to speedrunning, or does your average player learn to do that eventually as well?


Digging straight down is a speedrun strategy of last resort for modern versions. The typical progression to enter the Nether is to start by getting 7 iron (3 for an iron pick, 3 for a bucket, 1 for a flint and steel). You want to get that, grab some food, and be in the nether at around 3 minutes.

Doing that by digging down is not really feasible. There are two main routes.

In the village route, you spawn next do a village with a blacksmith. You get some iron or an iron pick from the blacksmith, kill the iron golem for the remaining iron, grab some wheat and then find a nearby lava pool or completable ruined portal. There are variations where you can trade with a villager for a bucket, but by then most speedrunners would have already reset.

The second and more typical modern route is using ocean. You locate a shipwreck which gives you iron and food. Then you find a ravine that has magma blocks (indicating there's lava underneath) and use that to build your portal. To my knowledge this isn't viable in 1.18's world gen though.

When speedrunners do go underground in the overworld, they'll usually try to find a cave first to find exposed iron and lava. Digging down is the last choice.


After doing the basics (stone tools, safe base, light and food), I'd always do that as soon as possible. Diamond stuff is very useful for later parts of the game, and you'll need a lot of redstone to build interesting circuits.

It's its own form of exploration, at least in the versions with fun cave generation: you always end up in some cave complex that you can then explore. But I'm happy to see that it's less needed now, because interesting caves are more commonly accessible from aboveground.


I've been playing since the release of beta and once it was understood in what layers diamond spawned more frequently it became very common. I've seen some people refuse to do it just because it kinda kills the wandering exploration vibes.


It's called strip/branch mining ( by players.

Everyone (including children) learns this method over time, be it by checking tutorials online, reading a wiki, or just naturally due to how simple and efficient it is.

You just had to mine straight down to y 8/13 (this changed in the new patch), make a long corridor and then just branch out of that corridor in lines.


At some point every player goes "Where, oh where in the world is the goddamned iron I need?", they google it, find ores spawn differently at certain heights and go from there.


Speedrunners don't do that, for one. They don't generally go searching for diamonds, only grabbing them if it's in an easy location like a chest.


It's also what I figured out when I was a kid.


You are mostly correct. My kids and I played Minecraft 1.18 survival even in beta. The new ore distribution rules, especially for iron, greatly increase the grindiness of the beginning of the game. We were compelled to immediately dig down to Y=16, where iron is most common (except for extremely tall mountains that are rare biomes).

We did not find the world "riddled with holes," though, and gaps in Minecraft have always been easy to traverse by simply placing blocks.

1.18 ore distribution sucks. Going caving virtually requires full armor due to mobs, but now to get full armor you need to go caving (or get extremely lucky).


My friends and I converted an existing world to 1.8 (hard difficulty). Even with full iron armor I got wrecked the first time I dug down to the cave beneath our base. Was not expecting the sheer number of mobs.


There is a ton of exposed iron on mountain surfaces.


It seems weirldly like the Minecraft devs tried Minetest and kinda randomly picked a couple of features to copy. The article describes how the new worldgen’s verticality and caves are reminiscent of Minetest, but inventory clutter is also very much like the default Minetest Game (which also has pretty useless copper, albeit as one of too many tiers of tool material, going into the same recipes as stone, iron, gold, diamond, etc.).

I much prefer the Minetest game NodeCore, which is playable with 8 inventory slots (the small HUD bar is all you get, because the game shuns all modal UI except a hints system for figuring out crafting and other gameplay mechanics without using a guide to see whole recipes, which are considered spoilers) and makes the fancy terrain more navigable by letting you scale cliffs and even climb overhangs without placing or mining anything.


It's extremely easy and quick to go >>1k blocks away from spawn looking for good territory if you want to, and I've done so now in three separate worlds. And I didn't have a bunch of resources when I did that either. Iron is easier to find than ever now that it spawns on the surface in mountain biomes. I got over a stack just taking the surface iron on a mountain. Sure, there were also copper veins, but there was plenty of iron. This was the first thing I did after punching down a tree and getting a stone pickaxe btw.

I do agree with you on inventory clutter, but that has been a problem for as long as I have played the game (since about 2010? I don't remember which version). I am disappointed that they have delayed the introduction of the bundle.


I haven't tried but I believe I can't play Minecraft anymore because all players have been forced to convert their Mojang accounts, which is something I simply don't want to do. Funny how a company can just take away something you purchased and you have no recourse.


The game files continue to work, even without an account actually. You can simply keep the old launcher and/or run the game directly.

While I agree with you in principal, the truth is that if you don't update the launcher it doesn't break - and if you do update the launcher it also doesn't break, it just requires you to complete an account migration.


Why does it matter who administers the account you use to log in to the game?


Nope. You can still play with a Mojang account. I do.


You can, but you have to use the old launcher, which is buried on the Minecraft site and avoid the dark pattern prompts to upgrade.






If you like voxel world generation, I strongly advise you to check out Veloren[1], it has a much much deeper worldgen that what Minecraft offers.

[1]: the game is still under development but it has been playable for more than a year now.


> Veloren is a multiplayer voxel RPG written in Rust.

This is such a meme by now, I have to assume it's an inside joke. Even games mention that they're "written in Rust" in the first sentence in their official presentation, what other programming language communities does this? Why would anyone except other Rust programmers care?


In the case of Veloren it’s rather relevant since it’s a community project and this way they actively recruit new contributors among people (interested in) using rust for gamedev.


In a winamp discussion thread a couple of days ago somebody linked to , and I had the same reaction to their "It's written in C++ using the Qt toolkit and GStreamer. Strawberry is compatible with both Qt version 5 and 6.”

It’s definitely weird for end-user software to promote itself based on internal implementation details, but it’s by no means a rust-exclusive phenomenon.


That line is the last line of their "About" section, and for open source software it is relevant since OSS projects often want people to contribute.


We mention it because:

1) The player community and the dev community are not distinct: we involve players in the development process and actively encourage them to become contributors.

2) There are many people that are interested in working with Rust, and it's often a big pull factor for new contributors.

3) Veloren is by far the biggest public domain game written in Rust: it's a good demonstration of what the language can do and how it performs at scale (Veloren's game server scaled up to a 48-thread machine with 181 players online in the same world during the last release party: not many games, let alone voxel games, can manage these numbers). We want to encourage other game developers to consider using it too.


Every sentence on the page is focused on developers, the community developing it, and technical details of the game. It seems relevant.


I'm not a programmer, but anytime I see something other than Java mention it usually bodes well for performance.


“Minecraft Java Edition” springs to mind :P

Otherwise I mostly only see this when it’s “for developers by developers”


Do you know what's also become a meme by now?


The Veloren developers seem to have forgotten to provide a page of information _anywhere_ giving a non-negligible overview of what the game is about. The main website has little more than a paragraph listing some other games as inspiration, followed by some Minecraft-like screenshots. It also has a manual which is clearly focused on developers and contributors, as it provides next to no information for a new person who just wants to know a bit more about the game without having to actually download it. How is it similar to Dwarf Fortress? How is it similar to Minecraft? How is it different? What do you do in it? Why is it so much work just to find an overview of the game and what it has to offer?


Have you checked out the wiki? There is a 'Getting Started' guide:

The game is extremely open-ended: there is no overarching objective, and you are free to interact with the world as you please. For most players, this means a combination of exploring, crafting, conquering challenges (dungeons, caves, etc.), finding rare items, and socialising with other players.

The game is 100% free so you have no reason not to jump in and give it a go if you're curious!

> How is it similar to Dwarf Fortress?

Veloren has an open procedural world with a history generation system. Although work on the history simulation is still very much ongoing, the world already more cohesive than those of many other voxel games. Our end goal is a world that feels as rich and as complex as that of Dwarf Fortress. Whether we end up succeeding is still to be determined!

> How is it similar to Minecraft?

Veloren is a voxel game, so... cubes and open-world exploration. Not much more to say about this one.


It's still under heavy development, AFAIK there is a long time vision but they are still far from there so it wouldn't make sense to discuss about it in the front page.

And at this point they are clearly looking for contributors much more than for players, that's why all the documentation you'll find is targeting the former.

Documentation and marketing material require a lit of work, and doing so continuously against a fast-moving target is a enormous effort.


Submit a pull request.


With what? A manual?

Parent is checking out a game because it was mentioned on HN and finds that the game’s web page doesn’t have a manual and your response is essentially: “Play the game sight-unseen, learn everything about it, write a manual and contribute it back to the project”? That’s not very realistic.

I could see creating an issue in the tracker, perhaps.


See also Minetest, which has several native map/terrain generators [1]. In addition, one can define other generators as mods using Lua [2]. Mods can define new biomes [4], place buildings [5], roads [3], etc.

If you add all that, though, you might there might be some glitches (roads passing through buildings..) but that's part of the fun.







I wouldn't say that it has deeper worldgen, but because it replaces voxel terrain with a lower lod after a certain distance, the world has a feeling of massive scale. You can see a far away mountain range on the horizon and actually journey there, then glide down it to an even further sea. Afaik the new site2 system also shows lod levels for generated structures.


The world feels massive because it is actually: a medium-sized hill in Veloren is taller than the entire horizontal dimension in Minecraft.

I'm personally not fond of the (pretty crude) lod view, especially because of the moiré in it.


Edit: vertical obviously.


This seems like a recreation of Cubeworld, which unfortunately seems pretty much dead now (but the developer disappeared for years at a time in the past too, so who knows). Too bad they changed a lot of the systems that made the game great for the Steam release.


Or more like open terrain generator for spigot, it's so much better than stock minecraft it's laughable.


Don't suppose you know what that image was rendered in? Guessing not Overviewer since the angle is weird and it doesn't do water shading like that...


Veloren is great. I'm pretty sure it's worldgen uses Simplex noise, rather than the Perlin noise like MC does.


It uses simulation, like Dwarf Fortress. Of course randomness is involved, but there's a real attempt at realism in the geology, etc.


Looks like a really cool open source project, will definitely check it out.


It's been a while since I've played Minecraft, so I can't speak to the newer terrain. But I always thought they could significantly improve the world gen.

Perlin noise is cool, but why not try to fully simulate plate tectonics, weather patterns, drainage? It would go a long way if deserts were in the rainshadow of mountain ranges, islands corresponded to real volcanic activity, etc.

I guess there is a limit to what you can do with the chunk-by-chunk approach, but I think a certain type of minecraft player stays basically put within a couple dozen chunks. Having worlds "load" on exploration would be a worthwhile tradeoff.

I get that this is a hard, but it is interesting, and it's a problem that would captivate an intelligent developer. They have like a million dollars, why not?


What do you mean having then load on exploration? That's the current state - a chuck isn't generated until you need to load it. This means when you take an old save to a new version any already-loaded chucks you're stuck with but any new exploration causes new world gen on the new version.


Yes, I think (maybe) the reason they use perlin noise is because it’s fast. I’m saying have an actual loading screen when you go to a new area and spend some CPU cycles doing advanced plate tectonics or whatever.


Oh wow. I did not realise this was the case. Thank you!


I think this actually might be a slight negative tbh - since everything is split into chunks and generated on demand, you lose the ability to do larger passes over multiple chucks at gen time. No real way to do tectonics or weather patterns as you don't necessarily know what's in your chunk neighborhood.


It is stunningly difficult to do something like this within procedural generation. Remember a given xyz coordinate has to generate given only the coordinate number itself and a seed. You couldn't do stuff like plate tectonics within an "infinite" world because that would require generating every block which is akin to generating every digit of pi


Hmm, I'm not sure that's true.

If you generate a noise function that dictates where the "plates" are, then the location of the "plates" is already predetermined from the construction of the function. Then the output of that function could be used to alter the mountain/terrain functions.

Thus, no need to pregenerate everything. You pregenerate all of the functions in advance, but not all of the outputs.


You could set up domain boundaries where everything in that domain is relatively isolated from other domains. When nearing an ungenerated domain, do the procedural + evolutionary generation and handle the borders blending. Domains could be tectonic plates. This would cause hitches on exploration, but they would be relatively infrequent and overall few since domains would be so large.


I also find it stunningly difficult, however, when I look at Sebastian Lague's channel [0] (no affiliation, just one of the best channels I follow), I feel like this could be a not so stunningly difficult project for him.



The article suggests that "more realistic" is correlated with "boring".

"Real" is an interesting concept. Any such simulation couldn't really be based on Earth geophysics. Who wants to spend months crossing the steppes? Real magma isn't something you find by accident. Real tunnels can collapse. And so on. So these intelligent developers will need to create a new system of geophysics.

I've read that Dwarf Fortress uses a process more like what you describe.


I think a SimClimate could become a hit. Maybe the dynamics could be gamified for kids, but I bet there are lots of adults who'd also want to just "spend months crossing the steppes" following the seasons pretty much.


Hmm are there any Climate Change strategy games?

There are games for nuclear war and pandemics that are fun and educational about the globe-spanning dynamics of those catastrophes. There are also games like Universal Paperclips that provide an intuition for the exponential dynamics of complex systems.

Seems like there should be a way to fuse these concepts around climate change.


> The article suggests that "more realistic" is correlated with "boring".

I wouldn't say so. With the increased mountain height I'd argue the terrain is more realistic than ever, and not boring.


I'm referring to this part of the article:

> In the b1.8 update [September 2011], the terrain generator was subject to a rewrite aimed to make the code simpler, and make the terrain more realistic. ... at the time it caused quite a controversy, a thread on the Minecraft Forums decried the new terrain as "boring".

What does "realistic" mean to you? Are you looking at it like a geologist? If we follow rland's idea to completion, Minecraft players should look for kimberlite pipes to find diamonds, etc.

I'm certain some people would love virtual prospecting like this. But I strongly doubt the game would be as popular.


Other people are pointing out that real == boring, but I disagree. Perhaps we wouldn’t want a literal real world, but hyper interactive terrain like you suggest could be used to explore new and interesting gameplay ideas.


You can do something like this by layering. E.g. generate a high level coarse map that sets some features like average elevation, and calculates climate from that and derives biome from that, as an input into the lower level worldgen


I believe this is exactly how minecraft's new "3D" worldgen works. It calculates a high-level terrain map, then a temperature map, and then biomes are placed on top of them that fit the characteristics.


Reminds me of the AutoBiomes paper! Waiting for something to be integrated into EG a terrain creation tool (no!-MC games), should produce pretty interesting results.


Doesn't the ultra-realistic mod terrafirmcraft already do exactly that?


To my knowledge, no. Terrafirmacraft does have strata of different stone types that stretch for large areas, but these aren't tectonic plates. It does generate biomes based on "lattitude" however.

As another comment mentions, dwarf fortress does do all of this though. I don't think this would work in minecraft though, I think minecraft needs a mix of terrain types near each other to be most interesting, and tectonic plates would spread it out.


One thing I always hoped, but that never happened: get people’s creations into the game. Randomly get vetted buildings as biomes. People have created insane things and it’s a shame they couldn’t have been integrated directly into the game. Even a few would have been cool. I guess one thing is that structures without life feels empty, so you’d have to generate random monsters or NPCs for each of them.


They _kind of_ have this, within the limits and goals of Minecraft.

They've spoken recently about one of their intentions for Survival is for the world to be a blank slate for players own creations. Adding more and richer structures detracts from that. There's also the part where it's still a game that needs balancing, and just importing arbitrary structures and resources detracts from Survival.

There is the marketplace which has community servers/worlds/levels that showcase what other people have built. I think that's the right level for this.

Otherwise, stuff like this is perfect for the data pack ecosystem they created.


I’m not sure I get your point. VAlves does it with cs:go and dota 2. Players create skins and all, and then vote for the ones they’d like to be incorporated. Sometimes valve incorporates some of the players’ creations.


I think you could completely automate this by training a neural network on local neighborhoods to generate a voxel probability distribution for the center voxel, so a randomly initialized world would converge to a world that has pieces that resemble the worlds that the network was trained on. So in 1D you would take a sequence of voxels ABCDE, and the network would learn to predict C from A,B,D,E.

Similar idea without neural networks:

I'm surprised no one applied this idea to Minecraft yet. There's even a 3D version:


Isn't that what servers are for? I joined a server once many years ago and found all kinds of truly amazing things that people had built. My favourite was a vast hollowed out mountain. The only way in was through a door right at the top of the mountain which caused you to descend into this beautifully lit cavern. Exploring this server was like exploring an art gallery.


It’s not. I’m talking about player structures being randomly incorporated into your own game.


That doesn't make much sense to me. The generated world is meant to be like nature while player structures are man-made things created by inhabitants of the world. Something that makes a bit more sense and possibly quite cool would be to incorporate ruins of famous structures into the generated world.


This might be more feasible-ish in a community effort?

I'm not entirely sure how the jigsaw system (the subsystem the game uses for building villages and bastions from a large palette of pre-builts using a set of rules and connectors) integrates with the custom world generator options, but that would seem the best bet for it in a vanilla game.

The problem with an "official" player creations is both vetting them and the social backlash because you didn't pick little Timmy's shack. As well as the disjointedness of theme that could occur with a player's playstyle.

I play a modpack. Create: Above & Beyond, they've added a bunch of new random structures to the game, but they're all within a similar theme of quasi-medieval for the surface structures, and a few of the underground ones stray a bit more modern.


Valve does it with cs:go and dota2. People submit skins and creations and you can vote. Valve then sometimes accept some of them and integrate them into the game. There’s no reason why players shouldn’t be able to create biomes or structures.


Other things to note. One big one is that the world height limit has increased from 256 blocks to 320, which allows for much more verticality.

I think the whole field of world generation is awesome. Perlin noise, and the concepts of "smooth noise" that have come from it, and the clever ways people have found to layer noise and use it to map biomes and heights, and spawn locations, etc. And then, even more impressively, with games like No Man's Sky, you have this concept applied to spherical geometry. I just think it's all so cool.

Incidentally, his (Sean Murray, creator of No Man's Sky) is the best talk I've ever seen on the topic:


The new terrain is great, but I have mostly exhausted my patience with Minecraft survival. I hope future releases will put some effort into reducing the grindyness. I know many things can be automated but they tend to rely on extremely arcane game mechanics rather than feeling like an intentional part of the game.

Some elements of Factorio were inspired by Minecraft. I think Minecraft could now use some inspiration from Factorio.


Have you considered modded to get a little more Factorio in your minecraft?

I'm playing the "Create: Above & Beyond" pack.

There is grind, the pack gives you a goal (build rocket, go to moon), but how you get there is up to you and the automation isn't quite so arcane as vanilla, and doesn't rely on "magic blocks" (Where you plonk down a single block and it magically does all the things) instead relying on you making a machine/factory line from multiple blocks.


I've pushed java far past its breaking point in previous modded minecraft worlds (30 second GC every 30 seconds, some AE2 leak when millions of items are stored).

I like many aspects of bedrock edition, most related to performance. I want a gregtech-like experience in bedrock so badly that I'd go as far as making it myself. I just wish the command interface wasn't so janky. When I've done test blocks and interactions in the past I've run into limitations that would prevent my vision. This was maybe 18 months ago, so I don't think much has changed yet.


Bedrock does have a Javascript-based behaviour pack system. No idea about it's capabilities.


The problem with those packs, especially ones like tekkit is that they just mash in a thousand different mods of barebones compatibility, some adding the same kind of ore or item slightly differently and you end up with a billion items, some of which are duplicates, no idea where to start from or what to use with what, and all in all a general unbalanced clusterfuck.

Create mod by itself is fine as it is imo, no extra stuff required.


I specifically called out Above & Beyond because it isn't a "kitchen sink" pack thrown together over a weekend with a dozen different game crashing interactions between mods, or 4 ways to create dupe item cycle. Or a hell inducing grindfest like the Gregtech focused packs are.

It's a pack with one goal (build rocket, goto moon), that puts you in the SANDBOX that is the core of minecraft and says, "here are the tools, here's a bit of guidance on what you should automate to progress, but figure it out from tools at hand and have fun in your own way".

The mods are from different authors, but all the recipes have been tweaked and tuned to integrate with each other and provide a cohesive progression curve. Even the textures have been tweaked to make the pack visually cohesive.


> no idea where to start from or what to use with what

A lot of recent packs give you a quest book to guide your progress and push you into the different mods slowly. You get maybe 70-80% through one mod's tech tree and then you're pushed to start on another section before you can progress into that last 20%. The Project Ozone series does a REALLY good job of making sure the different mods balance each other and interact in a way that feels good.

> you end up with a billion items

Yeah that gets bad. Most packs have some sort of inventory organization system (basically a computer you can stick all your items into with no organization and then search them up later) to help alleviate that.


The trailer for Create Above and Beyond made it look pretty integrated. They added a lot on top and did a lot of mod editing.


I honestly never understood the appeal of vanilla Minecraft. I only bought the game because of the mods! Mods like AE2 or Thermal Expansion is where the fun is at!


I think Mojang should reconsider its objection to adding an auto-crafting system into the game. There are legitimate concerns that, if it was too easy, it would make one of the defining aspects of the game obsolete, but I think it could be balanced so that users wouldn't bother doing it for simple items like pickaxes.

As an example, the "auto-crafter" block could be powered by chorus fruit, which means the player would not have access to it until after beating the Ender Dragon. That would also mean that truly automating crafting would also require an automated chorus fruit farm, which would give the system a minimum space requirement.


MC 1.19 has a new 'Allay' mob planned that will pick up and move items for the player. I can see people using this to create some interesting automated systems.


My kids don't play survival at all really and although I have kept trying for years I can't find an actual enjoyable game in there at all really. Like you say, too much grinding, and then it just starts to feel completely pointless and arbitrary.


My understanding of the height limit (originally 128 and later 256) was that the lighting algorithm was O(height limit), and so it was hard to increase it arbitrarily. And then most of the shape of the terrain is dictated by the height limit—you couldn’t really have big mountains full of false summits, and indeed the terrain generation didn’t really have many low-frequency features that spanned large areas (but maybe they are boring because they are slow to traverse). For comparison, the highest peak in the European part of the Netherlands is around 320m whereas the Minecraft limit wouldn’t let you above 192m above sea level. Denmark has a highest peak of 170m so it would fit but the game wouldn’t generate such a thing as you’d only have 22m above the top to build. I wonder if one way around raising the height limit would be to offset the base height of chunks relative to each other, but that might lead to strange behaviour in other parts of the game. Maybe Microsoft have figured out a better fix to allow increasing the limit.


The extremely low height limit is also influenced by gameplay: many voxel games let you run up slopes that vary by only a single vertical step, but Minecraft forces you to jump to climb, which is much slower. Nobody wants to jump 16,000 times just to get over a mountain range. On the other hand, while I actually really like the traversal and movement in Minecraft and I wouldn't want to change it, the current height and generation limits really are just too low to create anything truly majestic. The new update has increased the limits by 128 (64 blocks each above and below the prior limit, for new extreme Y coords of 320 and -64), which is a welcome change, but IMO they could have gone a bit higher.


Actually +64 above/+64 below is only the new default. Custom worlds will be able to increase the world height limits to +2032/-2032. That leaves a lot of room for activities. I think custom terrain gen data packs like the one mentioned above still have to be adjusted to actually use the full height.

I'm excited for mappers that can now build 1-1 scale recreations of real and fantasy locations like everest (~4000m from the base), minas tirith (300m), barad-dur (1400m), courisant senate building (600m), etc.


I've been playing Minecraft on-and-off for the better part of the decade now (since I was 9 or 10 wow!), I've been amazed at some of the recent changes that Mojang has made with improving game engine performance and now this new caving update.

I hope they once again push for a revival of modding multiplayer servers, although that is most likely a pipe dream because of the Bukkit DMCA issues in 2013-2014 :(.


Multiplayer modding is still very much alive and well. There is both the client-side route (where everyone uses a modpack tool like or the successor to Bukkit, Spigot (and its High-performance fork, PaperMC:

Spigot/Paper are bukkit-compatible and you can load modpacks from the site, or even, which is still operational to this day.

Source: I run several instances myself and I help develop a server management tool for it:


It was a pleasant surprise this morning when my son and I discovered the vast cave systems below Y=0 in our existing world! Very well done, really cool, we’re having a blast!


> I have stopped in recent years since I find it dangerously addictive

I forgot this, but that's actually why I stopped playing it. I was in uni around 2007-08 and realised it was using a serious amount of my time and energy. I just quit cold turkey back then. I've barely played any video games since this point, in fact.

What I loved about Minecraft initially was the exploration. I got into it during the beta (I think) when there was no mini-map or any "lifelines" you expect from a normal game. One thing that was possible was to go down into cave system, run out of torches, fall somewhere and just be trapped in the dark. The game was so simple there was a very real fear of losing your save file because you got trapped in a cave forever.

Are there any other games that have real stakes in them like the early Minecraft versions?


>I was in uni around 2007-08 and realised it was using a serious amount of my time and energy.

The first public version of Minecraft came out in May 2009.


Wow, that's strange, I could have sworn it was earlier. It must have been 2010-2011 when I stopped playing it then.


Surely you could always punch your way out? And an unlit cave was never pitch black as far as I'm aware.


Maybe, but I think there was still a risk you'd end up under water? I don't think it ever actually happened to me as I was genuinely scared of getting stuck so I was always prepared and careful.


That's what the hardcore mode in Minecraft is for: if you die, the save is deleted.


A fair few games have an Iron Man mode where they only save on exit, so you can’t revert.


They should really enable a sharded server networking protocol to allow some MMO elements. Each server would handle a limited area and players would automatically connect to a server when they go to a certain distance.

I'm sure players would pay to access it if it worked.


That was notch's original idea for portals in multiplayer I recall reading. A portal would take you to another server.


This is how the big multiplayer servers available via Bedrock edition work, not so much for distance, but for different game modes. Unfortunately players pay by buying predatory micro-transaction items, which I find near unacceptable in a game so squarely targeted at children.


Hypixel has done that to an extent. The way most servers do it nowadays with with a server proxy that shunts client requests to a specified server transparently -- you "world hop" using in-game mechanics (i.e. walking through a portal, clicking on a sign)

And as others have already mentioned, WorldQL is an effort to transparently replicate and shard the world without even requiring "world hopping"


I remember seeing a project like that (maybe even here, on HN), but unfortunately I can’t find it