Too many Diamonds in Unleashed

[un worry]

Here's the side-by-side comparison for a variety of blocks, grouped by Unleashed version (v.1 versus 1.1.3)

As you can see, the vast majority of metrics are closely correlated, so we have established a reasonable baseline for comparison

Summary of changes:

• Iron - double
• Redstone - up a half
• Certus - up a half
  
• Gold - triple
• Diamond - somewhere between double and triple

It really suggests that - despite config settings - actual generation of some ores have significantly increased since Unleashed was released

To the OP: Until this week, I had deliberately by-passed quarries, deciding to mine every block in my world by hand (okay, it was a pickaxe of the core) and focus on bees for resources. Even when I had diamond bees, and plenty in store, I would still go out on my weekly mining expedition and look for those elusive diamonds. I don't know if there are too many, but would personally prefer if they remain scarce.

Links
this summary: http://i.imgur.com/nA8Y9ZM.jpg
raw data: http://i.imgur.com/1It4FvZ.jpg

 

[Sphinx2k]

For info I checked the config(who thought we could do that instead of taking irrelevant samples) and TE config is set to the default for diamond:
diamond {
I:ClusterSize=10
I:MaxY=20
I:MinY=5
I:NumClusters=1
B:RetroGen=false
}

I don't think those values changed since unleashed was released.

Even if the Cluster Size would be the same as vanilla.
Vanilla Minecraft should still be Level 1- 15.
On the lowest levels you have often bedrock preventing spawn.

The Cluster Size is also not really a fixed value. Test it with something common like copper. I cant really figure out how it is related to the ammount of ore blocks it spawns. It is not a min value and not a max value more a average value.

 

[Zjarek_S]

For info I checked the config(who thought we could do that instead of taking irrelevant samples) and TE config is set to the default for diamond:
diamond {
I:ClusterSize=10
I:MaxY=20
I:MinY=5
I:NumClusters=1
B:RetroGen=false
}

I don't think those values changed since unleashed was released.

TE config generation is bugged, it will show last run config as default. If you delete world.cfg you will get real default config for it which should match default minecraft values (posted by me before).

 

[AlanEsh]

Not enough mid and late game recipes that I care about require diamonds, so whether they are plentiful or scarce only affects me for about a day either way. That said, I have found nothing but 8-9 block clusters of diamond ores when hand mining on our current Unleashed world.

I guess if you are old school hand-mining with an Iron pick for a full set of diamond armor and tools, this might seem Over Powered. But I need only about 30 diamonds to get my tech level to a point where the overall scarcity of diamonds has no impact. I'm much more limited by Glowstone, Certus and Nether Quartz, and Lapis(!), so I guess it doesn't bother me that diamonds aren't a bottleneck too.

 

[Enigmius1]

It is almost triple (2.8x), it is a significant enough sample that any error is trivial when talking in factors and, regardless of "configuration settings", it speaks to what is actually in the ground.

So, collecting data on one thousand and twenty diamonds, from an area of 4 million blocks within 244 chunks, comparing sets of 64 chunks across identical specifications and finding each set comparable and within a standard deviation, makes comparing the two larger sets against differing specs both significant and valid.

It's neither significant nor valid because you're not sampling blocks or diamonds, you're sampling chunks. Which means you're taking 64 samples and trying to claim it's significant relative to the statistical convention of 1000 samples.

This isn't subject to negotiation. Your sample sizes aren't even approaching significant. And your outcome reflects as much. You can't make up the rules as you go along; they've already been established. I'm not willing to discuss this with you any further.

 

[hiroshi42]

It's neither significant nor valid because you're not sampling blocks or diamonds, you're sampling chunks. Which means you're taking 64 samples and trying to claim it's significant relative to the statistical convention of 1000 samples.

This isn't subject to negotiation. Your sample sizes aren't even approaching significant. And your outcome reflects as much. You can't make up the rules as you go along; they've already been established. I'm not willing to discuss this with you any further.

The English language is unfit for mathematical concepts as it could be argued that you are both 'correct'.

What un worry means when he/she says that the test is significant is that the samples gathered are large enough to apply basic hypothesis tests to. These tests deal with samples of a population that has certain characteristics. A sample size of 64 is certainly large enough to run a basic t-test on although we don't have some of the data needed to run a proper test. We are missing the standard deviation here since all we are collecting is the overall averages.

What you are probably thinking of with the minimum sample size of 1000 is when we have no idea of the distribution/shape of the population we are sampling. Here we start to leave the nice sanitary world of mathematics and venture into the realms of the humanities where we have to worry about clumping, non-parametric data and people being untruthful when answering questions.

Fortunately we are dealing with data generated by a computer which is nice and uncluttered. Unfortunately we do not have enough information from un worry's test to actually do an analysis (there are really only 4 data points collected) and so we cannot say that it is definitive, it certainly appears to indicate that there are more diamonds floating about then there were.

As for the config/whatevers. Do they indicate how the game picks what size vein to place? From personal experience in vanilla I have never run into a vein with more than like 5 ores while with 1.1.3+ I have found that to be a lower limit on vein size. Maybe the weights were adjusted?

Also from the changelogs from 1.1.2 to 1.1.3 "Boosted Diamonds by 1 ore per cluster". It would also appear that from 1.1.1 to 1.1.2 that the flat bedrock option was enabled by default which as several people have stated might affect the amount of diamonds and other deep ores in a not insignificant way.

 

[Enigmius1]

Fortunately we are dealing with data generated by a computer which is nice and uncluttered. Unfortunately we do not have enough information from un worry's test to actually do an analysis (there are really only 4 data points collected) and so we cannot say that it is definitive, it certainly appears to indicate that there are more diamonds floating about then there were.

As for the config/whatevers. Do they indicate how the game picks what size vein to place? From personal experience in vanilla I have never run into a vein with more than like 5 ores while with 1.1.3+ I have found that to be a lower limit on vein size. Maybe the weights were adjusted?

Also from the changelogs from 1.1.2 to 1.1.3 "Boosted Diamonds by 1 ore per cluster". It would also appear that from 1.1.1 to 1.1.2 that the flat bedrock option was enabled by default which as several people have stated might affect the amount of diamonds and other deep ores in a not insignificant way.

The challenge that I have with theorycrafting discussions like this is that 90% of the disagreement stems from the method used to obtain the information. And then in order to pinpoint exactly what is going on, it turns into a spreadsheet instead of a game.

I remember a point in my current (1.1.3) unleashed world where I was in a deep crevasse and I could turn in three different directions and spot a diamond vein. Joe Average seeing something like that doesn't think in terms of how ore is actually generated or how ores are distributed. He just sees three diamond veins within easy viewing distance and arrives at the conclusion that it's a lot and it can only indicate a change to how many diamonds are generated.

By the time we sort out methodology and acceptable practices and run the tests and compile the data...wait, no. In discussions like this, there's never a consensus on methodology or acceptable practices, so let's re-word that.

By the time we sort out methodology and acceptable practices, ignoring the input from people who do not adhere to those methods and practices, we spend hours running the tests and compiling the data and if we're smart and on the ball, we arrive at the only conclusion that makes any sense: that it doesn't matter.

 

[Saice]

Without getting into long winded debate I'm with Enigmius on this. It does not mater with your total block count is in thsands or millions. For ore gen it is the chunk count that maters. Why? because it will only roll the dice on RNG X time Per Chunk. So yes there is only 64 samples and while there might actually be some variance there to get a true bead on whats happening you need hundreds of samples.

Its Per Chunk that is the sample here not Per Block or Per Cluster.

So yeah sure your numbers suggest something. But replace what your saying with "I rolled these dice 244 times and the average 3 times what it should be" and you get a better feel for why despite there being 4 million blocks over all it seems more RNG is RNG then oh my god guys when need to cut diamond spawn to 33% or all is doomed.

Also as aside the idea of taking 244 samples and only comparing 64 sets of pairs just to me really seems out of the norm for running stats.

 

[Flipz]

Stay calm and keep Minecrafting, folks. Analysis is good an fine, but no need to start sniping at each other.
This has been the topic's Be Nice warning.​

 

[un worry]

Without getting into long winded debate I'm with Enigmius on this. It does not mater with your total block count is in thsands or millions. For ore gen it is the chunk count that maters. Why? because it will only roll the dice on RNG X time Per Chunk. So yes there is only 64 samples and while there might actually be some variance there to get a true bead on whats happening you need hundreds of samples.

Its Per Chunk that is the sample here not Per Block or Per Cluster.

So yeah sure your numbers suggest something. But replace what your saying with "I rolled these dice 244 times and the average 3 times what it should be" and you get a better feel for why despite there being 4 million blocks over all it seems more RNG is RNG then oh my god guys when need to cut diamond spawn to 33% or all is doomed.

Also as aside the idea of taking 244 samples and only comparing 64 sets of pairs just to me really seems out of the norm for running stats.

LOL> Im not suggesting anything about altering diamond spawn. All Im observing is "OMG, look at the difference between early diamond generation and later versions" That's all I merely set out to examine what was actually in the ground, and discovered this variation between versions.

That's why the results are grouped into two sets - version 1 (128 chunks) and version 1.1.3 (116 chunks) ... as the basis for analysis.

120 odd chunks a side is a pretty decent amount of data for comparison, so any variation due to RNG is going to be relatively small-ish .. which is what we see at the top of the page: most ores are similar, in-line with expected random variation.

But some stand out - by a lot. Look at Iron ore, for example, one of the most frequently generated blocks in the ground. It is almost double in the later version. Clearly something has changed. Unfortunately, getting into debates about statistical method has led to quibbling, while ignoring the bigger picture. Iron, Gold and Diamonds generation has changed significantly since Unleashed was released. No amount of monte-carlo random arguments can satisfactorily explain it away. Just looking for a rational explanation, as "random" doesn't seem to cut it.

OT, finding diamonds is an early game focus for most of us - and I dont mind what the spawn rate is ... as long as there's some to be had down deep

 

[namiasdf]

So... If none of you know anything about statistics and probability, a RNG is a random number generator. There are several ways to implement this in terms of resource generations. Here are a couple.

(a) Every single block, roll a "dice". Depending on the value of the dice, set block value to "x", where x is the block ID for whatever is rolled. There are more faces to this die that are stone, than there are of ores, obviously. You roll a different dice depending on what y-value you are at, as copper doesn't spawn below "n" and diamond only spawns below "m".

(b) Generate an string of numbers, similar to a SEED. i.e. The number used to generate your world. That algorithm may contain several implementations that define a systematic way to which your world is randomly generated. i.e. Put swamp next to desert next to ocean, repeat three times then start next world gen algorithm. This algorithm would also contain the RNG that determines your ore generation.

In the end, the mod can only control the probability that an ore spawns. In order to determine whether an ore gen is OP you would need to take several samples of multiple independent chunks, even worlds and determine the mean value, standard deviation and other statistical values. Once those are known you can state the ore gen rate as x +/- y per chunk and compare it to (a) other ore gen statistics or (b) previous values of same ore gen, prior to mod addition.

I have 15 minutes till the bus comes, give me a break. I'm bored.

He knows what's up.

 

[Hoff]

He knows what's up.

lolwut. Was there a point to that?

 

[BeastFeeder]

Not to stir the pot but....I'm been amazed at how many diamonds I'm getting so early in my new unleashed world. I dont have any spreadsheets to prove it, but I have a lot of diamonds already--which I for one like.

I dont play minecraft/ftb for the "challenge" of it.

 

[Juanitierno]

It's neither significant nor valid because you're not sampling blocks or diamonds, you're sampling chunks. Which means you're taking 64 samples and trying to claim it's significant relative to the statistical convention of 1000 samples.

I think your statement is flawed. He is not sampling chunks.

Sampling chunks would be: "Out of the 64 sampled chunks, how many of them contained diamonds".

Hes sampling blocks. "Out of the 64x64xdiamond-able-layers blocks, how many where diamond blocks."

Granted, due to the rarity of diamonds very large samples are needed if you want a good % of security (sorry, dont know the terms in english), but even with the current samples, is more than enough to say, with a good % of security, that the amount of average diamonds has increased.

The number 1000 also only applies to certain scenarios and depends heavily on the original probability of occurrence. For examaple if theres an illness with 1:1000000 chance of ocurring, you would probably take samples of 1000 people and none would be ill, throwing off your numbers. On the other hand, for flipping a coin where the probability is 0.5, many less samples are needed to achieve a reasonable security.

Again, sorry for the poor vocabulary!

 

[RedBoss]

 

[Juanitierno]

Thats a nice image, Im saving it.

In any case, I was not talking about the rarity of diamonds, I just got interested into the statistics aspect of the problem and the arguments exposed.
I honestly find it more interesting to find a decent methodology to do a study of this kind of thing, than how many blue rocks can be found in X amount of time.

In this case this would have been more appropriate:

And im sorry for doing it.
Probably should have opened a new thread, or just kept it to myself.

 

[AlanEsh]

Your horse appears to be in need of some mechanical agitation.

 

[Dorque]

Not that it really matters in the long run, I have far more diamonds than I could ever use at this point (pretty sure I could fill a chunk with blocks of diamond thanks purely to MFR laser drills and Fortune) but my general suspicion would be that there are actually slightly fewer diamond ores generated in Unleashed when compared to Vanilla, due to the increased number of ores/non-stone blocks that can generate in the same location.

On the other hand, people may be finding an increase simply because of flattened bedrock.

In conclusion, who cares, you can start manufacturing diamonds by the end of the first week or so =P

 

[RedBoss]

To be honest, without EE the most limiting resource for me is ender pearls. Sure, once you beat the dragon, you can swim in them but up until then that's the only true scarce resource. IMO

 

[MigukNamja]

To be honest, without EE the most limiting resource for me is ender pearls. Sure, once you beat the dragon, you can swim in them but up until then that's the only true scarce resource. IMO

This. On top of that, the first person on a server to *get* to the End has to accumulate a good 10 to 16 Pearls, depending upon distance to nearest End portal and how many pearls you need to complete said portal.

Blaze Rods are a major check/chokepoint as well. I don't mind either as a check/choke point. It forces encourages proper progression.