Hacker News Reader

Comment by sourweasel | Hacker News Reader

>"The data provided is NOT a single playable Minecraft world, but rather a highly compressed collection of several world downloads of 2b2t."

I feel like HN needs to have a small model that compares post title to the article content and assigns it an accuracy score.

View on HN

The accuracy score of this one would probably be a lot higher than pulling that sentence out alone would lead one to believe.

What is the issue? You can still download the world files, but it would be inconvenient to store them in Minecraft's unoptimal format.

"The largest available Minecraft world" implies we're talking about one world. Then "totaling 15TB" implies the world is 15TB large which would be incredibly impressive and potentially involve multiple shards of minecraft servers stitched together. But 15TB of world files means there's multiple copies of the same data. Still impressive but not the same as a single 15TB world.

The only redundant part of this snapshot is the second 512k^2 snapshot of the overworld. The End and Nether snapshots are still meaningful. Excluding the 512k^2 snapshot, the size would be around 12TB.

And the actual size of the 2b2t world is likely much larger than 15TB. The data for this project is stored in a highly compressed form, much more efficient than the game's standard file format.

Microsoft can probably scale Minecraft up to support infinitely large sized worlds if they want to. Minecraft lazy loads the world in chunks so you there's no real limitation (aside from machine digit sizes)

This has already occurred. This has been since Minecraft infdev (2009, now 17 years old). Later on, a limit of 30 million blocks (30000km) in each direction from the origin was added, due to floating-point precision issues and integer overflows in world generation, which get more severe past this point. Integer overflows in world generation caused the infamous "far lands" in versions prior to Beta 1.8.

> “far lands” For those like me that didn’t know about this. Pics at bottom.

https://minecraft.fandom.com/wiki/Far_Lands

Maybe I'm misunderstanding what this means in that case:

> a 1,024,000² (1M²) area of the Overworld (Dec 25 2025 - Apr 13 2026), > a 512,000² (512k²) area of the Overworld (Nov 11 2024 - Dec 12 2024), > a 256,000² (256k²) area of the End (Jan 23 2026 - Feb 15 2026), > a 100,000² (100k²) area of the Nether (Jun 9 2025 - Jun 14 2025)

How do they both provide a single snapshot of the world while also providing a date range.

The word "snapshot" is not entirely accurate. They used several bots to manually download the world by flying around, and that collection process takes several months (as described in the blog post).

Model? For a http request and a Levenshtein distance calculation?

Levenshtein distance and semantic distance can be wildly different. Deed -> dead, for instance.

Reply on news.ycombinator.com