Reducing the VRF Variance for Parcels

Yes I think this small overlap in your distribution can be a fun feature, for rare very positive outcomes (much bigger than the originally proposed maximums), while eliminating the corresponding rekt experience. And it is a really small number: only about 2% would roll high enough to overlap (above the floor of the next size) given these numbers:


I’d also think that if this option wins, the important thing is the general shape/intent of the distribution, and Pixelcraft would still be able to tweak the exact params as needed.

It’s a huge difference from overlaps in the original 0.2 - 1.8 distribution (still assuming that was meant to be flat - I haven’t seen an official confirmation of that!). I just ran the numbers to look at the overlap there, i.e. a simple proportion of the flat range that overlaps the adjacent size ranges:

  • 62% of Humbles rolling up into the Reasonable range
  • 12% of Reasonables rolling up into the Spacious range
  • 15% of Reasonables rolling down into the Humble range
  • 1.5% of Spaciouses rolling down into the Reasonable range

proof of concept contract. use easySimulateBinomial with any integer seed to try some rolls. the average is 100 million in this simulation.


glmr based rerolls are self limiting, and will take from someone’s ability to compete for crests or speed up production. having four options(sell it, crest it, channel it, reroll it) of what to do with glmr, which is actually the rarest resource in the game, makes for very interesting strategy options.

In fact, the reroll protects the small player that we are worried might get burned and incentivizes them to stake the gex. They can stake a little as a hedge against low rolls, instead of staking to try get a crest. This also helps push people with no intent of getting a crest away from simply using their glmr to drive up crest prices.