Share our plausible theory for the asymmetry Karl noticed on pooltool.io battles

[nfrisby]

@karknu shared an unexpected observation on pooltool.io. A pool had a large number of concentrated "slot/height battles" (ie short forks) listed on pooltool.io, and moreover it won nearly all of them (eg 11/12).

Since the VRF tiebreaker is supposed to be "uniform", the lop-sided ratio is not what we were expecting. We chatted and came up with a plausible theory based on survivorship bias. It seems worthwhile to write this down, listing our assumptions and possible action items (eg calculating how many battles we expect to see, etc).

cc @amesgen

[nfrisby]

Some follow-up questions that our theory depends on: how does pooltool.io parse the node logs? Do pools report their block-producing node's logs or instead only their relay's logs?

[amesgen]

Relevant info:

• The pool in question
• Epoch 453 and 454 feature this prominently, with ~10 battle wins in each epoch, but only 1-2 losses, see the "Blocks" and "Orphans" tab.