Was digging into consecutive-block data after Foundry’s recent 7-block lottery.
I went from “luck?” to “interesting signal” by checking long streaks and consecutive-block lift.
Based on conversation in the previous thread, it looked like coincidence. But looking at all historical data at once, it looks like a structural issue.
7+ streaks7+ streaks
- Foundry has hit 33 streaks of 7+ consecutive blocks.
- Expected under pure hash-share odds: ≈11.
- That’s ~3× the rate you’d expect from a random model.
- Conclusion: not explained by hashpower alone.
Next question: 2+ consecutive blocks (Chart 2)Next question: 2+ consecutive blocks (Chart 2)
- If block win is independent, then
P(pool wins N+1 | pool won N)should be exactly pool hash share.
- Observed values:
- SpiderPool, SecPool, Poolin (and others) are consistently above that.
- Lift is roughly 2× expected for most pools.
- Interpretation: “winning a block makes it easier to win next block” in practice.
Why does this happen?Why does this happen?
- Propagation/latency dynamics (winner has new header/template first)
- Empty-block behavior
- Pool-specific sequencing/stacking mechanics
Don't have an answer, and we may get a definitive answer only post BNOC investigation, but the historical data definitely signals something
Anywat to learn more about mining pools, and how do they compare with each other. Check out Bitcoin Mining Pool. Here is the data to play with.
The dashboard shows
- every 7+ streak with specific block heights and dates.
- You'll notice Binance Pool and SpiderPool lean heavily on empty blocks, while Foundry doesn't do them at all.
- Plus plenty of other patterns
Your forensics dashboard is pretty awesome! I'm shocked that I didn't know about it. Great work on this! I think I need to spend a few days looking at it, there's so much there!
thanks. feedback welcome as you explore
i’ve been debating whether to expand the pool profile/behavior section. would be great if it actually helps miners decide where to mine!
I also like it!
Yes! Sv2 would be great to see - or even anything proprietary from the large pools like Ocean's proprietary solution. Giving up control may feel like a self-own for the pools, but the only thing they'd really lose is 0-fee pool distributions. Probably worth the upside of not being distrusted.
Re: your dashboard. I have a question:
Does the stated probability take into account the split-chaintip defense that was activated for Foundry when they found they had a competing tip?
agree on v2, and your thought process there.
To your question: no, the dashboard only records the winner of each block. It doesn't see the 'stale' blocks that were found but lost the race.
However, if a pool is consistently winning those 'split-tip' races, it confirms the suspicion: those 7-block streaks aren't just a statistical events, They are a result of structural edges.
One way to interpret this: while we attribute ~32% hashrate to Foundry, their effective hashrate, when they want it to be, is actually higher because of these 'other' technical edges
The 3x overrepresentation is wild. What gets me thinking is the propagation angle. If Foundry's winning a block and they already have their next template ready before anyone else even sees the new tip, that's not cheating, it's just physics. Speed of light stuff. Closer nodes, better peering, faster relay.
But here's what bugs me about it. That kind of structural edge compounds. The more blocks you win in a row, the more fees you capture from the mempool during that window where everyone else is still catching up. So it's not just "we find more blocks." It's "we find more blocks AND we get better fee selection during consecutive wins."
Curious if anyone's looked at the fee revenue difference between Foundry's consecutive blocks vs their standalone blocks. That would tell you whether the edge is just statistical or actually economically meaningful.
Great forensics work on this. The dashboard is a real public good.