1-block confirmation fee estimates are absurdly high for no good reason. What is going on here?

There appears to be something odd going on with fees being paid on Bitcoin transactions. As of writing, there have been a bizarrely high number of extremely high fee transactions over the last 24 hours, and there continue to be a large number of these (according to bitcoinfees.21.co). Others have noticed the absurdly high fees being suggested for 1-block confirmation; Mycelium is suggesting a $2.43 transaction fee to me for 1-block confirmation versus $0.10 for a 3-block confirmation, and Bitcoin Core is acting similarly at ~1102 sats/byte for 1-block confirmation versus ~63 sats/byte for 2-block confirmation. You might think this could be due to a volume spike, but it really isn't; there is in fact so little transaction volume that my node has dropped the minimum relay fee for its mempool. What actually seems to be the case is that there are just a large number of transactions paying needlessly large fees. Anything paying over ~60 sats/byte should be pretty much guaranteed to get into the next block given the current fee rates and volume, yet for some reason, multiple wallets are asking users to pay over 1000 sats/byte for next block confirmation. It seems that somehow, high fees have gotten various software to over-estimate the fee required for fast confirmation, resulting in people continuing to make these overpaying transactions, which continues the trend.

I'm also noticing another odd feature of this transaction mix--for some reason, the extremely high fee transactions do not seem zero out after a new block is found. Watching bitcoinfees.21.co, around 40-60 of these transactions seem to stay in the mempool (or quickly be put back in) after a block is found, which then will gradually increase in number until the next block and repeat the cycle. At first I figured that the trend was just reinforcing itself as people continued to pay the fees suggested to them by their wallets, but seeing a non-zero floor on the number of these high fee transactions makes me wonder if there is something else behind it. Maybe someone is intentionally throwing a bunch of high fee transactions at the network to manipulate fee rate estimates.

Is anyone else able to shed some light on what might be going on here?

EDIT: Just as I post, we find three blocks in quick succession, and this happens. Nearly everything paying above 21 sats/byte got cleared out, but the floor on extremely high paying transactions remains--what it looked like happened is that those transactions got eaten up, but new ones were quickly made to bring the number back up.

TL;DR There are a large number of extremely high fee (over 1000 sats/byte) transactions being made despite there being low transaction volume. It is possible that someone is manipulating fee estimates, as the number of transactions paying these rates seems to immediately refill to around 40-60 after blocks are found.

Submitted November 07, 2016 at 08:02AM by 4n4n4 http://bit.ly/2etlmAW