Quote from: TracerX on November 07, 2015, 05:56:46 AM

Quote from: eleuthria on November 07, 2015, 05:45:11 AM

EDIT: https://bitcointalk.org/index.php?topic=700411.msg11790734#msg11790734 - F2pool/Discus Fish publicly stating that even after SPV mining by chinese pools caused a fork in 2015, they had no intention to stop doing so, and don't believe that Antpool or BTC China will either. You can see clearly that all 3 pools are still doing it just by looking at the sheer number of empty blocks they mine each week.

I'll stop thread-jacking, but I've got one more question if you would be so kind. Under what circumstances could an empty block be mined "legitimately?"

A few years ago (I'd say anytime within the last 2 years this wouldn't apply), there were periods of slow network activity (1 tx every few seconds rather than 1-2/sec). If a pool got really lucky, the following events could happen:

1) Pool pushes out work to a miner that includes all transactions the pool knows about.
2) Miner happens to solve a block almost instantly after receiving this work and sends a winning share back to the pool.
3) Pool verifies the share and pushes out a new block, and has not yet received a single transaction, thus meaning the pool is not aware of a single unconfirmed transaction to include.
4) Pool pushes out new work with no transactions because it isn't aware of any.
5) Pool solves another block before it has sent out updated work with transactions in it.

Those events could only happen if the pool *just* created work and the miner solved it almost instantly, and the two were very low latency from each other. Obviously, the further back in time you go, the larger the window of opportunity, since transaction activity was significantly lower the further back you go.

Today, there's pretty much only one way a 1-tx block could legitimately happen:

1) Pool restarted its bitcoin node, which wipes out its knowledge of any unconfirmed transactions on the network
2) Pool software created a unit of work before the bitcoin node was notified of a single transaction.
3) Miner solves this unit of work before the pool has sent out new work that includes transactions.

I can't think of any other legitimate reason for a 1-tx block today. The network is *so* active at all times, that the odds of a pool being able to actually fit every transaction it knows about into a block, sending that work to a miner, and a miner solving that block before the pool has seen a single other transaction, creating a new block without transactions (because there aren't any), and then solving another block off that work, should be a once every few months occurrence (edit: and that's a VERY GENEROUS estimate of frequency). Not a multiple times in a single day occurrence like we see from the SPV mining pools.

Just going from my own evidence: BTC Guild, as far as I'm aware, did not have a single 1 TX block to its name from mid-2013 until it closed in June 2015. That includes a multi-month period where it was 35-45% of the network, had multiple chains of 6+ blocks in a row solved by its own bitcoin nodes, and multiple block solves that happened within seconds of each other. I'm trying to find the data somebody put together of pools with 1-tx blocks, I believe BTC Guild was 3 out of over 32,000 blocks, and I'm fairly certain those blocks were from 2012.