Stuck transactions

EDIT: The Bitcoin network is currently under a sophisticated DoS attack using the "fake sigOp" method...

Some weird shit is going on right now: mempool size is above 10 MB total, with a total size of transactions paying at least the default recommended fee rate (10 satoshi/byte) being 8.5 MB. (Numbers from here). So there is definitely enough fee-paying transaction to fill several blocks with.

However, most blocks are far from being full, they do not even reach arbitrary limits set by miners. For example, you can see block #385917, mined by a KnCminer, having size 912 KB, and the next block #385918 is only 487 KB. So we see that KnCminer is OK with mining blocks as big as 912 KB, but found only 487 KB of worthy transactions.

Thus we can conclude that some anti-spam filtration measures are on, but what are they?

I think we might benefit if miners state the policies they are using, as it gives wallet developers a chance to change the way they make transactions, or at least warn users if transaction won't be confirmed soon

I just got a transaction confirmed after 4 hours of waiting. It wasn't a low-fee transaction: it paid about 20 satoshi per byte, which is twice the default fee rate. I think I know why it was delayed: it had an output which is only 3400 satoshi, which is only slightly above dust threshold, perhaps it made transaction look spam-like.

It's frustrating that miners' policy is so opaque. Let's look at block #385927, which is mined by F2Pool. It is 340 KB in size, which is smaller than other blocks recently mined by F2Pool, so we know it didn't reach the block size F2Pool set. No block explorers I'm aware of show transaction fee distribution for a specific block, so we have to do this manually. However, we can take advantage of the fact that Bitcoin Core sorts transactions by fee rate, thus if F2Pool uses block constructor based on it we will see this pattern.

5 of the last 7 transactions in the block pay almost exactly 20 satoshi per byte, so this seems to be a cut-off rate. The last two pay less than that rate, however, they coalesce many outputs, so it might be a result of prioritization.

The question I have is why do we have to learn about it through guesswork, aren't miners themselves interested in Bitcoin being a reliable payment system?

It's hard to figure out what the fuck they want from us even using fairly sophisticated statistical tools.

Submitted November 30, 2015 at 04:24AM by killerstorm http://bit.ly/1lSXR8u