Remarkable sudden increase in confirmation time. Another fork?

[phlogistonq]

On the 5th of november the transaction confirmation time very suddenly increased dramatically, to about 18 minutes and did not recover fully since.

Why? What happened?
The only comparable event was the hard fork that happened on march 11th.
A difficulty increase happened on the 5th, but I doubt this caused so many miners to suddenly give up that it caused this increase.

Any theories?

[1714163806]

1714163806

[1714163806]

1714163806

[dree12]

On the 5th of november the transaction confirmation time very suddenly increased dramatically, to about 18 minutes and did not recover fully since.

Why? What happened?
The only comparable event was the hard fork that happened on march 11th.
A difficulty increase happened on the 5th, but I doubt this caused so many miners to suddenly give up that it caused this increase.

Any theories?

The transaction confirmation time is different from the average block time. In this case, the peak in average confirmation time is probably due to a recent sharp increase in transaction volume, which makes blocks more competitive.

[Carlton Banks]

The latest block solution Difficulty change happened on November 5th.

We've been experiencing a consistent 8-10 blocks an hour since August, this is much faster than where the difficulty adjustments try to maintain that rate (the target is 6 per hour). All due to the deployment of unprecedented amounts of hashing power to the mining network since August. We appear to have halted the rate increase in network hashing power for now, block discovery is back down to 6 per hour. Probably for only a few days at most.

[Come-from-Beyond]

Someone is trying Selfish Mining?

[Mike Hearn]

The best way to bring conf times back down is for miners to reconfigure their soft block size limits upwards (like to 1mb).

[Mota]

Someone is trying Selfish Mining?

I really do not think that this theory is applicable in reality. You would need more than the theoretized amount of the network because they did not factor in variance. Say bitcoinguild would try that: they have a block and don't report it and hash further. They don't hash the next block in 10 minutes, eligius finds a block and reports it. So far Selfish mining would still work. But what if they found 4 blocks already? Say variance is a bitch and the other pools find 5 blocks in a short timeframe. Then bitcoinguild has lost 4 blocks they could have mined. In theory Selfish mining could be a threat to the integrity of the blockchain, but IMO they did not factor in variance as the keyfactor.

[Come-from-Beyond]

It doesn't matter if it's applicable. Someone still could try. Btw, 50BTC could try that... https://bitcointalk.org/index.php?topic=277600.msg3268431#msg3268431

[yona]

still going on 

[yona]

no, this is not a 20%-30%  constant increase, the latest confirmed block was found 40 min ago

[Sukrim]

Blocks are overfilled and transactions are spammed all over the net like there is no tomorrow - there are simply not enough fees paid to get into blocks quickly.

Edit: See http://blockchain.info/charts/avg-block-size

[hannesnaude]

The transaction confirmation time is different from the average block time. In this case, the peak in average confirmation time is probably due to a recent sharp increase in transaction volume, which makes blocks more competitive.

That graph is transactions-per-block which you would expect to go up if there are more transactions or fewer blocks. However, no corresponding increase is visible on this graph of transactions-per-day, suggesting that there are indeed fewer blocks being produced. This has to be a sudden drop in hash power or variance. My gut-feel is that the swing is too large to be explained by variance, but one would have to do the math to be sure, preferably using only block times from after the OP to eliminate sampling bias.

[DannyHamilton]

At the moment there have been 424 blocks since the last difficulty increase.

The last difficulty increase happened 67.5833 hours ago.

At the expected average rate of 6 blocks per hour, that gives us an expected average number of blocks of 405 blocks.

There have been 19 more blocks than the average expected number of blocks by now.  This indicates that blocks are coming faster than 10 minutes on average, not slower.