Bitcoin Forum

http://s30.postimg.org/rckr8vpjj/Untitled.jpg

Anyone know why this started happening in April?  I thought the difficulty adjustments are supposed to keep it at 10? 

also, that's a completely valid image so i'm not sure why bitciontalk didn't recognize it.

Most probably the difficulty retarget wasn't great enough compared to the rise in hashing power. The difficult is designed to change every now and then to try and ensure that blocks are found every 10 minutes or so, but if there is an under-prediction then it may not be so. Also, with anything luck based there's always going to some variance which leads to a deviation from the expected value.

I really wish that image worked.  The image shows that it is not a matter of variance, and you can see the averages clearly separated.  If it was variance, it would look a lot more chaotic than this.  We have also been through many difficulty changes, and many chances to get back up to 10 mins.

It's a very clear drop, and have not yet come across a good explanation..

https://blockchain.info/charts/avg-confirmation-time

https://blockchain.info/charts/avg-confirmation-time.png

Odds are that blockchain.info's chart is simply wrong.  They had some technical difficulties a while back, it might be good for them to reparse the entire blockchain.

Here's an alternative charting service that doesn't draw their data from blockchain.info (most do, because APIs.)
http://bitinfocharts.com/comparison/confirmationtime-btc.html

If you're bored, it doesn't really take that much effort to parse the blockchain yourself.. it's just that it's a fairly big download and requires you to futz around with the clunky interface or use Bitcoin Abe or somesuch because things just couldn't have been stored in a reasonably standard database ;)

I might try it myself... need to sync the various chains anyway.

https://bitcoinwisdom.com/bitcoin/difficulty

Block Generation Time (average of last 504 blocks)
1 block: 8.4 minutes
3 blocks: 25.1 minutes
6 blocks: 50.2 minutes

First, it measures "average tx confirmation time" not "block time".
As most of the transactions with correct fee setting should get included in the next block , the "average tx confirmation time" which should be lower than "average block time".

As the difficulty was rising steadily in Jan 14 - Mar 14, the "average block time" in the period should be lower than 10 minutes.
However, the "average tx confirmation time" was obviously above 10 minutes.


Either the stat is not right, or blockchain.info includes lots of spamming no-fee tx in the calculation.

True - unfortunately the actual time between the transaction being broadcast and inclusion in the successful block is not something that is as easily found without having a well-connected node.  Also seems like an odd thing to track until transactions get significantly backlogged.  That would probably show mostly times along the same line as the block time (well-formatted transactions), then some higher ones for those who set low fees, even way higher ones for those that are generally frowned upon but some miner decides to work on them anyway, and those that never get included in the first place.

If it is indeed supposed to be an average between when they first detect the transactions being broadcast, and its inclusion in a block, then I guess it might have been the result of Core 0.9.1's changes (https://bitcoin.org/en/release/v0.9.0).  Would fit with the date anyway.