Measuring expected block timing for Difficulty Adjustment
Blocks are expected to be added to the Bitcoin blockchain every 10 minutes
However due to a variety of circumstances, this doesn’t always occur.
Speculation over the last day is around there being a fire at a large mining farm, however the validity of those claims is less relevant to the discussion because a variety of factors could occur at any time. We’ve seen flooding, fires, power outages and more all affect it.
This is something that has been discussed by Jared Tate, founder of DigiByte, on a number of occasions over the years. It was this slow adjustment that was the inspiration behind DigiShield in the first place, which DigiByte pioneered in early 2014. It was subsequently implemented by DOGE with the DigiByte developers assistance, and can now be found in everything from BCH / BSV to ZCash.
What is MultiShield?
MultiShield is a multi-algorithm version of DigiShield: a real-time difficulty adjustment algorithm from MentalCollatz.
Instead of taking 2 weeks to adjust as Bitcoin does (or 3.5 days for Litecoin), DigiByte adjusts it’s difficulty every single block. It is this difficulty that determines how often a block is expected to be found by the expected hash-rate, to maintain the intended block timing.
It does this asymmetrically and allows for the difficulty to rapidly rise and fall, to account for an influx / outflow of hashrate on any given blockchain.
NOTE: DigiByte now being MultiAlgorithm maintains 5x independent difficulties for each of the 5x algorithms that compete in every single block.
There are dozens of blockchain projects utilizing DigiShield, and for good reason: It’s simply more accurate with the block emission schedule, and it prevents things like chain freezes. This was why DOGE implemented DigiShield originally, after blocks were taking almost days to be found in early 2014.
This subject also came up recently when the Litecoin block reward halving event occurred and the block timing was completely crazy. It has since slowed significantly, putting block timings all over the show. It was for this reason that I submitted a proposal for Litecoin to adopt DigiShield.
A more uniform, reliable emission for block timing is beneficial for all. There is no down-side to having a more reliable emission curve.
Just how good is MultiShield?
I look at the last 24hrs worth for Bitcoin, Litecoin and DigiByte (With MultiShield). Using 144, 576 and 5760 blocks respectively to get the data, and compare the time it should have taken vs the time it actually took.
BTC
Current: 597311, 2019–09–30 22:53:25
144 blocks ago: 597167, 2019–09–29 20:52:41
Out by: 2 hours and 44 seconds (8.4%)
LTC
Current: 1711657, 2019–09–30 23:05:30
576 blocks ago: 1711081, 2019–09–29 21:49:12
Out by: 1 hour, 16 minutes and 18 seconds (5.3%)
DGB
Current: 9521659, 2019–09–30 23:02:48
5760 blocks ago: 9515889, 2019–09–29 23:00:52
Out by: 1 minute and 56 seconds (0.1%)
That’s quite a difference between the three!
Clearly the realtime difficulty adjustment of MultiShield is working quite well, keeping the block emissions within 0.1% of their expected emission schedule.
What if we look at 10 days worth?
Good question, let’s look back a bit further and see, taking the same 10 days worth from the aforementioned “current” blocks for all three projects.
BTC
1440 blocks ago: 595871, 2019–09–21 05:36:23
Out by: 6 hours, 42 minutes and 58 seconds (2.8%)
LTC
5760 blocks ago: 1705897, 2019–09–20 17:33:24
Out by: 5 hours, 32 minutes and 6 seconds (2.3%)
DGB
57600 blocks ago: 9464059, 2019–09–20 23:11:23
Out by: 8 minutes, 35 seconds (0.06%)
We can see BTC / LTC even out a further over the long term, but BTC swung from almost 7 hours too fast to 2 hours too slow.
These wild swings are not helpful when you are waiting for a transaction.
Visualized
Sometimes it’s easier to visualize how far off the expected block time these projects are, if we put it in to a graph form. I found this fascinating:
We can see here that DigiBytes variation both over 1 day or 10 days is far, far superior. As such, I have no hesitation in suggesting that DigiShield (Or for MultiAlgo: MultiShield) be should seriously be considered by any project.
