Happy New Year! — Qtum Mainnet Performance February 9–15
Greetings Qtum friends. For those new to the community, I was publishing a weekly report about Qtum Mainnet performance starting with the public launch last year, until the network stabilized and I ran out of interesting topics to write about. See links to that series in the references below.
Now there are two events to celebrate and the network is changing as the node count is jumping with a huge deployment in China. It is time to dust off the Excel formulas and update the charts, so here we go…
I am a community member and independent researcher, not affiliated with the Qtum Team, but I do appreciate their technical guidance and the robust discussions in the community.
First, Happy New Year, the year of the earth dog!
The Lunar New Year is observed in Singapore, China, and other Asian countries. May you have a happy, healthy and prosperous new year, and receive many red packets filed with QTUM block rewards.
This week there was also Valentine’s Day celebration on February 14 in the West, the day spouses, lovers and stalkers exchange cards, chocolates and flowers professing their true love, (promoted by the card, candy and flower industries). If you observe Valentine’s Day, I hope you had a sweet day with that special someone.
The Qtum Team sent a special Valentine, a mandatory update for the Qtum core wallets to version 0.14.14, with new code activated at block 100,000, reached at 03:25:04 UTC on February 14. This update is required for continuing proper performance of the Qtum core wallets (qtum-qt and qtumd) but does not affect the other wallets. If you have not updated (shame!) now would be a good time to do it. Make sure your wallet.dat file is backed up and just do the update.
Qtum Mainet reached block 100,000 at 03:25 UTC on February 14, 2018, from its initial launch on September 8, 2017.
Charts and Graphs
Data sources for this review of the Qtum Mainnet performance come from the Qtum Explorer, ripping the blockchain, logging from the qtumd server application and encrypted Telegram messages from the gazebo in the Bellagio Conservatory and Botanical Gardens.
Unique Reward Addresses
Your correspondent screen scrapes block reward addresses from the Qtum explorer, and uses Excel string formulas to crunch the data. For the months following Mainnet launch the typical number of unique reward addresses per day was the in range 250–290. Out of approximately 600 block rewards per day, this means that a little less than half of the block rewards were won by small wallets, and a little more than half were won with multiple block rewards per day by bigger wallets (typically wallets with 20,000 QTUM, or much more). For the last 7 days, unique addresses per day peaked at 292 on February 12th and averaged 278 per day.
In November and December unique reward addresses per week were in the 950 range. For this recent 7-day period, there were 1,028 unique addresses, showing a slight increase potentially from the new China nodes.
Transactions per Day
The number of transactions per day can be seen at QTUM Explorer.io and at the bottom of the Qtum Explorer home page. In the months after Mainnet launch the number of transactions was typically in the 3,000–8,000 range, but now the network has grown and is recently in the 6,000–8,000 range:
I would be remiss if I didn’t point out the growth of transaction fees. You can see the fees for the last 24 hours on the Explorer Stats page. The incentive and compensation for stakers is the block reward and a (one-tenth) share in the transaction fees for each block. Last time I analyzed the transaction fees in mid-December the transaction fees were peaking at 8% of the block rewards.
What difference two months (and a whole lot of smart contract deployments) makes. You can see below that daily Qtum transaction fees are now half the daily block reward fees (in the example below for a 24-hour period there were 1,579 QTUM in fees and typically 2,400 QTUM for block rewards), and we are just getting started with DAPPs (Distributed Applications) and smart contracts. With the expected growth of Qtum DAPPs we could see the transaction fees exceeding the block reward fees later this year.
A special shout out to blocks 100,135 and 100,783, with huge transaction fees of 8.1 QTUM, driven by smart contract calls with transaction fees of 8 QTUM. I’m sure the stakers appreciate these juicy fees.
Average Block Time
Average block time has held steady in the 144 second range, which gives approximately 600 blocks a day. There is variation in the block spacing, caused by randomness in the SHA256 hash algorithm which is used to select the Proof of Stake block reward winner, with shortest blocks of a few seconds and longest blocks in the range of 30+ minutes. For the first two weeks of February there were a total of 8,752 blocks, 15 with more than 20 minutes spacing, and the longest block was #98,210 at 31:12.
Wallet Sizes for Block Award Winners
For this next section I reprise some previous analyis to determine the actual wallet weight for wallets winning block rewards. This helps answer the question of wether small wallets can win block rewards, and the answer is — absolutely. It will be interesting if this analysis can tease out anything about the new Chinese nodes.
Before looking at the results we can do a thought exercise. Suppose the network weight was 20,000,000. There would be some distribution of block reward winning wallets where the really big wallets were winning block rewards proportional to their heavyweight status. Now in the thought exercise, suppose someone were to add another 20,000,000 network weight of wallets, but split into tens of thousands of small wallets.
If you understand that the probabity of winning a block reward depends only on the weight of a wallet vs. the network weight, then the thousands of new small wallets must win half of the block rewards. This would show up as a bunch of small wallets unexpectedly winning half of the block rewards, and cutting the original wallet’s winnings by half.
I have described the methodology for determining block reward winner balances in a previous report, so I will just say that the data is captured using a screen scraping macro. During the process I noticed that about 3% of the winning wallets held 1 or more tokens.
For the period February 9 to February 15, there were 1,028 unique addresses winning block rewards. The table below shows the groups of block reward winners:
The smallest wallet winner for this period scored block #99,071 with a wallet weighing in at 8.096 QTUM, the block reward giving a nice 50% return!
This histogram chart below shows the distribution of block reward winners with weights of 100 to 1000, in buckets of 10:
Notice there a bunch of wallets with sizes of 400 and 500. As compared to the histogram for wallets of this same size range in early November a large group of 250 QTUM wallets has gone missing, and there are still big groupings at 400 and 500. Many of these 400 QTUM wallets were funded in the last week or two (their first transaction is only 5,000 to 8,000 blocks ago). These wallets are very new, and very lucky to be winning a block reward in two weeks when the expected time to a block reward for a 400 QTUM wallet is about 90 days.
The probability of block reward mathematics says that these lucky new 400 QTUM wallets are probably part of much larger cohort of 400 QTUM wallets, and an educated guess is they are the new nodes we see in China. The Qtum nodemap currently shows about 1,200 nodes in China, up from about 180 at the start of the year. But I can’t be sure — let’s see what it looks like next month.
Network Weight is used by wallets to display the average time to a block reward for a given weight in a Qtum wallet. You can also see a nice Qtum staking return calculator that uses network weight here.
Network Weight is calculated by the wallets as a moving average of the staking difficulty of recent blocks. The actual network weight (the total number of coins staking) is unknowable for a decentralized network, and this calculation of network weight by the wallets is useful, but suffers from large random swings. The chart below shows the difficulty (how difficult it is for the stakers to win the next block reward) for the first two weeks of February:
This chart doesn’t show an increase in network weight during the first part of February, but analyzing the block rewards received by really big wallets of known weight we can give a better calculation of network weight.
Think of it this way: if wallets holding 6,100,000 coins win 23.8% of the block rewards over a week, then the network weight must be 6.1 million / 0.238 = 25.6 million. Using this formula, I examined the block rewards for big wallets over the period February 9 to 15, where big wallets with a known balance of 6.1 million won 1,001 block rewards. This means the current network weight is in fact 25.6 million, and I encourage you to ignore the network weight provided by the wallet calculation, which is saying 16.5 million early on New Year’s Day.
The network weight of 25.6 million represents an annual return for staking wallets of 3.4%:
There is a nice list of the biggest staking wallets at QTUMExplorer.io (thank you Calbert).
Yet Another Simulator
Curious about how the algorithms controlling difficulty adjustment and block spacing interact with other consensus parameters, I wrote a full-featured Qtum system simulator, which comes in handy to predict changes from the new nodes.
Based on the wallet weight analysis above I set the simulator to add 50,000 wallets of 400 QTUM each, spread over a time period of 30,000 blocks (about 2 months). The chart below shows growth of true network weight and the wallet-calculated network weight during this period. The blue line shows true network weight (the simulator knows the actual weight of each simulated wallet) vs. the yellow line shows how the network weight would be calculated by the wallet.
If this is the deployment configuration for the new nodes (50k nodes of 400 QTUM each) which is only a guess at this point, the simulator shows that Mainnet performance metrics such as average block interval and block spacing variation remain unchanged, and network weight adds 20 million. This simulation run of 30,000 blocks ending with 51,500 wallets took 12.5 hours on an i5/2.3 GHz CPU, generated 20 blocks with spacings over 20 minutes, had an average block spacing of 143.5 seconds, and ending true network weight of 45 million.
There was some concern on the social channels this week about “concentration” by a single entity owning so many nodes, but rest assured Qtum is a completely decentralized system. All the nodes (wallets) operate completely independently, as guaranteed by the open source software. As this deployment continues it will increase the security and resiliency of the Qtum network. Having more nodes than bitcoin and Ethereum combined will be a remarkable advantage for Qtum.
Happy New Year! Happy Valentine’s Day! Happy Block 100,000!
1. See my past reports, with these topics:
Testnet and Qtum Mainnet Performance January 1–8, The Qtum Test Network (testnet). (Rio de Janeiro, Brazil), Published on Jan 7, 2018
Qtum Mainnet Results December 25–31, Using Virtual Private Networks (VPNs) for staking. (Seoul), Published on Dec 31, 2017
Qtum Mainnet Results December 18–24, Transaction fees. Published on Dec 24, 2017
Qtum Mainnet Results Dec 11–17, Passphrases, a passphrase guessing script. (Genève, Switzerland), Published on Dec 17, 2017
Qtum Mainnet Results Dec 4–10, The memory pool (mempool) and processing of unconfirmed transactions. (Sydney), Published on Dec 10, 2017
Qtum Mainnet Results Nov 27 — Dec 3, Orphan blocks (Hong Kong — Lantau Island), Published on Dec 3, 2017
Qtum Mainnet Results November 20–26, The QTUM.explorer.io Contracts page (Berlin), Published on Nov 26, 2017
Qtum Mainnet Results November 13–19, A really simple example of the SHA-256 hash algorithm and Difficulty (Berlin — Friedrichshain), Published on Nov 20, 2017
Qtum Mainnet Results November 6–12, Distribution of wallets winning block rewards, the Big Five Network Weight (Beijing), Published on Nov 12, 2017
Qtum Mainnet Results October 30 — November 5, The SHA-256 hash algorithm, the Target and Difficulty (Seoul), Published on Nov 5, 2017
An Introduction to Qtum Proof of Stake Mining — A Racing Story, An ELI5 story about PoS mining (Berlin Marathon), Published on Oct 29, 2017
Qtum Mainnet Results October 23–29, Foundation wallets leave the field (Berlin), Published on Oct 29, 2017
Qtum Mainnet Results October 1–8, Unofficial staking FAQ, Published on Oct 26, 2017
Qtum Mainnet Results October 16–22, Network weight, time to reward statistics and Nodemap (Vivid Sydney 2017), Published on Oct 22, 2017
Simulating Proof of Stake Mining for Qtum, A simulation to pick the correct size of UTXOs for PoS mining, Published on Oct 21, 2017
Qtum Proof of Stake Mining, My version 1.0 simulator for PoS mining, Published on Oct 16, 2017
Qtum Mainnet Ignition Results October 9–15, Introduction to Mainnet charts and graphs (Singapore), Published on Oct 15, 2017
2. Here are some knowledgeable YouTube (English) vloggers on crypto who publish daily vlogs:
Ivan Liljeqvist, — (Ivan on Tech) news coverage, technical details
Michael Gu, — (Boxmining) news coverage, interviews with founders & new projects
Finally, a different take on drone footage at the Bellagio Hotel and Casino — CES 2018 Intel Drone Light Show