Exciting times are ahead with HF20 coming to STEEM on the 25th of September. For most of us here on steem this does not mean to much but there are some fundamental changes taking place that will need a little consideration by witnesses on new parameters that need to be set.
Repository
https://github.com/steemit/steem
HF20, known as Velocity, makes changes to the account creation process works. Currently accounts can be set up with delegation, as Steemit Inc. have been doing for some time. This method of account creation will cease.
Instead the options will be Non-Discounted Accounts and Discounted Accounts.
Non-Discounted Accounts will be created by paying the Account creation fee in STEEM. From what I understand, this STEEM will be burned when the account is created.
The account creation fee is set by witnesses. Currently, this identified as ‘reg fee’ on https://steemian.info/witnesses and is multiplied by 30 to get the actual fee. So, a 0.20 fee would be 6.0 STEEM. At current prices this is very low, considering some months ago most witnesses had this set at 0.10 or 3 STEEM and the price was 10 times higher.
When HF20 happens, a onetime multiplication of x30 will take place on the account creation fees since this multiplication will no longer take place at account creation stage.
Is it time for witnesses to consider reviewing the account creation fee now that the price of steem is so low?
Discounted Accounts. This is where things get interesting and is a new way to create accounts. With discounted accounts the fee is paid with the new resource credit that will also be introduced with HF20. It’s a two step process. First the recourse credits will be used to Claim a discounted account. Once a discounted account is claimed it will be stored in their pending claims account until step two. Step two involved the actual set up of the discounted account.
Some points to note about pending claims. These do not expire, they are not transferable and there is also no upper limit to the amount one can have. But the do require recourse credits, and with SMT’s on the way, this could be a limited resource, especially for those with low SP. This could very well become an activity only carried out by highly staked accounts.
Is it a case therefore that the account creation fee should be valued not against steem but RCs?
By removing the delegation method and introducing the discounted accounts, a lot of the onus of account creation moves to witnesses. Between them the witnesses need to decide how many discount account creation tokens are created each day and the total number allowed at any one time.
Do you think the number of account creation tokens should be limited and capped and at what levels?
You can read more about these changes on
https://steemit.com/hf20/@steemitblog/velocity-hardfork-account-creation-guidelines
Exploratory Analysis
I have carried out an exploratory analysis with the aim of gathering account sign up information in a way that it can be sliced and diced so people can see when and who has been setting up accounts. Hopefully by looking at the numbers you will be better informed when considering what parameters should be set and why.
I have used Steemsql to gather the data and I have taken 2017 and 2018. Full details of the queries can be found at the bottom of the report.
The full report is available here and is fully interactive. Don’t be afraid to click around, you won’t break it.
https://app.powerbi.com/view?r=eyJrIjoiMzE5YTU1MzEtZWQ3Ni00ZWUxLWI0NmItNDc1MjFiNzYzZmZhIiwidCI6ImRhZWNkYWM0LTYyNjgtNGQyNy05Nzk0LWFiMjA0ODdkZTgyNyIsImMiOjh9
Overview
The overview page of the report contains some general information
When filters are NOT placed on this report we can see that for 2017 and 2018 combined we have over 1M new accounts. Of which 19.12% have their about profile complete and 9.19% have completed a link in the profile to a website or other source. If you have followed my monthly new user report, you will know that these values are important as it is an indication of the quality of the account.
43.43% of accounts have posted, 21.25% have posted 10 or more times and 9.63% have posted 50+ times.
I have then created an account classification based on the account actives.
Fully active accounts have both posted and voted at some stage
Some accounts have posted and never voted
Some accounts have voted and never posted
Wallet holder accounts have made a transfer but have not posted or voted
Not active accounts have never posted, voted or made a transfer.
The number and % of these accounts is seen in the pie chart.
I have also added recovery account information as it gives some indication on who is setting up accounts.
I have used the accounts create tables to gather information on who has set up each account
96.67% accounts have 
The gif shows how you can add and remove filters and interact with the dashboard.
Time Series
The second page of the report shows the time series for accounts set up, the top charts shows the months and the bottom chart shows the weeks and the same period in the prior year. This page is also fully interactive and you can filter the data by years or Account Activity type.
Forecast
Based on weekly account set up numbers I have used Power BI to predict the number of new accounts needed each week for the next 6 weeks
Fee
On this page you will find an interactive time series for 2018 on the fees paid to create accounts and the delegations given.
Delegations:
On this page you will find an interactive time series for 2018 for delegations given to new accounts so they can transact.
Explore Fees_delegation
This page will give further details on the fees paid and delegations given. it has been set up so you can explore groupings between fees, delegations and account creation.
Conclusion
The number of accounts being set up each month has declined since Jan 18 and the median monthly new user for year to date is 60K and the weekly median is 16.5K.
The vast majority of accounts have a recovery account of Steem. After HF20 I would expect to see this change considerably for new accounts as DApps will have the ability to register accounts using the discounted account model. This will in turn reduce the number of applications going though steemit inc and steemit.com. It will therefore be up to each DApps to decide on the account application process.
There are a number of things for witnesses to consider in addition to the data when deciding on these parameters. For example, if a cap is set on the total number of account creation tokens is circulation at any one time, how can the witnesses prevent illegitimate grabbing and stockpiling leaving a diminished amount for more legitimate use? Obviously, this could be countered by witnesses by increasing the cap and will have to be monitored.
Is it also worth considering how many new users we are equipped to handle? Are we ready to just open the flood gates? Or, should we keep it at a trickle, so we grow some awesome dolphins and orcas, improving the base users and current distribution first?
How many new discount account tokens should we issue each day and why?
Data and Queries
To prepare this analysis I connected to steemsql. The SQL query to get the accounts information was
Select *FROM Accounts (NOLOCK)where created >= CONVERT(datetime,'01/01/2017')and the SQL query to get the accounts involved in transfer details was
let Source = Sql.Database("vip.steemsql.com", "DBSteem", [Query="select [from], timestamp#(lf)from TXTransfers#(lf)where #(lf)timestamp >= CONVERT(datetime,'01/01/2017')#(lf)"]), #"Changed Type" = Table.TransformColumnTypes(Source,{{"timestamp", type date}}), #"Appended Query" = Table.Combine({#"Changed Type", #"transfers (2)"}), #"Removed Duplicates" = Table.Distinct(#"Appended Query", {"from"})in #"Removed Duplicates"The SQL query used to get account creation information is
Select *FROM txAccountcreates (NOLOCK)where timestamp >= CONVERT(datetime,'01/01/2017')To identify the accounts activity
the accounts table records the number of posts made, this allows us to identify accounts that have and have not posted.
The accounts table does not record the number of votes. However it does record the last vote date. adjustments had to be made to this as accounts that did not vote, recorded their last vote data as the account creation date.
To get the wallet holders only, any accounts that did not post or vote, but have made a transfer have been included as wallet holders.
Non active accounts have not voted, not posts and have not made a transfer.