Document Metadata
• Creators: @ned; @theoretical
• Developers: @theoretical; @vandeberg; @itwasntme; @zgredek; @pychol-mychol;
@small.minion; @youkaicountry; @picokernel
• Contributors:
@sneak; @vandeberg; @valzav; @youkaicountry; @justinw;
@goldibex; et al.
• Sketch designs: @pkattera
• Copyright (c) Steemit, Inc. 2017
• GitHub: https://github.com/steemit/smt-whitepaper/blob/master/smt-manual/
manual.md
Smart Media Tokens (SMTs)
A Token Protocol for Content Websites, Applications, Online
Communities and Guilds Seeking Funding, Monetization and
User Growth
Steem’s Smart Media Tokens (SMTs) give anyone the power to launch and sell Proof-ofBrain [1] tokens, which are tokens distributed by “upvote” and “like”-based algorithms
and can be integrated with websites to align incentives and spur growth, while websites are
empowered to adopt sustainable, currency-centric revenue models. This model has been
tested and continues to be proven by steemit.com, busy.org, chainbb.com, dsound.audio,
dtube.video and other Steem interfaces, which are monetizing content, tokens and media
in a way never before seen.
Several popular token protocols, such as Ethereum’s ERC-20, allow you to create and
launch arbitrary tokens, but no protocol enables content businesses to leverage those tokens by aligning incentives between users and applications. Due to suboptimal transaction
cost structures that incur fees for basic actions such as voting or posting, misalignment
of interests between meta and core tokens that aren’t built for influencing distributions
based on Proof-of-Brain, private key hierarchies that don’t cater to social versus financial
operations, and slow transaction speeds that are out of sync with real-time websites - none
of these protocols could ever provide an acceptable user experience for content websites,
such as Twitter, Reddit (even subreddits) or The New York Times.
For content websites and tokens, incentive alignment between websites and users comes
from a steady, as well as decentralized and mathematically guaranteed, release of new
tokens, and incentives that must be allocated to the users - including bloggers, vloggers,
commenters and curators. The distribution of new tokens occurs based on stake-weighted
voting to prevent gaming and eliminate the need for a counterparty. Quality user experience comes from tokens that can be transacted safely (through separate private keys for
distinct sets of actions), without fees, and at real-time speeds. Further incentive alignment comes from a company’s ability to raise capital in ICOs. All Smart Media Tokens
have built-in ICO support, should the issuer wish to launch one.
1
Table of Contents
Document Metadata
1
Smart Media Tokens (SMTs)
A Token Protocol for Content Websites, Applications, Online Communities and
Guilds Seeking Funding, Monetization and User Growth . . . . . . . . . .
1
1
Introduction
Leveraging Tokens for Autonomous User Growth . . . . . .
New Fundraising Opportunities . . . . . . . . . . . . . . . .
Immediate Liquidity . . . . . . . . . . . . . . . . . . . . . .
Shared Bootstrap Tools . . . . . . . . . . . . . . . . . . . .
Monetizing with Shared Token Rewards . . . . . . . . . . .
Can My Entity Participate in SMTs? . . . . . . . . . . . . .
Use Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1 - Content Publishers - Single Token Support . . . . .
2 - Forums - Multiple Token Support . . . . . . . . . .
3 - Comments Widget for Online Publishers . . . . . .
4 - Sub-Community Moderators and Managers . . . .
5 - Arbitrary Assets - Tokens Representing Real World
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
Assets
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5
5
6
6
6
6
7
7
7
8
9
10
11
Owner’s manual
Create a control account . . . . . . . . . . . .
Control account security . . . . . . . . .
Token consensus . . . . . . . . . . . . .
Token Generation and Initialized Parameters
SMT object creation . . . . . . . . . . .
SMT pre-setup . . . . . . . . . . . . . .
SMT setup . . . . . . . . . . . . . . . .
Token units . . . . . . . . . . . . . . . .
Unit ratios . . . . . . . . . . . . . . . .
Cap and min . . . . . . . . . . . . . . .
Hidden caps . . . . . . . . . . . . . . . .
Generation policy data structure . . . .
Examples and rationale . . . . . . . . .
Launch . . . . . . . . . . . . . . . . . .
Full JSON examples . . . . . . . . . . .
Inflation Parameters . . . . . . . . . . .
Named token parameters . . . . . . . .
Parameter constraints . . . . . . . . . . . . .
SMT vesting semantics . . . . . . . . . . . . .
Content rewards . . . . . . . . . . . . . . . .
Curve definitions . . . . . . . . . . . . . . . .
Target votes per day . . . . . . . . . . . . . .
SMT Setup GUI Sketch . . . . . . . . . . . .
Votability and Rewardability . . . . . . . . .
Static Token Parameters . . . . . . . . . . . .
Mandatory token parameters . . . . . . . . .
SMT interaction with existing operations . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
12
12
13
13
13
13
14
15
16
16
16
17
18
18
20
21
29
33
34
34
35
35
38
39
40
41
42
42
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Automated Market Makers for SMTs
Setup . . . . . . . . . . . . . . . . . . .
Basic Definitions . . . . . . . . . .
Notes on Conventions . . . . . . .
Finite Trades . . . . . . . . . . . . . . .
Basic Definitions . . . . . . . . . .
Computing the Restoring Trade . .
Computing the Equilibrium Price .
Example . . . . . . . . . . . . . . .
Infinitesimal Trades . . . . . . . . . . .
Setting up the Problem . . . . . .
Solving the DE’s . . . . . . . . . .
Qualitative discussion . . . . . . . . . .
FAQ . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
43
43
43
43
44
44
44
44
45
46
46
46
47
47
Costs of SMT Operations And Bandwidth Rate Limiting
Fee-less Operations Necessary for Quality User Experience . . . . . . . . . . . .
50
51
Decentralized Exchange
Automatic Order Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Diverse Asset Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Zero Trading and Transfer Fees . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
51
51
52
Augmenting SMTs with Additional Native Contracts
Community Building with Paid Positions . . . . . . . . . . . . .
Democratic SMTs using Whitelist Oracles . . . . . . . . . . . . .
Secondary ICOs for Contiguous Fundraising . . . . . . . . . . . .
Bandwidth Sharing with SMTs Based on Reserve Liquidity Pools
52
52
53
53
53
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
What Makes SMTs Better Suited to Application-Specific Blockchains, such
as Steem, than Application-General Blockchains, such as Ethereum?
SMTs are Safer and More Cost Effective in Application-Specific Blockchain Environments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
SMTs on Steem have Aligned Proof-of-Brain Incentives with the Core Token .
SMTs on Steem Have Transaction Pricing that Contributes to a Quality User
Experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
SMTs Benefit from a Blockchain that has Scaling Processes Programmed to a
Specialized Set of Applications . . . . . . . . . . . . . . . . . . . . . . . .
SMTs Benefit from a Blockchain with Content Management System (CMS)
Primitives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Increasing Market Demand for STEEM with SMTs and Implicit Value
Drivers rather than Fees
STEEM Purchased for Transaction Bandwidth Enables Maximally Profitable
Participation across SMTs . . . . . . . . . . . . . . . . . . . . . . . . . . .
STEEM Supply is Locked into Liquidity Pools by Automated Market Makers .
STEEM and SMT Demand Increases with Advent of New Powers of Influence .
STEEM Demand Increases with Proliferation of SMT ICOs . . . . . . . . . . .
Steem: The World’s Advertising Network . . . . . . . . . . . . . . . . . . . . .
Steem Ecosystem Support for SMTs
Integrating SMTs into Websites and Apps . . . . . . . . . . . . . . . . . . . . .
3
53
54
54
55
55
56
56
56
56
57
57
57
58
58
APIs and Documentation . . . . . . . . . . . . . . . . . . . . . . . . . . .
Shared Tools for Account Creation, Key Signing, and Wallet Functions . .
58
58
Conclusion
58
References
58
Appendix
Implementation Notes . . . . . . . . . . . . . .
SMT naming standards . . . . . . . . . .
Asset directory standards . . . . . . . . .
UI guidelines for SMT names . . . . . . .
Operational guidelines for asset directories
Asset directory formats . . . . . . . . . .
Unit Tests . . . . . . . . . . . . . . . . . . . . .
4
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
58
58
59
59
60
60
60
61
Introduction
Smart Media Tokens (SMTs) is a proposal to build a consensus-level token issuance protocol on the Steem blockchain. Inspired by the revolutionary properties of the STEEM
token, including automatic distributions to content creators, SMTs will be an upgrade
beyond previously created blockchain token issuance protocols due to carefully designed
token sale programmability, automated liquidity providers, decentralized token markets
and dynamic token distribution parameters, as well as a large ecosystem of tools (open
source wallets, shared key signing tools, etc.) for integrations at website and application
layers.
SMTs are an evolution of the successful relationship established between STEEM and
the social websites sitting atop of it, such as steemit.com, which has grown to be a top
2100 website in Alexa rankings in less than one year, solely from integrating the incentive
model of STEEM. With SMTs, any website or content library across the internet may have
one or more tokens integrated into its interface to facilitate fundraising and autonomous
growth.
These tokens are designed to allow website operators flexibility during the integration of
the token into their community by choosing from many parameters that may be structured
creatively at outset or refined over time. Any tokens launched as SMTs shall benefit from a
blockchain ecosystem built with an inbuilt decentralized exchange, as well as an ecosystem
of open-source applications and libraries to support successful deployment, fundraising,
and growth.
Leveraging Tokens for Autonomous User Growth
SMTs are a breakthrough for bridging the world’s content applications to tokens in a way
that aligns incentives between the users of a network and the entrepreneurs building the
applications. By leveraging the concepts of inflation (new token emissions) and token allocations by post-based voting, SMTs exist in a manner where value must be distributed
to users who are participating in their related content networks and applications. Entrepreneurs may now create tokens to integrate with their blog, application, or an entire
network of applications and topics. With SMTs, the entrepreneurs have the flexibility
to decide on the economics of the tokens they integrate into their products, from the
inflation rates to the algorithms that distribute the tokens.
Two unique properties align incentives and make SMTs “smart and social” compared to
other tokens (such as bitcoin, ether and ERC-20s). The first is a pool of tokens dedicated
to incentivizing content creation and curation (called the “rewards pool”). The second
is a voting system that leverages the wisdom of the crowd to assess the value of content
and distribute tokens to it. These two unique properties when combined are referred
to as Proof-of-Brain, which is an entendre based on Proof-of-Work, meant to emphasize
the human work required to distribute tokens to community participants. Proof-of-Brain
positions SMTs as a tool for building perpetually growing communities, which encourage
their members to add value to the community through the built in rewards structure.
Entrepreneurs and established entities may rely on SMTs to grow their content network
because of the automated and continuous generation of new tokens that are allocated
to producers of content by the holders of the existing tokens, through the process of
competitive voting. As the tokens are distributed to users of the network, the interests of
5
existing token holders are further aligned with content creators, the businesses running
the applications, and the entrepreneurs that support them. These unique properties of
the tokens’ economics continue to provide incentives for new users to join and participate
in growing the network. Any application, whether it is an existing publisher behemoth or
a stealth-mode social media startup, will be able to integrate and leverage these special
tokens for their own growth.
New Fundraising Opportunities
Blockchain-based tokens, beginning strongly with the advent of ERC20 on Ethereum,
represent a new manner of bringing capital into an organization through the process of
Initial Coin Offerings (ICOs). ICOs are an opportunity for one group to sell an initial
supply of tokens, privately or publicly, for-specific-purpose, for-profit or not-for-profit.
Depending on how these tokens are sold, different regulatory bodies could see them as
commodities, securities, derivatives, or as none of the above. Regardless, it is clear we
have seen north of one billion dollars (USD) raised through ICOs in 2017, and to support
this trend, it is possible to conveniently launch and sell tokens via the built in ICO contract
of SMTs. The launch of SMTs can be structured for ICOs with hard, soft, and no caps,
and can be tailored to receive STEEM and cryptocurrencies on other blockchains.
Immediate Liquidity
By leveraging a recently designed automated market maker concept [2], SMT-based ICOs
allow a portion of STEEM tokens received to be sent into an SMT’s on-chain, off-orderbook market maker in order to provide liquidity to the SMT at a specified reserve ratio.
Beyond the social and specialized distribution mechanisms of SMTs, this feature advances
the concept of automated market makers by pairing it alongside SMT’s decentralized
markets, which also facilitate bids and asks by human participants. The combination
of these two markets enables on-chain and trustless exchange opportunities for market
makers while enabling liquidity for token users.
Shared Bootstrap Tools
SMTs may be created with reward pool parameters tuned for “Shared Influence” between
Steem Power and other vesting SMTs, which means a SMT creator may specify that
Steem Power can control a portion of the SMT’s rewards pool for an unlimited or limited
amount of time, with increasing or decreasing influence. Altogether, Shared Influence may
allow SMTs to be wholly or partially bootstrapped by the interest of existing and active
Steem or other SMT community members. Through these tools, community managers
and entrepreneurs launching a token may leverage existing user bases to accelerate the
distribution of the SMT to a target market.
Monetizing with Shared Token Rewards
All Steem based interfaces have the option of splitting token rewards among a set of
arbitrary recipients, which could include an interface, community manager, referrer, a
paid position donation pool, and more. An interface can also provide this optionality
6
of how to split the tokens to the authors. The number of potential Reward Sharing
beneficiaries is initially soft capped by block producers at eight while the feature proves
its use, however the blockchain is capable of handling up to 256 beneficiaries per post.
Can My Entity Participate in SMTs?
An SMT can be launched by a person or entity; they only need 1 USD to cover the network
fee (this fee prevents spam and unused tokens while accruing value to the network), and
a namespace on Steem - which can be obtained by registering at anon.steem.network,
steemit.com, steemconnect.com, or any other Steem sign-up service.
Once an account name to register the token with is secured, the account issues the token
by using a Steem-based command line tool or any tool created in the future to support
token launches. The token can be structured to support an initial sale or distribution of
the token. Certain properties of an SMT, such as its inflation rate, must also be defined
by the person or entity creating the token. These properties dictate how the token is used
inside applications and respective communities.
From launch, the token becomes immutable on the blockchain, and leveraged correctly,
the token can have dramatic effects on the growth of businesses that choose to integrate
these tokens.
Use Cases
We have identified five ways in which existing businesses and future entrepreneurs can
leverage specially designed SMTs to transform the internet. Among these use cases you
may discover other ways of structuring and leveraging tokens inside applications. This list
is by no means exhaustive, and we will update this paper as more use cases demonstrate
their value.
1 - Content Publishers - Single Token Support
A mainstream media website’s growth has been slowing and they are looking for ways
to get ahead of the changing tech landscape. The website migrates to a Disqus-like
application based on Steem, or taps directly into Steem APIs for a custom integration.
Now their subscribers can be rewarded with cryptocurrency while commenting. When
the website is ready, they can issue their own token through the comments interface - the
token will allow them to 1) raise capital by selling tokens 2) catalyze autonomous growth.
7
Figure 1: Single Token Content Publishers
2 - Forums - Multiple Token Support
An up-and-coming forum business is looking to integrate cryptocurrency to create cash
flow and spark growth to get the business to the next level, however they are not cryptocurrency security experts and would prefer not to host a cryptocurrency wallet. They
issue an SMT and integrate it into their website. Focusing solely on the social aspects, the
forum business can integrate other applications, such as SteemConnect into their forum
to handle the wallet and transfer capabilities. This allows them to focus on their business
(growing communities) without focusing on the security aspects of cryptocurrency. The
forum enables additional tokens to be exposed or launched, to represent specific topics
of discussion. The ability to launch these tokens can be retained by the company behind
the website, or granted to the website’s community managers. Tokens dedicated to the
website’s specific topics will further spur autonomous growth of the website niche by niche.
An example of this multi-token model could eventually be found in organizations such as
ChainBB (chainbb.com) if it were to enable its own globally available token on its domain,
8
as well as narrowly available tokens for specific community niches - such as “gardening.”
Figure 2: Multiple tokens Forum
3 - Comments Widget for Online Publishers
One of the ways in which publishers will be onboarded faster to SMT integrations is by
offering a Steem-based comments widget that can easily be integrated into existing blogs
that are built on software such as WordPress and Blogger. The developer employing
the widget would be able to take a percentage of the tokens (called “Shared Rewards”)
distributed to the commenters for themselves, thereby creating a business opportunity for
the next generation of Disqus-like companies that are cryptocurrency enabled. It would
alleviate the burdens of transaction signing support, private key management, wallet
functionality, and hosting costs for the publisher - by outsourcing all of these functions
to the comments widget maintainer.
9
Figure 3: Comment Widget
4 - Sub-Community Moderators and Managers
Imagine you are a moderator for a specific topic inside a forum, such as a Reddit “subreddit” or a Steemit “community”. If a website integrates SMTs for these specific topics,
then the topic moderator/s can launch these tokens to empower the subscribers of their
topic, raise funds, and increase the quality of content curation for the community.
10
Figure 4: Sub-community
5 - Arbitrary Assets - Tokens Representing Real World Assets
Let’s examine an instance in which an entrepreneur is looking to provide liquidity in the
Steem ecosystem. The entrepreneur can issue an SMT without inflation properties, and
imply that they will provide structure to peg it to USD (or any other debt, contract, or
asset), making it like an IOU or basic derivative. The structure they provide to the asset
includes buying and selling it near $1, similar to Tether. The entrepreneur sets up bank
wire capabilities for buying and selling, and takes a small % on each transaction. The
derivative trades against STEEM, and also brings capital into the ecosystem to be used
across all tokens.
11
Figure 5: IOU Asset Token Exchange
Owner’s manual
This manual will explain the nuts and bolts of how SMTs work. The intended audience
is technical users who want to create their own SMT.
Create a control account
The first step to creating an SMT is to create a control account for the SMT. Any STEEM
account may serve as a control account, however it is highly recommended to create a
dedicated account solely for the purpose. It is also highly recommended that a control
account does not post, vote, or hold any STEEM, SBD, or other tokens (other than a
small amount of vested STEEM for transaction bandwidth).
12
The control account’s name will not occupy a high visibility position in most user interfaces, so it does not much matter if the control account’s name is not the best match for
the SMT brand.
Control account security
Security on the control account is important for persons who plan to use the account post
launch:
• The control account should use 2-of-3 or 3-of-5 multi-signature security.
• The control account’s authorities should have other accounts, not specific keys, as
multi-signature members.
• For additional security, each of the accounts in the control account’s multi-signature
group should itself use multi-signature security.
• A subset of keys should be kept offline, in air-gapped machines.
• Transactions should be generated by an online interface, and physically transferred
to the air-gapped machines via removable media.
• Signatures should be returned via physically removable media to the online system
for transmission via the UI.
Of course, once authorities are set up, you should verify the account is still able to transact.
It is advisable to test your authorities and transaction signing setup using a testnet, or a
less-important account on the main network.
Once the token is launched, you may consider burning the account’s keys by assigning
them to @null, initiating a token for which the dynamic properties can never be adjusted.
Token consensus
Since tokens participate in atomic transactions also involving STEEM, they have been
designed as part of the STEEM blockchain’s consensus.
Token Generation and Initialized Parameters
SMT object creation
The first operation to be executed is an smt_create_operation. This operation creates
an SMT object in the blockchain state. After executing the smt_create_operation, the
newly created SMT object is not yet fully configured.
Most of the configuration occurs in subsequent operations (smt_set_setup_parameters_operation,
smt_setup_inflation_operation and smt_setup_operation). These later operations
may occur in the same transaction, but they may also occur at any later point in time.
struct smt_create_operation
{
account_name_type control_account;
asset
smt_creation_fee;
asset_symbol_type symbol;
extensions_type
extensions;
};
13
Numerical asset identifiers
An SMT is referred to by a numerical asset identifier or NAI, consisting of two at-signs
followed by nine decimal digits, for example @@314159265. The blockchain enforces that
the identifier placed by a UI into the smt_create_operation must match the result of
the get_next_smt_identifier RPC. Therefore, an NAI cannot be chosen freely by the
SMT creator. It is not even possible to “mine” a “vanity NAI” (analogous to the “vanity
Bitcoin address” some people use).
The reason for this restriction is that the blockchain designers want to discourage users
from using the consensus level identifiers as symbol names, and instead use a nonconsensus directory system to attach human meaningful symbols to assets. Distinguishing
a “namesquatter” from the legitimate owner of a brand is not something that a blockchain
can do, especially if the squatter is willing to pay the SMT creation fee.
SMT naming
The solution to the namesquatting problem is to publish an asset directory mapping NAIs
to names. An asset directory is non-consensus, meaning that all blockchain operations
are serialized only with NAIs. Asset names are only used for UI presentation.
A UI may include an asset directory as a file, URL, or a blockchain account which publishes
directory entries with custom operations. The publisher of an asset directory should
ensure that directory entries meet whatever standards of legitimate brand ownership the
publisher chooses to enforce.
SMT creation fee
Issuing a smt_create_operation requires payment of smt_creation_fee. The amount
required is set by the smt_creation_fee field of dynamic_global_properties_object.
This field may contain a value in STEEM or SBD. If specified in SBD, an equivalent
amount of STEEM will be accepted, at the current price feed.
Initially, smt_creation_fee will be set to 1 SBD, and no means will be provided
to update it. Updates to the smt_creation_fee amount may occur in future hardforks, however, so user-agents should read the smt_creation_fee value from the
dynamic_global_properties_object. User-agents should not assume the fee will
always be 1 SBD and they should be prepared to charge a separate fee paid to the
user-agent if the aim of the interface is to enable only a curated set of tokens.
The fee is destroyed by sending it to STEEM_NULL_ACCOUNT.
SMT pre-setup
Two pre-setup operations are included:
smt_setup_inflation_operation and
smt_setup_parameters. These operations must be issued after smt_create_operation,
and before smt_setup_operation. They may be issued in the same transaction, or in
prior blocks.
The reason pre-setup operations are not made a part of smt_setup_operation is to allow
a large number of pre-setup operations to be executed over multiple blocks.
14
SMT setup
Each SMT has an associated descriptor object which has permanent configuration
data. This data cannot be changed after launch! The descriptor is set by the
smt_setup_operation:
struct smt_setup_operation
{
account_name_type
control_account;
asset_symbol_type
smt_name;
int64_t
max_supply = STEEM_MAX_SHARE_SUPPLY;
smt_generation_policy
initial_generation_policy;
time_point_sec
time_point_sec
time_point_sec
time_point_sec
generation_begin_time;
generation_end_time;
announced_launch_time;
launch_expiration_time;
extensions_type
extensions;
};
The symbol precision in smt_setup_operation is authoritative. It may differ from, and
will override, any previously specified operations’ precision. Subsequently issued operations must have matching precision.
The operation must be signed by the control_account key. The named SMT must have
been created earlier by the control_account. The symbol’s embedded decimal places
may be distinct from prior smt_setup_operation.
The decimal_places field is used by UIs to display units as a number of decimals.
The generation_begin_time is when participants can begin to contribute to the ICO.
It is allowed to be in the future so users have time to study the ICO’s final terms before
the ICO begins.
The generation_end_time is when the ICO stops accepting contributions, and
the announced_launch_time is when the ICO token is created (assuming the ICO
reached the minimum participation level). Some pause is allocated between the
generation_end_time and announced_launch_time to allow for the possibility of
ICOs that wish to have hidden caps that aren’t revealed while the ICO is open for
contributions. It also gives the ICO creator time to use the final ICO numbers to aid in
pre-launch business activities.
At launch_expiration_time, if the ICO has not yet launched, all contributors will be
automatically refunded (with virtual operations) and the ICO will be cancelled. The
symbol will remain reserved to the specified control_account. However, in order to
launch the token, an smt_create_operation must be issued and the smt_creation_fee
must be paid again.
15
Token units
Initial token generation is driven by a contributions of STEEM units from contributors.
To simplify rounding concerns, a contribution must be an integer number of STEEM
units. The ICO creator sets the size of a STEEM unit - it can be large or small. It is
better to keep the unit small (for example, 1 STEEM or 0.1 STEEM), as this allows the
ICO to be accessible to the maximum possible audience.
A STEEM unit also specifies a routing policy which determines where the STEEM goes
when the token launches. (STEEM for tokens which do not launch may be refunded on
demand.) The routing policy may split the STEEM in the unit among multiple parties.
When the ICO occurs, the tokens are generated in token units. Multiple token units are
generated per STEEM unit contributed. Token units also have a routing policy.
The units and their routing policies are specified in the smt_generation_unit structure:
struct smt_generation_unit
{
flat_map< account_name_type, uint16_t >
flat_map< account_name_type, uint16_t >
};
steem_unit;
token_unit;
Each (key, value) pair in the flat_map determines the routing of some satoshis. The
total STEEM/tokens in each unit is simply the sum of the values.
Unit ratios
When an SMT launches, token units are created for STEEM units in a R-for-1 ratio. The number R is called the unit ratio. Maximum and minimum allowable values
for R are specified respectively in the min_unit_ratio and max_unit_ratio fields of
smt_generation_policy.
The maximum number of token units that can be created in the ICO is limited to
max_token_units_generated, a parameter which is set by the ICO creator. (More tokens can be created after the token has launched, but this later creation is called inflation
and is not considered to be part of the ICO.)
The unit ratio is set to the largest integer that would not result in exceeding
max_token_units_generated for the number of STEEM units actually contributed.
Cap and min
ICOs may specify a minimum number of STEEM units min_steem_units. If the ICO
does not reach min_steem_units before generation_end_time, then it does not occur,
and contributors become eligible for refunds.
Likewise, ICOs may specify two maximum numbers of STEEM units: A hard cap and
a soft cap. Units in excess of the soft cap have different routing for their STEEM and
tokens. STEEM units in excess of the hard cap are rejected and do not generate any
SMTs.
16
The effects of the soft cap are divided proportionally among all contributors. I.e. if a
ICO has a soft cap of 8 million STEEM, and 10 contributors each contribute 1 million
STEEM, then 0.2 million of each user’s STEEM is routed via the soft cap’s policy.
The effects of the hard cap fall solely on the last contributors. I.e. if a ICO has a hard
cap of 8 million STEEM, and 10 contributors each contribute 1 million STEEM, then
the first 8 users fully participate in the ICO, and the last 2 users are refunded 1 million
STEEM.
Hidden caps
The min and hard caps are hidden in the generation policy. This means that these
numbers are fixed at setup time, but the ICO creator has the option to keep them secret.
This functionality is implemented by a commit/reveal cryptographic protocol: A hash
called the commitment is published at setup time, and the actual amount must match
the commitment. (A nonce is also included in the hash to prevent an attacker from finding
the hidden cap with a brute-force guess-and-test approach.)
The SMT designer may wish to pre-publish a guarantee that the hidden values are within
a certain range. The lower_bound and upper_bound fields provide this functionality: A
revealed amount that is not in the specified range is treated the same as a hash mismatch.
struct smt_cap_commitment
{
share_type
lower_bound;
share_type
upper_bound;
digest_type
hash;
};
struct smt_revealed_cap
{
share_type
uint128_t
};
amount;
nonce;
struct smt_cap_reveal_operation
{
account_name_type
control_account;
smt_revealed_cap
cap;
extensions_type
extensions;
};
All caps are hidden, but the cap may be revealed at any point in time. Therefore, an
ICO with a non-hidden minimum or cap may be implemented by simply including the
smt_cap_reveal_operation in the same transaction as the smt_setup_operation. UIs
should provide functionality for this.
A UI should provide one or more of the following means to ensure the nonce and amount
are recoverable:
• Force the user to type in the amount and nonce again, as confirmation they have
been backed up.
17
• Set nonce to some deterministic function of the private key and public data, for example nonce = H(privkey + control_account + lower_bound + upper_bound
+ current_date).
• Provide functionality to brute-force the uncertain fields when the nonce is known
(e.g. the current date and amount).
• Require the amount to be low-entropy to facilitate brute-forcing when the nonce is
known (e.g. a number between 1-999 times a power of 10).
Generation policy data structure
The SMT generation policy data structure looks like this:
struct smt_capped_generation_policy
{
smt_generation_unit pre_soft_cap_unit;
smt_generation_unit post_soft_cap_unit;
smt_cap_commitment
smt_cap_commitment
min_steem_units_commitment;
hard_cap_steem_units_commitment;
uint16_t
soft_cap_percent = 0;
uint32_t
uint32_t
min_unit_ratio = 0;
max_unit_ratio = 0;
extensions_type
extensions;
};
Note, the max_token_units_generated parameter does not appear anywhere in the operation. The reason is that it is actually a derived parameter: max_token_units_generated
= min_unit_ratio * hard_cap_steem_units.
Additionally, the smt_generation_policy is defined as a static_variant, of which
smt_capped_generation_policy is the only member:
typedef static_variant< smt_capped_generation_policy > smt_generation_policy;
This typedef allows the potential for future protocol versions to allow additional generation policy semantics with different parameters.
Examples and rationale
Example ICO
ALPHA wants to sell a token to the crowd to raise funds where: 70% of contributed
STEEM goes to the Alpha Organization Account (@alpha_org), 23% of contributed
STEEM goes to Founder Account A (@founder_a), and 7% of contributed STEEM goes
to Founder Account B (@founder_b).
ALPHA defines a STEEM unit as:
steem_unit = [["alpha_org", 70], ["founder_a", 23], ["founder_b", 7]]
18
This STEEM-unit contains 100 STEEM-satoshis, or 0.1 STEEM.
For every 1 STEEM contributed, an ALPHA contributer will receive 5 ALPHA tokens,
and Founder Account D will receive 1 ALPHA token. This five-sixths / one-sixth split is
expressed as:
token_unit = [["$from", 5], ["founder_c", 1]]
This ratio is defined in the following data structure:
struct smt_generation_unit
{
flat_map< account_name_type, uint16_t >
flat_map< account_name_type, uint16_t >
};
steem_unit;
token_unit;
This token-unit contains 6 ALPHA-satoshis, or 0.0006 ALPHA (if ALPHA has 4 decimal
places).
Next we define the unit ratio as the relative rate at which token_unit are issued as
steem_unit are contributed. So to match the specification of 6 ALPHA per 1 STEEM,
we need to issue 1000 ALPHA-units per STEEM-unit. Therefore the unit ratio of this
ICO is 1000. This unit ratio is placed in the min_unit_ratio and max_unit_ratio fields
of the smt_capped_generation_policy data structure:
min_unit_ratio = 1000
max_unit_ratio = 1000
A special account name, $from, represents the contributor.
Also supported is
$from.vesting, which represents the vesting balance of the $from account.
Why unit ratios?
Why does the blockchain use unit ratios, rather than simply specifying prices?
The answer is that it is possible to write ICO definitions for which price is ill-defined. For
example:
•
•
•
•
"$from" does not occur in token_unit.
"$from" occurs in both token_unit and steem_unit.
A combination of "$from" and "$from.vesting" occurs.
Future expansion allows new special accounts.
All of these ICO definitions have a unit ratio, but defining a single quantity to call “price”
is complicated or impossible for ICOs like these.
UI treatment of unit ratios
As a consequence of the above, the concept of “ICO price” is purely a UI-level concept.
UIs which provide an ICO price should do the following:
• Document the precise definition of “price” provided by the UI.
• Be well-behaved for pathological input like above.
• Have a button for switching between a unit ratio display and price display.
19
Hidden cap FAQ
• Q: Should my ICO have a cap?
• A: Some set of people stay away from uncapped ICOs due to perceived “greed”, or
want a guaranteed lower bound on the percentage of the ICO their contribution will
buy. If you want this set of people to participate, use a cap.
• Q: Should my cap be hidden?
• A: Some people like the transparency and certainty of a public cap. Other people
think a hidden cap creates excitement and builds demand. One possible compromise
is to publish the previous and next power of 10, for example “this ICO’s cap is
between 1 million and 10 million STEEM.”
• Q: How do I disable the cap?
• A: Set it so that the cap would occur above STEEM_MAX_SHARE_SUPPLY.
Launch
The effective launch time is the time at which tokens become transferable. Two possibilities occur based on the timing of revealing of the hard cap:
• When min_steem_units and hard_cap_steem_units are revealed before the
announced_launch_time, the launch is an on-time launch. The launch logic is
executed by the blockchain as soon as announced_launch_time arrives, regardless
of further user action.
• When min_steem_units and hard_cap_steem_units have not been revealed before
the announced_launch_time, the launch will be a delayed launch. The launch logic
is executed by the blockchain when min_steem_units and hard_cap_steem_units
have been revealed.
• If the launch is delayed, then any contributor may use smt_refund_operation to
get their STEEM back at any time after announced_launch_time, and before the
launch logic is executed.
The reasons for this design are as follows:
• The hidden cap isn’t published immediately (that’s the definition of hidden).
• Publishing the hidden cap is an action that must be done by the ICO creator (again,
any action requiring non-public information to occur cannot happen automatically
on a blockchain).
• If the ICO creator never acts, then the launch logic will never execute.
• In the case of such a malicious or unresponsive ICO creator, contributors’ STEEM
would effectively be trapped forever, and they would never receive any tokens.
• To keep the STEEM from being trapped in this way, the smt_refund_operation
is implemented.
struct smt_refund_operation
{
account_name_type
contributor;
asset
amount;
extensions_type
extensions;
};
20
Note, users are not required to use smt_refund_operation; each individual contributor
must opt-in to receiving a refund. If the ICO creator publicizes a legitimate reason they
failed to publish before announced_launch_time, it is possible that all/most contributors
will voluntarily choose not to use smt_refund_operation. In this case, the launch will
occur as soon as the ICO creator publishes the hidden values.
The launch logic considers a contribution followed by a refund to be equivalent to not
having contributed at all. Therefore, when a delayed launch occurs, each contributor will
be in exactly one of the following two states:
• The contributor has executed smt_refund_operation, received their STEEM back,
and will not participate in the ICO.
• The contributor has not been issued a refund, and will participate in the ICO.
It is possible for a delayed launch to have exceeded its min_steem_units value at the
announced launch time, but subsequently falls below its min_steem_units value as a
result of refunds. In such a case, the ICO will not occur; it will be treated as if it had
never reached its min_steem_units.
Full JSON examples
ALPHA
This example builds on the ALPHA example from earlier. This ICO has the following
characteristics:
•
•
•
•
•
•
•
•
70% of contributed STEEM goes to Alpha Organization Account (@alpha_org).
23% of contributed STEEM goes to Founder Account A (@founder_a).
7% of contributed STEEM goes to Founder Account B (@founder_b).
Minimum unit of contribution is 0.1 STEEM.
For every 1 STEEM contributed, the contributor gets 5 ALPHA (@contibutor_a).
For every 1 STEEM contributed, Founder Account C gets 1 ALPHA (@founder_c).
No minimum, hard cap, or soft cap.
No post-launch inflation after launch.
21
Figure 6: Alpha ICO Flow
These are the operations for the ALPHA launch:
[
["smt_setup",
{
"control_account" : "alpha",
"decimal_places" : 4,
"max_supply" : "1000000000000000",
"initial_generation_policy" : [0,
{
"pre_soft_cap_unit" : {
"steem_unit" : [["alpha_org", 70], ["founder_a", 23], ["founder_b", 7]],
"token_unit" : [["$from", 5], ["founder_c", 1]]
},
"post_soft_cap_unit" : {
"steem_unit" : [],
"token_unit" : []
},
"min_steem_units_commitment" : {
"lower_bound" : 1,
"upper_bound" : 1,
"hash" : "32edb6022c0921d99aa347e9cda5dc2db413f5574eebaaa8592234308ffebd2b"
},
22
"hard_cap_steem_units_commitment" : {
"lower_bound" : "166666666666",
"upper_bound" : "166666666666",
"hash" : "93c5a6b892de788c5b54b63b91c4b692e36099b05d3af0d16d01c854723dda21"
},
"soft_cap_percent" : 10000,
"min_unit_ratio" : 1000,
"max_unit_ratio" : 1000,
"extensions" : []
}
],
"generation_begin_time" : "2017-08-10T00:00:00",
"generation_end_time" : "2017-08-17T00:00:00",
"announced_launch_time" : "2017-08-21T00:00:00",
"smt_creation_fee" : "1000.000 SBD",
"extensions" : []
}
],
["smt_cap_reveal",
{
"control_account" : "alpha",
"cap" : { "amount" : 1, "nonce" : "0" },
"extensions" : []
}
],
["smt_cap_reveal",
{
"control_account" : "alpha",
"cap" : { "amount" : "166666666666", "nonce" : "0" },
"extensions" : []
}
]
]
Some things to note:
• We disable the soft cap by setting soft_cap_percent to STEEM_100_PERCENT =
10000.
• post_soft_cap_unit must be empty when the soft cap is disabled.
• The unit ratio does not change so min_unit_ratio / max_unit_ratio must be set
accordingly.
• We disable the hidden caps by using a zero nonce and setting lower_bound
==
upper_bound.
• We still need to reveal the caps with smt_cap_reveal_operation.
• The hard cap specified is the largest hard cap that does not result in created tokens
exceeding STEEM_MAX_SHARE_SUPPLY.
BETA
The BETA token is created with the following rules:
• For every 5 STEEM contributed, 3 STEEM go to founder account Fred.
23
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
For every 5 STEEM contributed, 2 STEEM go to founder account George.
10% of the initial token supply goes to founder account George.
20% of the initial token supply goes to founder acconut Henry.
70% of the initial token supply is divided among contributors according to their
contribution.
Each STEEM unit is 0.005 STEEM.
Each token unit is 0.0010 BETA.
The minimum raised is 5 million STEEM units, or 25,000 STEEM.
The maximum raised is 30 million STEEM units, or 150,000 STEEM.
Each contributor receives 7-14 BETA per STEEM contributed, depending on total
contributions.
George receives 1-2 BETA per STEEM contributed, depending on total contributions.
Harry receives 2-4 BETA per STEEM contributed, depending on total contributions.
If the maximum of 30 million STEEM units are raised, then min_unit_ratio
=
50 applies.
The maximum number of token units is min_unit_ratio times 30 million, or 1.5
billion token units.
Since each token unit is 0.0010 BETA, at most 1.5 million BETA tokens will be
generated.
If 75,000 STEEM or less is contributed, the contributors George and Harry will
receive the maximum of 14, 2, and 4 BETA per STEEM contributed (respectively).
If more than 75,000 STEEM is contributed, the contributors, George and Harry
will receive BETA in a 70% / 10% / 20% ratio, such that the total is fixed at 1.5
million BETA.
As a consequence of the hard cap, the contributors, George and Harry will receive
at least 7, 1, and 2 BETA per STEEM contributed (respectively).
This example is chosen to demonstrate how the ratios work. It is not a realistic example,
as most ICOs will choose to either set min_unit_ratio = max_unit_ratio like ALPHA,
or choose to use a large max_unit_ratio like BETA.
[
[
"smt_setup",
{
"control_account" : "beta",
"decimal_places" : 4,
"max_supply" : "1000000000000000",
"initial_generation_policy" : [0,
{
"pre_soft_cap_unit" : {
"steem_unit" : [["fred", 3], ["george", 2]],
"token_unit" : [["$from", 7], ["george", 1], ["henry", 2]]
},
"post_soft_cap_unit" : {
"steem_unit" : [],
"token_unit" : []
},
"min_steem_units_commitment" : {
"lower_bound" : 5000000,
"upper_bound" : 5000000,
24
"hash" : "dff2e4aed5cd054439e045e1216722aa8c4758b22df0a4b0251d6f16d58e0f3b"
},
"hard_cap_steem_units_commitment" : {
"lower_bound" : 30000000,
"upper_bound" : 30000000,
"hash" : "f8e6ab0e8f2c06a9d94881fdf370f0849b4c7864f62242040c88ac82ce5e40d6"
},
"soft_cap_percent" : 10000,
"min_unit_ratio" : 50,
"max_unit_ratio" : 100,
"extensions" : []
}
],
"generation_begin_time" : "2017-06-01T00:00:00",
"generation_end_time" : "2017-06-30T00:00:00",
"announced_launch_time" : "2017-07-01T00:00:00",
"smt_creation_fee" : "1000.000 SBD",
"extensions" : []
}
],
[
"smt_cap_reveal",
{
"control_account" : "beta",
"cap" : { "amount" : 5000000, "nonce" : "0" },
"extensions" : []
}
],
[
"smt_cap_reveal",
{
"control_account" : "beta",
"cap" : { "amount" : 30000000, "nonce" : "0" },
"extensions" : []
}
]
]
This spreadsheet will make the relationship clear.
GAMMA
The GAMMA token is like BETA, but with one difference: The large max_unit_ratio
means that the maximum issue of 1.5 million tokens is reached very early in the ICO.
This ICO effectively divides 1.5 million GAMMA tokens between contributors (provided
at least 5 STEEM is contributed).
[
[
"smt_setup",
{
25
"control_account" : "gamma",
"decimal_places" : 4,
"max_supply" : "1000000000000000",
"initial_generation_policy" : [0,
{
"pre_soft_cap_unit" : {
"steem_unit" : [["fred", 3], ["george", 2]],
"token_unit" : [["$from", 7], ["george", 1], ["henry", 2]]
},
"post_soft_cap_unit" : {
"steem_unit" : [],
"token_unit" : []
},
"min_steem_units_commitment" : {
"lower_bound" : 5000000,
"upper_bound" : 5000000,
"hash" : "dff2e4aed5cd054439e045e1216722aa8c4758b22df0a4b0251d6f16d58e0f3b"
},
"hard_cap_steem_units_commitment" : {
"lower_bound" : 30000000,
"upper_bound" : 30000000,
"hash" : "f8e6ab0e8f2c06a9d94881fdf370f0849b4c7864f62242040c88ac82ce5e40d6"
},
"soft_cap_percent" : 10000,
"min_unit_ratio" : 50,
"max_unit_ratio" : 300000,
"extensions" : []
}
],
"generation_begin_time" : "2017-06-01T00:00:00",
"generation_end_time" : "2017-06-30T00:00:00",
"announced_launch_time" : "2017-07-01T00:00:00",
"smt_creation_fee" : "1000.000 SBD",
"extensions" : []
}
],
[
"smt_cap_reveal",
{
"control_account" : "gamma",
"cap" : { "amount" : 5000000, "nonce" : "0" },
"extensions" : []
}
],
[
"smt_cap_reveal",
{
"control_account" : "gamma",
"cap" : { "amount" : 30000000, "nonce" : "0" },
"extensions" : []
}
26
]
]
DELTA
In this ICO we have one million DELTA tokens created for the founder, and none for
contributors. A modest contribution of 0.1 STEEM can be made by any user (including
the founder themselves) to trigger the generation.
[
[
"smt_setup",
{
"control_account" : "delta",
"decimal_places" : 5,
"max_supply" : "1000000000000000",
"initial_generation_policy" : [0,
{
"pre_soft_cap_unit" : {
"steem_unit" : [["founder", 1]],
"token_unit" : [["founder", 10000]]
},
"post_soft_cap_unit" : {
"steem_unit" : [],
"token_unit" : []
},
"min_steem_units_commitment" : {
"lower_bound" : 10000000,
"upper_bound" : 10000000,
"hash" : "4e12522945b8cc2d87d54debd9563a1bb6461f1b1fa1c31876afe3514e9a1511"
},
"hard_cap_steem_units_commitment" : {
"lower_bound" : 10000000,
"upper_bound" : 10000000,
"hash" : "4e12522945b8cc2d87d54debd9563a1bb6461f1b1fa1c31876afe3514e9a1511"
},
"soft_cap_percent" : 10000,
"min_unit_ratio" : 1000,
"max_unit_ratio" : 1000,
"extensions" : []
}
],
"generation_begin_time" : "2017-06-01T00:00:00",
"generation_end_time" : "2017-06-30T00:00:00",
"announced_launch_time" : "2017-07-01T00:00:00",
"smt_creation_fee" : "1000.000 SBD",
"extensions" : []
}
],
[
"smt_cap_reveal",
27
{
"control_account" : "delta",
"cap" : { "amount" : 10000000, "nonce" : "0" },
"extensions" : []
}
],
[
"smt_cap_reveal",
{
"control_account" : "delta",
"cap" : { "amount" : 10000000, "nonce" : "0" },
"extensions" : []
}
]
]
Vesting contributions
It is possible to send part or all of contributions to a vesting balance, instead of permitting
immediate liquidity. This example puts 95% in vesting.
"token_unit"
: [["$from.vesting", 95], ["$from", 5]]
Burning contributed STEEM
In this ICO, the STEEM is permanently destroyed rather than going into the wallet of
any person. This mimics the structure of the Counterparty ICO.
{
"steem_unit" : [["null", 1]],
"token_unit" : [["$from", 1]]
}
Vesting as cost
In this ICO, you don’t send STEEM to the issuer in exchange for tokens. Instead, you
vest STEEM (to yourself), and tokens are issued to you equal to the STEEM you vested.
{
"steem_unit" : [["$from.vesting", 1]],
"token_unit" : [["$from", 1]]
}
Non-STEEM & Hybrid ICO’s
ICOs using non-STEEM contributions – for example, SBD, BTC, ETH, etc. – cannot
be done fully automatically on-chain. However, such ICOs can be managed by manually
transferring some founder account’s distribution to buyers’ Steem accounts in proportion
to their non-STEEM contribution.
28
Inflation Parameters
Creation of SMT after launch is called inflation.
Inflation is the means by which the SMT rewards contributors for the value they provide.
Inflation events use the following data structure:
struct smt_inflation_unit
{
flat_map< account_name_type, uint16_t >
};
token_unit;
// Event: Support issuing tokens to target at time
struct token_inflation_event
{
timestamp
schedule_time;
smt_inflation_unit unit;
uint32_t
num_units;
};
This event prints num_units units of the SMT token.
Possible inflation target
The target is the entity to which the inflation is directed. The target may be a normal
Steem account controlled by an individual founder, or a multi-signature secured account
comprised of several founders.
In addition, several special targets are possible representing trustless functions provided
by the blockchain itself:
• Rewards. A special destination representing the token’s posting / voting rewards.
• Vesting. A special destination representing the tokens backing vested tokens.
Event sequences
Traditionally blockchains compute inflation on a per-block basis, as block production
rewards are the main (often, only) means of inflation.
However, there is no good reason to couple inflation to block production for SMTs. In fact,
SMTs have no block rewards, since they have no blocks (the underlying functionality of
block production being supplied by the Steem witnesses, who are rewarded with STEEM).
Repeating inflation at regular intervals can be enabled by adding interval_seconds and
interval_count to the token_inflation_event data structure. The result is a new
data structure called token_inflation_event_seq_v1:
// Event seq v1: Support repeatedly issuing tokens to target at time
struct token_inflation_event_seq_v1
{
timestamp
schedule_time;
smt_inflation_unit unit;
asset
new_smt;
29
int32_t
uint32_t
interval_seconds;
interval_count;
};
The data structure represents a token inflation event that repeats every interval_seconds
seconds, for interval_count times. The maximum integer value 0xFFFFFFFF is a special
sentinel value that represents an event sequence that repeats forever.
Note, the new_smt is a quantity of SMT, not a number of units. The number of units is
determined by dividing new_smt by the sum of unit members.
Adding relative inflation
Often, inflation schedules are expressed using percentage of supply, rather than in absolute
terms:
// Event seq v2: v1 + allow relative amount of tokens
struct token_inflation_event_seq_v2
{
timestamp
schedule_time;
smt_inflation_unit unit;
uint32_t
num_units;
int32_t
uint32_t
interval_seconds;
interval_count;
asset
uint32_t
abs_amount;
rel_amount_numerator;
};
Then we compute new_smt as follows from the supply:
rel_amount = (smt_supply * rel_amount_numerator) / SMT_REL_AMOUNT_DENOMINATOR;
new_smt = max( abs_amount, rel_amount );
If we set SMT_REL_AMOUNT_DENOMINATOR to a power of two, the division can be optimized
to a bit-shift operation. To gain a more dynamic range from the bits, we can let the shift
be variable:
// Event seq v3: v2 + specify shift in struct
struct token_inflation_event_seq_v3
{
timestamp
schedule_time;
smt_inflation_unit unit;
int32_t
uint32_t
interval_seconds;
interval_count;
asset
uint32_t
uint8_t
abs_amount;
rel_amount_numerator;
rel_amount_denom_bits;
};
Then the computation becomes:
30
rel_amount = (smt_supply * rel_amount_numerator) >> rel_amount_denom_bits;
new_smt = max( abs_amount, rel_amount );
Of course, the implementation of these computations must carefully handle potential
overflow in the intermediate value smt_supply * rel_amount_numerator!
Adding time modulation
Time modulation allows implementing an inflation rate which changes continuously over
time according to a piecewise linear function. This can be achieved by simply specifying the left/right endpoints of a time interval, and specifying absolute amounts at both
endpoints:
// Event seq v4: v3 + modulation over time
struct token_inflation_event_seq_v4
{
timestamp
schedule_time;
smt_inflation_unit unit;
int32_t
uint32_t
interval_seconds;
interval_count;
timestamp
timestamp
lep_time;
rep_time;
asset
asset
uint32_t
uint32_t
lep_abs_amount;
rep_abs_amount;
lep_rel_amount_numerator;
rep_rel_amount_numerator;
uint8_t
rel_amount_denom_bits;
};
Some notes about this:
• Only the numerator of relative amounts is interpolated, the denominator is the same
for both endpoints.
• For times before the left endpoint time, the amount at the left endpoint time is
used.
• For times after the right endpoint time, the amount at the right endpoint time is
used.
Code looks something like this:
if( now <= lep_time )
{
abs_amount = lep_abs_amount;
rel_amount_numerator = lep_rel_amount_numerator;
}
else if( now >= rep_time )
{
abs_amount = rep_abs_amount;
rel_amount_numerator = rep_rel_amount_numerator;
31
}
else
{
// t is a number between 0.0 and 1.0
// this calculation will need to be implemented
// slightly re-arranged so it uses all integer math
t = (now - lep_time) / (rep_time - lep_time)
abs_amount = lep_abs_amount * (1-t) + rep_abs_amount * t;
rel_amount_numerator = lep_rel_amount_numerator * (1-t) + rep_rel_amount_numerator * t;
}
Inflation operations
The inflation operation is specified as follows:
struct smt_setup_inflation_operation
{
account_name_type
control_account;
timestamp
smt_inflation_unit
schedule_time;
inflation_unit;
int32_t
uint32_t
interval_seconds = 0;
interval_count = 0;
timestamp
timestamp
lep_time;
rep_time;
asset
asset
uint32_t
uint32_t
lep_abs_amount;
rep_abs_amount;
lep_rel_amount_numerator = 0;
rep_rel_amount_numerator = 0;
uint8_t
rel_amount_denom_bits = 0;
extensions_type
extensions
};
The setup_inflation_operation is a pre-setup operation which must be executed before
the smt_setup_operation. See the section on pre-setup operations.
Inflation FAQ
• Q: Can the SMT inflation data structures express Steem’s current inflation scheme?
• A: Yes (except for rounding errors).
• Q: Can the SMT inflation data structures reward founders directly after X
months/years?
• A: Yes.
• Q: I don’t care about time modulation. Can I disable it?
32
• A: Yes, just set the lep_abs_amount == rep_abs_amount and lep_rel_amount_numerator
== rep_rel_amount_numerator to the same value, and set lep_time = rep_time
(any value will do).
• Q: Can some of this complexity be hidden by a well-designed UI?
• A: Yes.
• Q: Can we model the inflation as a function of time with complete accuracy?
• A: The inflation data structures can be fully modeled / simulated. For some issue
structures, the amount issued depends on how much is raised, so the issue structures
cannot be modeled with complete accuracy.
Named token parameters
Some behaviors of STEEM are influenced by compile-time configuration constants which
are implemented by #define statements in the steemd C++ source code. It makes sense
for the equivalent behaviors for SMTs to be configurable by the SMT creator.
These parameters are runtime_parameters and setup_parameters. The setup_parameters
are a field in smt_setup_operation; they must be set before smt_setup_operation, and
cannot be changed once smt_setup_operation is executed. The runtime_parameters
are a field in smt_set_runtime_parameters_operation, and they can be changed by
the token creator at any time.
These operations are defined as follows:
struct smt_set_setup_parameters_operation
{
account_name_type
flat_set< smt_setup_parameter >
extensions_type
control_account;
setup_parameters;
extensions;
};
struct smt_set_runtime_parameters_operation
{
account_name_type
flat_set< smt_runtime_parameter >
extensions_type
control_account;
runtime_parameters;
extensions;
};
Currently the following setup_parameters and runtime_parameters are defined:
struct smt_param_allow_vesting
struct smt_param_allow_voting
{ bool value = true;
{ bool value = true;
typedef static_variant<
smt_param_allow_vesting,
smt_param_allow_voting
> smt_setup_parameter;
struct smt_param_windows_v1
{
33
};
};
uint32_t cashout_window_seconds = 0;
// STEEM_CASHOUT_WINDOW_SECONDS
uint32_t reverse_auction_window_seconds = 0;
// STEEM_REVERSE_AUCTION_WINDOW_SECONDS
};
struct smt_param_vote_regeneration_period_seconds_v1
{
uint32_t vote_regeneration_period_seconds = 0; // STEEM_VOTE_REGENERATION_SECONDS
uint32_t votes_per_regeneration_period = 0;
};
struct smt_param_rewards_v1
{
uint128_t
content_constant = 0;
uint16_t
percent_curation_rewards = 0;
uint16_t
percent_content_rewards = 0;
curve_id
author_reward_curve;
curve_id
curation_reward_curve;
};
typedef static_variant<
smt_param_windows_v1,
smt_param_vote_regeneration_period_seconds_v1,
smt_param_rewards_v1
> smt_runtime_parameter;
UIs which allow inspecting or setting these parameters should be aware of the type and
scale of each parameter. In particular, percentage parameters are on a basis point scale
(i.e. 100% corresponds to a value of STEEM_100_PERCENT = 10000), and UIs or other
tools for creating or inspecting transactions must use the basis point scale.
Parameter constraints
Several dynamic parameters must be constrained to prevent abuse scenarios that could
harm token users.
• 0 < vote_regeneration_seconds < SMT_VESTING_WITHDRAW_INTERVAL_SECONDS
• 0 <= reverse_auction_window_seconds + SMT_UPVOTE_LOCKOUT < cashout_window_seconds
< SMT_VESTING_WITHDRAW_INTERVAL_SECONDS
SMT vesting semantics
SMTs have similar vesting (powerup / powerdown) semantics to STEEM. In particular:
• SMTs can be “powered up” into a vesting balance.
• SMTs in a vesting balance can be “powered down” over 13 weeks (controlled by static
SMT_VESTING_WITHDRAW_INTERVALS, SMT_VESTING_WITHDRAW_INTERVAL_SECONDS
parameters).
• Voting is affected only by powered-up tokens.
• Vesting balance cannot be transferred or sold.
34
Additionally, some token inflation may be directed to vesting balances. These newly
“printed” tokens are effectively split among all users with vesting balances, proportional
to the number of tokens they have vested. As the number of tokens printed is independent
of users’ vesting balances, the percentage rate of return this represents will vary depending
on how many tokens are vested at a time.
Content rewards
Tokens flow from SMT emissions into the reward fund. The blockchain uses algorithms
to decide:
• (1) How to divide the token-wide rewards among posts.
• (2) How to divide rewards within a post among the author and curators (upvoters)
of that post.
The algorithms to solve these problems operate as follows:
• (1) Posts are weighed against other posts according to the reward curve or rc.
• (2a) The curators collectively receive a fixed percentage of the post, specified by the
curation_pct parameter.
• (2b) The author receives the remainder (after applying any beneficiaries or limited/declined author reward).
• (2c) Curators are weighted against other curators of that post according to the
curation curve or cc.
Figure 7: Flow of initial tokens and SMT emissions
Curve definitions
The reward curve can be linear or quadratic. The linear reward curve rc(r) = r passes
the R-shares (upvotes) through unchanged. The quadratic reward curve rc(r) = rˆ2 +
35
2rs has increasing slope.
For an illustration of the meaning of reward curves, imagine grouping the most-upvoted
posts as follows:
• Section A consists of the top 10% of posts by upvotes.
• Section B consists of the next 10% of posts by upvotes.
Here’s how the rewards differ:
• With either reward curve, Section A posts will have greater rewards than Section
B posts, since they have more upvotes.
• With the quadratic reward curve, Section A posts will have an additional boost
relative to Section B posts, since Section A posts will get more rewards per upvote.
• With the linear reward curve, Section A and Section B will get the same reward
per upvote.
Possible curation curves are:
• Linear cc(r) = r
• Square-root cc(r) = sqrt(r)
• Bounded cc(r) = r / (r + 2s)
To help visualize, here are some plots called pie charts. Each colored area represents how
curation rewards are divided among curators with equal voting power.
36
Figure 8: Reward curves and curation curves
• The rectangular vertical column shows the immediate reward upon making an upvote.
• The colored area extending to the right shows how the rewards of a curator grow
as later curators vote.
• When both curves are linear, everyone gets the same curation reward regardless of
37
which post they vote on.
• In the case of rc_linear + cc_sqrt and rc_quadratic + cc_bounded, the same
height rectangles means everyone gets about the same initial curation reward, call
this ICR=.
• In the case of rc_linear + cc_bounded, the rectangles are decreasing in height.
This represents a progressive handicap against voting for already-popular posts, call
this ICR-.
• In the case of rc_quadratic + cc_sqrt and rc_quadratic + cc_linear, the
rectangles are increasing in height. Call this ICR+.
Fundamentally, curation is making a prediction that upvotes will occur in the future. As
reward system designers, our criterion for selecting a curve should be to reward successful
predictions. Which curve satisfies this criterion depends on the relationship between
current and future upvotes.
• If a post’s future upvotes are independent of its current upvotes, we should choose
an ICR= curve.
• If a post’s future upvotes are positively correlated with its current upvotes, we should
choose some ICR- curve, ideally somehow tuned to the amount of correlation.
• If a post’s future upvotes are negatively correlated with its current upvotes, we should
choose some ICR+ curve, ideally somehow tuned to the amount of correlation.
In practice, independence or a modest positive correlation should be expected, so an ICR=
or ICR- curve should be chosen. For STEEM itself, curation was originally the quadratic
ICR=, as of the Steem hard fork 19 it is the linear ICR=.
Target votes per day
Each account has a voting_power, which is essentially a “mana bar” that fills from 0%
to 100% over time at a constant rate. That rate is determined by two parameters:
• (a) The time it takes to regenerate the bar to 100%, vote_regeneration_period_seconds.
• (b) The voting_power used by a maximum-strength vote.
The vote_regeneration_period_seconds is specified directly. For (b), instead of
specifying the voting power of a maximum-strength vote directly, instead you specify
votes_per_regeneration_period. Then the maximum-strength vote is set such that a
user casting that many max-strength votes will exactly cancel the regeneration.
38
SMT Setup GUI Sketch
39
Figure 9: SMT Configuration
Votability and Rewardability
In this section, we introduce the concepts of votability and rewardability.
•
•
•
•
A token is votable for a comment if the balance of that token influences the comment.
For a given vote, each votable token of the comment is either rewardable or advisory.
If a token is rewardable, then the vote affects the comment’s reward in that token.
If a token is advisory, then the vote does not affect the comment’s reward in that
token.
Advisory votes do not affect rewards or voting power. However, the ranking algorithms
and estimated reward calculations still apply advisory votes, so UIs may display advisory
posts accordingly.
The votable token set is determined by allowed_vote_assets which is a comment_options_extension.
struct allowed_vote_assets
{
flat_map< account_name_type, votable_asset_info >
};
votable_assets;
struct votable_asset_info_v1
{
share_type
max_accepted_payout
= 0;
bool
allow_curation_rewards = false;
};
typedef static_variant< votable_asset_info_v1 >
votable_asset_info;
The following rules are applied to determine whether tokens are votable:
• STEEM is votable for every post.
• A token is votable for a post if it appears in the post’s votable_assets.
• Otherwise, the token is not votable for this post.
And these are the rules for whether a token is rewardable:
• In order to be rewardable for a post, a token must be votable for that post.
• If, for some post/token, that post’s max_accepted_payout of the token is zero, then
the token is not rewardable for that post.
• If some voter (i.e. upvoter / downvoter) has a zero balance of a token, then that
token is not rewardable for that voter’s votes.
• If the max_accepted_payout for any non-STEEM token is nonzero, then
the max_accepted_payout for STEEM/SBD must be at least the default
max_accepted_payout.
Implementation notes:
• For an advisory vote, all rewards are zero, including curators and beneficiaries. This
is because the blockchain applies the max_accepted_payout cap before the curator
/ beneficiary computations.
40
• Currently (as of Steem hard fork 19), the Steem blockchain does deduct voting
power for advisory Steem votes. This behavior will be changed in a future Steem
hard fork (Steem issue #1380).
• At most two tokens may be specified in votable_assets. This means that each
post is voted with at most three tokens (including STEEM).
• The default max_accepted_payout is stored in max_accepted_steem_payout_latch
member of dynamic_global_properties_object.
Clients should populate
max_accepted_payout of a post based on this member, in case the default value
changes in a future version.
No consensus level restriction forces any particular post to have any particular
allowed_vote_assets. As a consequence, any post may mark itself as eligible to be
rewarded in any token. However, UI’s may impose their own non-consensus validation
rules on allowed_vote_assets, and hide posts that violate these non-consensus
validation rules.
For example, in a Hivemind community with a corresponding token, there may be a validation rule that the allowed_vote_assets specified in each post within that Hivemind
community must include the token of that community. This is a non-consensus validation rule, since the entire concept of a post existing within a Hivemind community is a
non-consensus concept. Since it is a non-consensus validation rule, no consensus logic can
enforce it. However, UIs that are aware of Hivemind communities may refuse to index or
display posts that violate this validation rule.
Static Token Parameters
Static parameters are configuration constants that affect the behavior of SMTs, but are
deliberately excluded from smt_setup_parameters or smt_runtime_parameters. The
reason they are designed to be non-configurable is that allowing these parameters to
significantly deviate from the values used for STEEM would result in significant risks,
such as:
•
•
•
•
May
May
May
May
result in a very complicated implementation.
result in extreme end-user frustration.
threaten the security and stability of the token.
threaten the security and stability of STEEM.
Here is the list of such static parameters:
• SMT_UPVOTE_LOCKOUT_HF17 : Static – This value locks out upvotes from posts at a
certain time prior to “CASH OUT”, to prevent downvote abuse immediately prior
to “CASH OUT.”
• SMT_VESTING_WITHDRAW_INTERVALS : Static
• SMT_VESTING_WITHDRAW_INTERVAL_SECONDS : Static
• SMT_MAX_WITHDRAW_ROUTES : Static
• SMT_SAVINGS_WITHDRAW_TIME : Static
• SMT_SAVINGS_WITHDRAW_REQUEST_LIMIT : Static
• SMT_MAX_VOTE_CHANGES : Static
• SMT_MIN_VOTE_INTERVAL_SEC : Static
• SMT_MIN_ROOT_COMMENT_INTERVAL : Static
• SMT_MIN_REPLY_INTERVAL : Static
• SMT_MAX_COMMENT_DEPTH : Static
41
• SMT_SOFT_MAX_COMMENT_DEPTH : Static
• SMT_MIN_PERMLINK_LENGTH : Static
• SMT_MAX_PERMLINK_LENGTH : Static
Mandatory token parameters
The token parameters set by smt_setup_parameters or smt_runtime_parameters have
default values. A few STEEM-equivalent parameters are specified by smt_setup_operation
fields. These are the parameters which do not have a default value, and thus, must be
specified for every asset.
• SMT_MAX_SHARE_SUPPLY : Set by smt_setup_operation.max_supply
• SMT_BLOCKCHAIN_PRECISION : Set by pow(10, smt_setup_operation.decimal_places)
• SMT_BLOCKCHAIN_PRECISION_DIGITS : Set by smt_setup_operation.decimal_places
SMT interaction with existing operations
• comment_payout_beneficiaries : The existing comment_payout_beneficiaries
will only redirect STEEM. In the future, comment_payout_beneficiaries functionality which allows redirecting SMT rewards may be added.
• comment_options : max_accepted_payout, allow_votes only affects STEEM, see
here to restrict max_accepted_payout for assets. allow_curation_rewards affects
all tokens.
• vote_operation : Multiple tokens in the comment’s votable set vote.
• transfer_operation : Supports all SMTs.
• Escrow operations: Do not support SMTs.
• transfer_to_vesting_operation : Supports all SMTs that support vesting.
• withdraw_vesting_operation : Supports all SMTs that support vesting.
• set_withdraw_vesting_route_operation : Does not support SMTs.
• account_witness_vote_operation : SMTs do not affect witness votes.
• account_witness_proxy_operation : SMTs do not affect witness votes.
• feed_publish_operation : Feeds may not be published for SMTs.
• convert_operation : SMTs cannot be converted.
• Limit order operations : Limit orders are fully supported by SMTs trading against
STEEM.
• transfer_to_savings_operation : SMTs support savings.
• decline_voting_rights_operation : Affects SMT votes as well as STEEM votes.
• claim_reward_balance_operation : Restrictions on this operation are relaxed to
allow any asset in any of the three fields, including SMTs.
• delegate_vesting_shares_operation : Supports all SMTs that support vesting.
• Multisig Native: There is nothing “special” about the handling of SMT operations
signed by multiple signatures. If you set up your account to require multi-signature
security, then everything your account signs will need to be signed with multiple
signatures, as you specified. This includes operations your account does as a control
account managing an SMT, and operations your account does as a user holding SMT
tokens.
42
Automated Market Makers for SMTs
Automated Market Makers are smart contracts, largely based on the Bancor Protocol [2],
that may be constructed during the initial ICO setup of an SMT for providing perpetual
liquidity to an SMT community. For simplicity, Automated Market Makers in Steem may
only trade between STEEM and any given SMT.
Setup
Basic Definitions
In this article, we’ll let s represent a quantity of STEEM, let t represent a quantity of
some token (SMT), and let p represent a price, such that pt is STEEM-valued (i.e. if
MYTOKEN is trading at p = 0.05 STEEM / MYTOKEN then t = 120 MYTOKEN has
a value of pt = (0.05 STEEM / MYTOKEN) · (120 MYTOKEN) = 6 STEEM.
Suppose we have a market maker (or any economic agent) with a two-asset “portfolio”
(inventory) of s STEEM and t tokens. If the price of tokens is t, then we may measure of
the value of this portfolio, in units of STEEM, as v(p, s, t) = s + pt.
One common portfolio management policy is to require that STEEM should be some
constant fraction r of the portfolio, i.e. s = rv(p, s, t) where 0 < r < 1. We call this
policy the constant portfolio ratio or CPR policy, and the equation s = rv(p, s, t) is the
CPR invariant.
A different portfolio management policy, discussed by the Bancor whitepaper, is called
CRR or constant reserve ratio. To discuss CRR, let us notate the total number of tokens
in existence as T . The CRR invariant is then defined as s = rv(p, 0, T − t).
Notes on Conventions
We must discuss where our convention varies from the Bancor whitepaper. At some times,
when some user Alice interacts with the market maker, Alice will remove some tokens
from her balance to get STEEM from the market maker’s balance. On the other hand,
Bob may add some tokens to his balance in exchange for sending STEEM to the market
maker’s balance.
Bancor takes the convention that in this example, the market maker destroys tokens in
its interaction with Alice, and creates tokens in its interaction with Bob. The Bancor
convention suggests the market maker is not an ordinary actor, but needs system-level
“special powers” – specifically, the privilege to operate the token printing press – in order
to function.
In this paper, we adopt the convention that the tokens sent by Alice to the market maker
are not destroyed, but are instead added to the inventory (balance) of the market maker.
Likewise, the tokens sent to Bob by the market maker are not created out of thin air; they
already exist and are merely transferred from the inventory of the market maker to Bob.
Thus, we show that the market maker is essentially an ordinary economic agent acting
according to a deterministic algorithm – it doesn’t actually need “special powers”!
43
Finite Trades
Basic Definitions
A trade is a change in the market maker’s balance from (s, t) → (s+∆s, t+∆t). The price
at which the trade occurs is defined as p = −∆s
∆t . We restrict ourselves to well-formed
trades where either ∆s = ∆t = 0, or ∆s and ∆t are both nonzero and have opposite sign.
Theorem: A trade at price p conserves value at price p. More rigorously, if ∆s, ∆t
represent a trade at price p, then v(p, s, t) = v(p, s + ∆s, t + ∆t).
Let us more rigorously define the market maker’s state as a tuple M = (s, t, T, r). Given
some price p, we may define the restoring trade at p (also called a relaxing trade or a
relaxation) to be a trade which occurs at price p and results in a state that satisfies the
CRR invariant.
Computing the Restoring Trade
The restoring trade consists of functions ∆s(M, p) and ∆t(M, p). We may actually compute these functions from the definition of price and the CRR invariant:
∆s = −p∆t
s + ∆s = rv(p, 0, T − (t + ∆t))
⇒ s − p∆t = rv(p, 0, T − t − ∆t)
= rp(T − t − ∆t)
= rpT − rpt − rp∆t
⇒ rp∆t − p∆t = rp(T − t) − s
rp(T − t) − s
⇒ ∆t =
rp − p
(
)(
)
1
s
=
− r(T − t)
1−r
p
⇒ ∆s = −p∆t
(
)
1
=
(rp(T − t) − s)
1−r
Computing the Equilibrium Price
Given a state M , there exists some price peq (M ) for which the restoring trade is zero;
call this price the equilibrium price. We may compute the equilibrium price by setting
∆s = 0:
44
∆s = (
0
)
1
=
(rpeq (T − t) − s)
1−r
⇒ rpeq (T − t) − s = 0
⇒ rpeq (T − t) =
⇒ peq
=
s
s
r(T − t)
Theorem: Relaxation is idempotent. That is, after relaxing at price p, the equilibrium
price of the resulting state is p, and a second relaxation at price p will be a zero trade.
Example
Example: Suppose M = (1200, 3600, 12000, 0.25) and p = 0.5. Then of the T = 12000
TOKEN in existence, t = 3600 TOKEN is held by the MM, so T −t = 12000−3600 = 8400
TOKEN are “circulating” (i.e. exist in balances outside the MM). These circulating tokens
are worth p(T − t) = 4200 STEEM total, so they “should be” backed by a target reserve
level of rp(T − t) = 4200 ∗ 0.25 = 1050 STEEM.
In this example, there is “too much” STEEM in the reserve, so relaxation will buy tokens
in the market. This sale will cause two effects: It will decrease the reserve STEEM, and
also decrease circulating tokens. The decrease in circulating tokens, in turn, causes the
target reserve level to decline. For every 1 STEEM used to buy tokens, the target reserve
level declines by r STEEM; since r < 1 eventually the declining reserve will “catch up”
to its more slowly declining target level.
(
)
1
The above algebra shows that we will catch up at ∆s = 1−r
(rp(T − t) − s) and
(
)(
)
1
s
∆t = 1−r
p − r(T − t) . Running the calculations with the numbers defined in this
example gives ∆s = −200 STEEM and ∆t = 400 TOKEN.
Let’s check that these computed values ∆s = −200, ∆t = 400 (a) represent a trade with
price 0.5, and (b) that the CRR invariant holds for the new state Mnew = (s + ∆s, t +
∆t, T, r). Calculating p = −∆s
∆t we indeed get p = 0.5. After this trade executes, the
market maker has snew = s + ∆s = 1200 − 200 = 1000 STEEM, and tnew = t + ∆t =
3600 + 400 = 4000 tokens.
To check condition (b), that the CRR invariant holds, we effectively repeat the analysis
in the initial paragraph of this example with the new numbers. We know Mnew =
(1000, 4000, 12000, 0.25) and p = 0.5. Then of the T = 12000 TOKEN in existence,
tnew = 4000 TOKEN is now held by the MM, so T −tnew = 12000−4000 = 8000 TOKEN
are now circulating. These circulating tokens are worth p(T − tnew ) = 4000 STEEM total,
so they “should be” backed by a target reserve level of rp(T − t) = 4000 ∗ 0.25 = 1000
STEEM. Since the target reserve level indeed exactly matches the actual reserve level of
snew = 1000 STEEM, we conclude that the CRR invariant is satisfied after this relaxing
trade.
45
Infinitesimal Trades
This section is fairly technical; the reader will need a good grasp of calculus and differential
equations to follow the results.
Setting up the Problem
Suppose we satisfy the invariant condition at some price p = pe q; by the CRR invariant
s = rv(p, 0, T − t) = rp(T − t). Suppose the price then increases to p + ∆p and a relaxing
trade ∆s, ∆t occurs at this new price.
In this section we consider the limiting situation where ∆p is infinitesimally small, so we
will use Leibniz notation (dp for a small change in p, ds for a small change in s, dt for a
small change in t).
Solving the DE’s
By applying the substitution p ← p+dp to the expression for ∆s computed in the previous
section, we obtain an expression which simplifies to a separable DE which can be solved:
ds
=
=
=
=
1
⇒ dp =
p
∫
1
⇒
dp =
p
⇒ ln(p) =
⇒p
1
(r(p + dp)(T − t) − s)
1−r
1
(rp(T − t) + rdp(T − t) − rp(T − t))
1−r
1
r(T − t)dp
1−r
( )
s
1
dp
1−r p
(
)
1
(1 − r)
ds
s
∫
1
(1 − r)
ds
s
(1 − r) ln(s) + C0
= k0 s1−r
Similarly for t, we can start from dt = −ds/p and again obtain and solve a separable DE:
46
= −ds/p
r
= −
(T − t)dp/p
1−r
( )
r
1
1
dt = −
dp
⇒
T −t
1−r p
(
)
1
1−r
1
⇒ dp =
dt
p
r
t−T
∫
∫
1−r
1
1
dp =
dt
⇒
p
r
t−T
1−r
⇒ ln(p) =
ln |t − T | + C1
r
1−r
⇒ p = k1 (T − t) r
dt
Qualitative discussion
In a CRR market maker, where does the “backing” for newly emitted tokens come from?
One option is to lower the reserve ratio r. This option results in no immediate market
activity, but will weaken the response of the market maker to any future price changes.
This is called the “pay later” option.
Another option is to change the dynamical system’s initial conditions, i.e. edit the constants of integration. This option will cause the equilibrium price peq to drop, meaning
the market maker will more aggressively sell tokens to replenish the reserve. If order
books are deep compared to the amount of emission, and there are adequate buyers for
the tokens, then the sales will be able to replenish the reserve and keep the equilibrium
price near its old value; the deep order books provide resistance to the price change being
driven by the market maker. If order books are thin compared to the amount of emission,
and there are few/no buyers for the tokens, then the equilibrium price will fall, breaking through the thin orders and lowering the market price. Even though few/no many
tokens were sold, so even though the absolute amount of STEEM in the reserve is still
nearly/exactly the same as before, the reserve’s value relative to the now-lower market
cap of the token has increased to the reserve ratio. This option is the “pay now” option.
FAQ
Q: What is the relevance of constant portfolio ratio policy?
A: It may become a supported market maker policy in the future.
Q: Can the reserve ratio go over 100 percent?
A: No.
Q: Can the reserve ratio be exactly 100 percent?
A: Not with the system described in this paper. It might be possible to code as a special
case.
47
Q: In a CRR market maker, where does the “backing” for newly emitted tokens come
from?
A: As blockchain designers, we have two options for sourcing the “backing”. One option
is to lower the reserve ratio r. This option results in no immediate market activity, but
will weaken the response of the market maker to any future price changes. This is called
the “pay later” option.
Another option is to change the dynamical system’s initial conditions, i.e. edit the constants of integration. This option will cause the equilibrium price peq to drop, meaning
the market maker will more aggressively sell tokens to replenish the reserve. If order
books are deep compared to the amount of emission, and there are adequate buyers for
the tokens, then the sales will be able to replenish the reserve to its target level while
keeping the equilibrium price near its old value. The deep order books provide resistance
to the price change being driven by the market maker.
If order books are thin compared to the amount of emission, and there are few/no buyers
for the tokens, then the equilibrium price will fall, breaking through the thin orders and
lowering the market price. Even though few/no many tokens were sold, so even though
the absolute amount of STEEM in the reserve is still nearly/exactly the same as before,
the reserve’s value relative to the now-lower market cap of the token has increased to the
reserve ratio. This option is the “pay now” option.
Q: Where’s the “don’t pay” option?
A: You have to come up with some answer to where the “backing” for newly emitted
tokens will come from. Unless there’s no emission. Or unless there’s no “backing” for any
tokens. So the “don’t pay” option would be to have an SMT with either no emission, or
no market maker.
Q: Don’t fractional exponents require floating point to implement?
A: Only if you need fairly high precision (we don’t), don’t care about bit-for-bit reproducibility across compilers, OS’s, CPU’s, etc (we do), and need to do massive numbers
of calculations quickly (we don’t). A fast, approximate, all-integer implementation is
possible.
Q: Does this market maker interact with the order book through the existing limit order
system, or is it a separate set of operations?
A: In theory, it could be implemented either way. However, the likely outcome is that
the market maker will be implemented outside of order-book markets to allow its code to
be modularized. In practice, if implemented as a completely separate subsystem, people
will run arbitrage bots which will trade away any price differences between the reserve
system and the existing market system.
Q: Where do the market maker’s initial token balances come from?
A: ICO units can specify the market maker as a destination. An ICO creator may direct
a percentage of their ICO’s STEEM contributions to the MM by specifying the market
maker similarly to specifying a founder. Or may use the soft cap system to specify all
STEEM above a pre-determined amount goes to the ICO. Likewise, a fixed or percentage
amount of tokens can be added in the ICO to increase the MM’s token balance.
Q: Can someone send STEEM or tokens to the market maker?
A: Yes.
48
Q: What are the side effects of sending STEEM or tokens to the market maker?
A: The constants of integration are re-initialized, meaning the equilibrium price will
change. The market maker will become more aggressive about selling the asset.
Q: Can’t this cause manipulation or appropriating the market maker’s inventory to private
profit?
A: Sending assets to the market maker does cause it to engage in trading activity which
affects the price. However, dumping an identical amount on the market will result in a
larger amount of trading activity and a larger effect on the price. If Eve is willing to
spend her tokens/STEEM to manipulate prices, she would prefer the strategy of simply
dumping tokens/STEEM on the market, as that strategy is more cost-effective for her.
Q: Does the market maker’s activity generate profits (losses)?
A: It depends on how you measure “profits”. If you measure the value of STEEM and
tokens in some external third currency such as US dollars or bitcoins, the market maker’s
inventory, valued in that currency, can definitely increase or decrease. If people voluntarily
send STEEM or tokens to the market maker, such activity definitely increases the value
of the market maker regardless of your measurement.
Another way to define profits is by the constants of integration. If both of the constants
of integration increase, or one increases while the other remains the same, a tiny increase
occurs with each trade when the market maker is in “taker” mode.
Q: What is “taker” mode? How can a market maker be set to operate in “taker” mode?
A: When orders execute, the order used to set the price is called the maker; the maker’s
counterparty is the taker. In the STEEM on-chain market (and on almost all trading
platforms) the older order is always the maker.
When the market maker is in taker mode, its actions are always considered to be taker
orders, which execute at the price specified by the user acting as its counterparty – this
price is always at least a little bit more favorable than the market maker is willing to
accept. When the market maker is not in taker mode, its actions are always considered
to be maker orders, which don’t generate changes in the constants of integration.
Taker mode is a runtime parameter that can be set by the SMT’s control account.
Q: Who benefits from the profits of a market maker in taker mode?
A: Maybe nobody, or maybe everybody. It’s decentralized.
Q: OK, if my SMT reaches a steady price, the STEEM in the reserve is basically locked
up forever. That seems not cool. How do I set it up so that this STEEM can be unlocked
for the benefits of my SMT users?
A: Set the DRR (decaying reserve ratio) setup parameter. If you set DRR, then the reserve
ratio will slowly drop over time to a pre-set value, using its excess STEEM reserves to
buy excess tokens. Setting DRR is an excellent, fair, decentralized way to return excess
capitalization to contributors in a more-popular-than-anticipated ICO that raises more
than the sponsor can effectively spend.
Q: If the reserve ratio can change over time due to pay-later emissions or DRR, it’s not
really a constant reserve ratio, is it?
49
A: No, they’re not. The reserve ratio’s called “constant” because it’s constant over the
short-term, in normal conditions, or in the conditions in Bancor which is where it was
named. But the name could be regarded as slightly misleading.
Q: If the constants of integration can change over time, they’re not really constants either,
are they?
A: No, they’re not. They’re called constants of integration because that’s their mathematical role in the calculation that introduces them. Maybe they’ll be differently named
in a future version of this paper.
Q: Can I specify a contribution to a DRR to be a pay-later contribution, that increases
its reserve ratio, the increase to be eventually negated over time by future decay? Why
would I want to?
A: Yes. This is effectively contributing to the market maker, subject to the condition
that it’s not allowed to immediately dump a portion of the contribution. It’s useful if
you want to make a large contribution to a market maker without causing it to create a
disturbance by immediately dumping a significant fraction of your contribution onto the
market.
Q: Can I specify a DRR with emission to use pay-later for emissions when the RR is
decaying?
A: Yes.
Q: Is the market maker specified here equivalent to a Bancor token changer?
A: No. A Bancor token changer has multiple reserve ratios that must sum to one hundred
percent, and involves a third token that effectively represents equity in the token changer.
This paper’s market maker has none of these features.
Q: I want to have an initial “price discovery” period where people trade without action
from the market maker, then have tokens and STEEM from the ICO gradually flow in
over time to the market maker so it has a delayed, slow start from zero to full power. Can
I do it?
A: This is called “gradual seeding” and it may be supported.
Q: What about numerical stability?
A: A market maker will be restricted to only operate when its balances exceed a certain
minimum for both assets. Also, reserve ratios will be restricted to a certain range, all
the mechanisms that can set / increase / decrease a reserve ratio will be restricted to not
allow it to move outside the range. Tentative numerical experiments suggests these limits
should be about 10,000 satoshis of both assets, 5 percent and 50 percent, respectively.
These values are subject to change based on future experimentation, worst-case analysis,
and testing.
Costs of SMT Operations And Bandwidth Rate Limiting
Like STEEM, SMTs can be transferred on the Steem blockchain with zero fees. Steem
replaces fees with bandwidth rate limiting based on the percentage of STEEM an ac-
50
count has staked, which means the blockchain calculates how much STEEM an account
has temporaily vested to determine how much bandwidth the account is permitted for
transfers, posting, and other operations across a period of time. In a future version of
Steem, possesion of an account name could permit some small degree of bandwidth to
allow for even greater user experience.
Fee-less Operations Necessary for Quality User Experience
Because of bandwidth rate limiting, Steem may never charge applications or users transaction fees for basic operations such as voting, posting, and transferring tokens. This lack
of fees allows Steem based apps to compete with their non-blockchain counterparts, such
as Facebook or Reddit, which certainly do not charge fees for actions such as ‘Like’ and
‘Upvote’. If these applications did charge fees, adoption would suffer.
Decentralized Exchange
One of the valuable features of SMTs is their immediate access to functioning unmanned
markets against the liquid asset, STEEM.
Automatic Order Matching
The Decentralized Exchange (DEX) structures of Steem allow assets to be automatically
matched for best possible price when bids and asks overlap, unlike other DEXs - which
require a “man in the middle” or user-agent to match orders. Automatic, rather than
middle-man-facilitated, order matching is important for the security of Steem-based assets,
and for the replicability and safety of DEX interfaces.
Diverse Asset Types
There are several assets that SMT users and creators will have access to by way of the
Steem DEX: STEEM, SBD, SMTs, and Simple Derivatives (IOUs). These neighboring
assets can increase the visibility and network effect of all created SMTs.
STEEM is the gateway token for assets issued on Steem, staying relevant by acting as
the bandwidth usage measuring stick across Steem’s SMTs. STEEM is also the common
denominator asset, acting as a trading pair for all of Steem’s SMTs.
SBD (Steem Blockchain Dollars) are an experimental asset on Steem that relate to the
US Dollar, originating with Steem’s launch in 2016. It is unclear if SBD will bring value
to holders of USD as they will compete, possibly poorly, with USD IOU tokens; however,
SBDs will bring value to speculators.
SMTs as described in this proposal are an important part of growing the token ecosystem,
and bringing crypto assets to the mainstream. SMTs will trade against STEEM across
the DEX.
Simple Derivatives (IOUs) will be possible via SMT issuance. For instance, if an SMT
is issued without inflation or rewards pool properties, then the issuer can reliably back
51
the token with another real world asset such as bitcoin or USD. In this instance, the
issuer could create a business functioning as a gateway, by trading their IOU for BTC
or USD. Users would buy the IOU to gain access to the Steem DEX. This market would
add diversity and value flow to the Steem ecosystem, while adding to the DEX’s network
effect.
Zero Trading and Transfer Fees
The Steem DEX is the first DEX to exist without trading fees, to the benefit of SMT
creators and traders alike. This is made possible by bandwidth rate limiting (described
in the original Steem Whitepaper and Bluepaper), as the process by which the blockchain
calculates transaction “prices” on a per byte basis, and deducts transaction bandwidth
available to an account temporarily. These “prices” are an internal blockchain accounting
and do not debit any token balances.
Augmenting SMTs with Additional Native Contracts
There are several potentially valuable programmable contracts that are not in the immediate scope of SMTs, however, these contract capabilities can be created as modular,
follow-on projects that increase the creativity entrepreneurs and communities may apply
to growth of SMT ecosystems.
Community Building with Paid Positions
SMT communities may be bolstered with paid positions, guild roles, or jobs that are
defined in programmable, native smart contracts and matched with continuously elected
participants. Rewards received through the elected position come from some portion of
the token’s Founder allocations or donations that are sent to a paid position contract.
Paid position contracts may be defined for length of position, frequency and volume of
payments, particular token used for stake-weighted elections, percentage of the token
required for a participant to be elected, and how tokens in paid position contracts are
socialized or forfeited given no participant is elected.
The paid roles may be leveraged to support various applications, games, and businesses
built around an SMT. A contract for a paid position, the postion’s reward schedule, and
the voting thresholds required to elect an account into a paid position may be created by
anyone for a fee. To establish the purpose of these positions, job descriptions or constitutions that encourage adherence to performance expectations may be established by the
issuer or the token’s community. There can be an unlimited number of paid positions,
and paid position contracts can receive any amount of a token’s Founder allocations or
community donations. The types of paid positions that may be employed includes everything from front end developer, to evangelist, including educational content creator,
business development representative, and many roles that have yet to be imagined.
52
Democratic SMTs using Whitelist Oracles
SMTs represent completely open access to tokens, however, some entities may wish to
enable one-whitelisted-account, one-vote-per-post and X-number-of-target-votes-per-day
algorithms to increase their token’s potential for accurate wisdom-of-the-crowd content
discovery mechanics and the democratic nature of their token community. To incorporate
this, the Rewards Pool for a token will need to have a manageable whitelist that can be
enabled only at launch. Whitelist management may be handled by the entity launching
the token or outsourced to an identity management service, such as Civic or Jumio. The
service would need to publish a feed of Steem usernames for known/identified people into
the Steem blockchain, along with periodic updates to ensure accuracy of the whitelist.
As the blockchain pays rewards to a token, it verifies the account receiving the token is
on the whitelist, otherwise the tokens are returned to the reward pool.
Secondary ICOs for Contiguous Fundraising
Entrepreneurs leveraging SMTs to finance ventures may want to have the option to perform token auctions after the initial launch of the token. The entrepreneur can reserve
Founders tokens at launch and earmark them for later sale, however, they may want to
auction these tokens rather than sell them into Bid/Ask order books or sell them OTC.
To enable secondary auction-style ICOs, a secondary auction contract may be established.
This contract requires definitions for when an ICO begins and how long it lasts, as well as
lockup periods for the tokens purchased. The lockup period allows the tokens to be sold
at a discount to the open markets and attract investment capital that would otherwise
stay out of the market. The entrepreneur will send tokens to this contract prior to the
beginning of the auction and the tokens will be distributed to the auction participants
immediately following the close of auction period.
Bandwidth Sharing with SMTs Based on Reserve Liquidity Pools
SMTs that use ICOs to create Automated Market Makers to boost token liquidity will
inherit bandwidth rights proportionate to the amount of STEEM in the Automated Market Maker’s reserve pool. This bandwidth inheritance confers transaction rights from
STEEM to all the of the “powered up” and vested SMT, basically permitting SMT owners to transact proportionate to their stake of SMT without owning STEEM outright.
Bandwidth Sharing based on liquidity pools enables new tokens to operate with an even
higher degree of independence while still contributing proportionate value to STEEM.
What Makes SMTs Better Suited to ApplicationSpecific Blockchains, such as Steem, than ApplicationGeneral Blockchains, such as Ethereum?
Throughout the history of software and hardware development, it has been observed
that specialized systems have the potential to greatly outperform generalized systems.
An example of this can be seen in GPUs outperforming CPUs through specialization,
which was followed by ASICs outperforming GPUs for particular tasks. In turn, some
53
wonder how a specialized blockchain, such as Steem, which hosts application-specific programmability, and static mechanics embedded in consensus, is more suited to SMTs than
application-general, open-programmability blockchains, such as Ethereum, which hosts
turing-complete (“infinitely”) programmable smart contracts in a layer beyond consensus, and has shown its use for discovering new cryptocurrency concepts. Without delving
into Steem’s advantages in network effect and developer team experience, the advantages
for SMTs on Steem can be seen through a set of computer science, consumer safety, and
economic perspectives.
SMTs are Safer and More Cost Effective in Application-Specific
Blockchain Environments
The value of SMTs in a native, specialized-programmability environment, such as Steem,
comes from reliability of the code and efficiencies created by that reliability, whereas
application-general platforms, such as Ethereum and Tezos, require costly and highlyassumptive audits on each new token and issuer to be deemed safe. Some of these
application-general protocols claim to have formal verification, which is valuable, however, the majority of the audit cost remains due to the need to audit the issuer’s choice
of token mechanics, choice of client for writing the code, and semantics of custom code
written to the token. Enabled by the purposeful design of its code, Steem enables SMTs
to support static (versus dynamic) crypto-economic properties that can be tuned after
the token’s launch without each change potentially harming their token holders. The
purposeful delineation between economic properties that should be static versus dynamic
makes the necessary token audits for safety simple and inexpensive to accomplish.
To elucidate this issue, imagine someone is offering you 20% of their currency in exchange
for $100 USD. You will have additional questions for the seller - essentially questions to
audit tertiary realities of the deal, such as: “does the seller maintain a right to print more
currency and therefore dilute me?” In SMTs, holders of SMTs will be able to rely on the
core economics of the SMTs they purchase due to static nature of the SMTs economic
properties - such as emissions or inflation rates, which cannot be changed by the issuer
after launch. Therefore, there can be no unexpected new currency emissions to harm
the consumer. In application-general, open-programmability blockchain protocols, such
as Ethereum and Tezos, there can be no such platform-spanning design principles and
reliabilities that protect consumer safety.
SMTs on Steem have Aligned Proof-of-Brain Incentives with the
Core Token
Unlike STEEM, core tokens (such as ETH) that do not carry Proof-of-Brain content
rewards, cannot offer monetization, primed active user-base, shared influence, and bootstrapping benefits to new SMT communities. STEEM, on the other hand, is able to lend
its reward pool features and primed-user base to new networks, to help them bootstrap,
market, and become successful independent clusters of participants on the network. Conversely, some entrepreneurs will identify and choose a strategy to employ SMTs largely
independent from STEEM, and like ERC20 to Ethereum, SMTs can run while only having STEEM run in the background to calculate the necessary bandwidth for transaction
costs.
54
SMTs on Steem Have Transaction Pricing that Contributes to a
Quality User Experience
Whether operating with bandwidth rate limiting, or outright fees, no general purpose
blockchain will price transactions effectively for more than a small fraction of its applications, and SMTs would have reduced user experience (UX) on application-general
blockchains (such as Ethereum) as a result. The clear example is that on blockchains
such as Ethereum, there are outright fees for all transactions, however, no content publisher would expect users to pay fees to leave comments or likes on their articles. With
SMTs on Ethereum, those fees would be required, which makes Ethereum a non-starter
as an SMT platform.
Unlike Ethereum, some open-programmability blockchains of the future may use bandwidth rate limiting as transaction costs, however, bandwidth rate limiting requires fine
tuning to meet the UX requirements of specific applications. As an example, in Steem,
bandwidth rate limiting is specifically tailored to support content applications and their
user interactions by leveraging bandwidth rights according to two objects: amount of
token ownership, and account ownership - and it’s taken over a year of production-level
research to refine the optimal bandwidth allowances to each. In general purpose, openprogrammability platforms, the burden and the need for accurate pricing may hinder the
ability for applications to have their users’ actions priced appropriately, and the problem may be exacerbated as a greater myriad of potential application experiments come
to exist, stretching and sharing the blockchain’s resources. Therefore, blockchains that
support native application-specificity may yield more suitable transaction pricing, as it
pertains to the UX with tokens in related applications.
SMTs Benefit from a Blockchain that has Scaling Processes Programmed to a Specialized Set of Applications
In blockchain scaling there are cutting-edge concepts of “sharding” (originated by Vitalik Buterin and the Ethereum project) and “multi-threaded parallelism” (originated by
Michael Vandeberg of Steem) that refer to how blockchains may scale by allowing multiple
operations to occur at once. General purpose platforms (such as Ethereum) are a great
test bed for these approaches to scaling, however, a platform that takes advantage of all
the product-market fit discovered by Ethereum, that then applies it to a more specialized,
iterative-upgrading model, such as Steem, can scale its processes more effectively to meet
the demand discovered by that product-market fit.
Looking to the 90s and early 00s for analogy, when the computer science world started
writing code specifically optimized for GPUs, the boundary pushing for greater scale occurred through FPGAs: field programmable gate arrays, which are chips that allow the
programmability of the set of logic gates into the form of any conceivable circuit, allowing for effectively a prototype ASIC (albeit with higher power consumption). This is not
quite the same performance per watt as an ASIC, but orders of magnitude faster than
a CPU for particular tasks. As these platforms move to more and more generalizations,
such as the idea that any contract may call on any other contract, they will move further away from ability to optimize for scale, as contracts that call on all other contracts
can reduce the capacity for multi-parallel processing to single-core processing. By analogy, like CPUs do not optimize better than GPUs, platforms like Ethereum, GEOS, and
Tezos do not optimize better than Turing-incomplete application-specific blockchains like
55
Steem. These CPU-like blockchains will be bottlenecked by unpredictable processing requirements, while the ultimate blockchain platforms will be specially-designed, like Steem,
and will scale by optimizing in the way FPGAs were optimized for parallel algorithms.
SMTs Benefit from a Blockchain with Content Management System (CMS) Primitives
Unlike application-general blockchains, such as Ethereum, that inherently avoid
application-specific primitives at the core of the protocol, Steem offers a structured
public content database for storing plain text and generic structured data in tandem with
content primitives that developers can build from: Account Names, Posts, Comments,
Votes and Account Balance. These primitives benefit the blockchain-based applications
by helping to establish application-interoperability and rapid developer on-boarding.
Without these primitives, second order databases need to be structured specifically for
a blockchain-based application, which may give rise to many second-order applicationspecific databases competing with each other. The rise of multiple second layer content
databases splits the potential network effect for the blockchain as a content management
system (CMS), and reduces the potential for application-interoperability, which provides
consumer safety benefits by allowing end users to move fluidly from one blockchain-based
application to another.
Increasing Market Demand for STEEM with SMTs and
Implicit Value Drivers rather than Fees
There are several new value drivers to STEEM with the creation of SMTs.
STEEM Purchased for Transaction Bandwidth Enables Maximally
Profitable Participation across SMTs
With the advent of SMTs, there is growing demand for users to hold STEEM, because
users need to increasingly hold STEEM in order to participate, consume, and use Steem
services at a rate maximally commensurate with their growing potentials in respective
SMT ecosystems. Put simply, as power users are growing their earning potential in SMT
communities, they need more STEEM to achieve the bandwidth allowance needed to
perform at their highest possible rate of return in SMT ecosystems. At an application
level, the demand for bandwidth may be satisfied by users or by businesses, which can
delegate surplus bandwidth to their users.
STEEM Supply is Locked into Liquidity Pools by Automated Market Makers
Each SMT that leverages Automated Market Makers augments the ratio of demand for
STEEM to available supply of STEEM. The effect of the Automated Market Maker to
STEEM is that each Automated Market Maker represents a permanent holding pool for
STEEM, which represents a decrease in available supply. Given demand were to stay
56
equal, the price of STEEM is caused to rise with the advent of each new Automated
Market Maker.
STEEM and SMT Demand Increases with Advent of New Powers
of Influence
From a potential utility perspective, demand for STEEM increases as each SMT is created
with Influence Sharing for Steem Power over a SMT’s rewards pool. The advent of each
trace of Steem Power-based Shared Influence over an SMT’s Reward Pool gives new rights
and usage to STEEM, which in turn drives demand for STEEM. These rights can also
be granted from SMT to SMT, and the flow of value follows an identical pattern.
STEEM Demand Increases with Proliferation of SMT ICOs
At a platform level, other cause for demand may include exclusive financing opportunities,
such as ICOs, which attract new capital into ecosystems, first flowing into the base asset,
STEEM, and then flowing into SMTs. Increased capital in the ecosystem due to ICOs
always presents an opportunity for net positive capital retained in STEEM, and at worst,
a wash on the value of the base asset, where all of the STEEM is sold by the organization
making the offering. The example of the worst case scenario is that an ICO occurs and
$100 USD buys STEEM to buy the ICO’d SMT, then 100% of the STEEM received by the
ICO is sold for USD - and no explicit net effect related to the value of STEEM. However,
even when the net effect contribution of an ICO to the value of STEEM is apparently
zero, it is an implicit net benefit in terms of attention received by STEEM and the Steem
ecosystem, if we consider all new attention valuable. Further, it is reasonable to expect,
based on the behavior of ICOs in Ethereum, that the majority of the STEEM received
by the ICO’ing organization will continue to be held on a speculative or promissory basis,
therefore creating holding value.
Steem: The World’s Advertising Network
Along with these new value creating mechanisms, it is imperative to recognize the original
value created for STEEM as an implicit attention and advertising network that now
applies to all SMTs that utilize Proof-of-Brain rewards. Smart Media Tokens, such as
STEEM, have inherent curation properties, such as their Rewards Pools, that give them
reliability and credentials as an implicit advertising network. The Rewards Pool in SMTs
demands that fully-SMT-integrated interfaces, such as steemit.com, respect the pending
SMT payouts on posts and then rank these posts from highest to lowest pending payout
in pages often referred to as “Trending” - such that the posts can be audited by the
community of SMT holders. The effect of this, which applies equally to STEEM as other
SMTs, is a sorted “Trending” page that users (bloggers, vloggers, advertisers) can reliably
use to evaluate the potential returns on buying higher placement on the page to attain
more attention, and then these participants make decisions to buy or rent STEEM and
SMTs to promote content. Through this process, as advertisers choose to buy and
rent STEEM/SMTs to gain exposure, demand for STEEM/SMTs increases. These value
driving properties can be described in a way similar to “Ethereum: the world computer”,
but instead as “Steem: The world’s advertising network.”
57
Steem Ecosystem Support for SMTs
Integrating SMTs into Websites and Apps
APIs and Documentation
To be continuously updated for SMTs. Current Steem APIs exist here: http://steem.
readthedocs.io/en/latest/index.html and https://steemit.github.io/steemit-docs/
Shared Tools for Account Creation, Key Signing, and Wallet Functions
Several shared tools exist to support applications that wish to outsource signup, transaction signing, and wallet functions - such as SteemConnect. SteemConnect enables
applications to support SMTs while the applications are backed by entrepreneurs who
may have little to no cryptocurrency experience.
Conclusion
Through a combination of specialized designs for open asset-issuance, bandwidth rate limiting as transaction costs, permanent-availability of content, real-time transaction speeds,
autonomous distribution of tokens, decentralized exchange, automated market making
and ICO contracts, Steem offers the premier token protocol for publishers across the
internet.
References
[1] Steemit, Inc., 2017. Steem Bluepaper. A protocol for bringing smart, social currency to publishers and content businesses across the internet. (https://www.steem.io/
steem-bluepaper.pdf)
[2] Eyal Hertzog, Guy Benartzi & Galia Benartzi, 2017. Bancor Protocol. Continuous
Liquidity and Asynchronous Price Discovery for Tokens through their Smart Contracts.
(https://www.bancor.network/static/bancor_protocol_whitepaper_en.pdf)
Appendix
Implementation Notes
Here is a timeline / state diagram of the events in an SMT launch:
58
Figure 10: Timeline of SMT Launch
SMT naming standards
• An SMT name should consist of 3-10 uppercase ASCII letters (A-Z).
• An SMT name should not equal STEEM, SBD or VESTS.
Asset directory standards
A directory maps each NAI to one of the following states:
Listed
Deprecated
Unlisted
Blacklisted
Each possible asset name is mapped to one of the following states:
Free
Reserved
59
A Listed or Deprecated NAI has an associated name, which should be listed as Reserved
in the mapping.
UIs may provide asset directory union functionality to augment directories by combining
multiple asset directories into a single asset directory. Asset directory union should use
the following algorithm to resolve situations where an NAI is listed differently by different
directories:
• (1) If the NAI is Blacklisted in any component directory, return Blacklisted.
• (2) If the NAI is Listed or Deprecated in multiple component directories, and
all of the component directories do not agree on the associated name, return
Unlisted.
• (3) If the NAI is Listed in at least one component directory, return Listed.
• (4) If the NAI is Deprecated in at least one component directory, return
Deprecated.
• (5) Return Unlisted.
Likewise, here are the rules for resolving names listed differently by different directories:
• (1) If the name is Reserved in any component directory, return Reserved.
• (2) Return Free.
A dynamic directory (based on a URL or blockchain account) should not be cached more
than 5 minutes.
UI guidelines for SMT names
• A UI may, but need not, have a default asset directory.
• A UI may choose to hide unlisted NAIs.
• A UI should allow users to override or augment the UIs defaults with their own
asset directories.
• A UI should reconsider hiding unlisted NAIs in which the user has actively transacted.
Operational guidelines for asset directories
• An asset directory should not confuse users by setting a well-known NAI to refer to
a different name, or setting a well-known name to refer to a different NAI.
• An asset directory should make the process for listing clear to both SMT creators
seeking to add their asset to the directory, and UI developers considering adding
the directory to their UI.
Asset directory formats
URL and file-based asset directories will be a JSON format. The details will be developed concurrently with the implementation. Blockchain-based asset directories will use
a custom JSON operation. Again, the details will be developed concurrently with the
implementation.
60
Unit Tests
The details of the unit tests will be developed concurrently with the implementation.
61