ERC-2848 - My Own Messages (MOM)

Created 2020-08-02
Status Stagnant
Category ERC
Type Standards Track
Authors

Simple Summary

My Own Messages (MOM) is a standard to create your very own public, always updated, unstoppable, verifiable, message board.

Abstract

My Own Messages (MOM) use Ethereum as a certification layer for commands and multihash of your messages. It don't use smart contracts but simple self-send transactions with specific payload attached.

To ge more insights, you can test a live client, watch a full video overview and demo and read a brief presentation.

Motivation

As a developer or pool's owner, I'd like to send messages to my users in a decentralized way. They must be able to easily verify my role in the smart contract context (owner, user, and so on) and they must be able to do it without relying on external, insecure and hackable social media sites (Facebook, Twitter, you name it). Also, I'd like to read messages from my userbase, in the same secure and verifiable manner.

As a user, I want a method to easily share my thoughts and idea, publish content, send messages, receive feedback, receive tips, and so on, without dealing with any complexity: just write a message, send it and it's done. Also, I want to write to some smart contract's owner or to the sender of some transaction.

As an explorer service, I want to give my users an effective way to read information by smart contract owners and a place to share ideas and information without using third party services (i.e. Etherscan uses Disqus, and so on)

And in any role, I want a method that does not allow scams - transactions without values, no smart contract's address to remember or to fake - and it does not allow spam - it's cheap but not free, and even if you can link/refer other accounts, you cannot send them messages directly, and others must explicitly follow and listen to your transactions if they want to read your messages.

Main advantages:

I know, My Own Messages (MOM) sounds like mom. And yes, pun intended :)

Specification

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 when, and only when, they appear in all capitals as shown here.

Clients following MOM standard MUST allow users to send and to read MOM transaction, creating an updated message list for each address the users are interested in.

Reading MOM transactions, MOM clients MUST be able to show the current and updated message list, and they SHOULD be able to show also all the message history if users ask for it.

Apart from message list, MOM clients SHOULD be able to download the content of the messages and to show them to the user.

Clients SHOULD allow users to choose and set the source to download content from, and they SHOULD be able to use common Content Addressable Networks - i.e. IPFS or Swarm - or HTTP servers. If content is downloaded from HTTP servers, clients MUST check the content against the declared multihash.

As the default setting, clients MUST consider text/markdown (RFC 7763) as the media type of the content represented by a multihash, and in particular Markdown text in UTF-8 without BOM.

Clients MAY let users choose to parse messages considering other content types. In this case they SHOULD cast a warning to users stating that a content type other than text/markdown is used while processing messages.

It's RECOMMENDED that clients inform users about the actual setting of the default content type.

MOM transactions

Clients MUST assume that invalid MOM transactions don't exist. If a transaction does not strictly follow the MOM standard, clients MUST ignore it and they MUST NOT consider it a MOM transaction at all.

Because there can be security implications parsing data sent by users, clients SHOULD NOT try to keep track or interpret transactions as invalid MOM transactions.

Valid MOM transaction's data structure

ATTRIBUTE VALUE
to MUST be the same account signing the transaction.
value MUST be 0 wei.
data MUST be at least 2 bytes. The first byte MUST be operational code and following bytes MUST be based on the operational codes listed below.

List of supported operations and messages

Each operational code has one or more parameters, and all parameters MUST be considered mandatory.

Optional parameters don't exist: if parameters for the specific operational code are not all present or they don't follow the rules, clients MUST ignore the transaction completely.

Messages MUST be always referenced with the multihash of their content.

Operations are divided into two sets: CORE and EXTENDED operations.

Core operations

OPERATION CODE PARAMETERS MEANING EFFECT
ADD 0x00 multihash Add a message. The parameter MUST be the multihash of the message. Clients MUST add the message to the message list of the sender.
UPDATE 0x01 multihash, multihash Update a message. The first parameter MUST be the multihash of the message to be updated. The second parameter MUST be the multihash of the updated message. Clients MUST update the message list to show the updated message.
REPLY 0x02 multihash, multihash Reply to a message. The first parameter MUST be the multihash of the message to reply to. The second parameter MUST the multihash of the message. Clients MUST insert a new message in the message list and they MUST preserve the relationship with the referenced message.
DELETE 0x03 multihash Delete a message. The parameter MUST be the multihash of the message to delete. Clients MUST remove the message from the message list.
CLOSE ACCOUNT 0xFD multihash Close an account. The parameter MUST be the multihash of the message with the motivations for closing the account. Clients MUST add the message with motivations to the message list and they MUST NOT consider MOM messages sent by that address to be valid anymore, ever. In other words, MOM clients MUST ignore any other transaction sent by that address while creating the message list. This is useful when users want to change account, for example because the private key seems compromised.
RAW 0xFF any The parameter MUST be at least 1 byte. Content type is not disclosed and it MUST NOT be considered as text/markdown. Clients MUST add the message to the message list but they MUST NOT try to decode the content. Clients SHOULD allow users to see this message only if explicitly asked for. This operation can be used for blind notarization that general client can ignore.

Note about DELETE operational code

Please note that sending a DELETE command users are not asking to actually delete anything from the blockchain, they are just asking clients to hide that specific message because it's not valid anymore for some reasons. You can think of it like if users say: I changed my mind so please ÐApps don't show this anymore. As already stated in the specifications above, clients MUST follow this request by the author, unless expressly asked otherwise by the user.

Please also note that, because it's usually up to the author of a message to be sure the content is available to everyone, if a DELETE message was sent it's very likely the content referenced by the multihash isn't available anymore, simply because probably it's not shared by anyone.

Extended operations

OPERATION CODE PARAMETERS MEANING EFFECT
ADD & REFER 0x04 multihash, address Add a message and refer an account. The first parameter MUST be the multihash of the message. The second parameter MUST be an address referenced by the message. Clients MUST add the message to the message list and they MUST track the reference to the specified account. This can be useful to invite the owner of the referenced account to read this specific message.
UPDATE & REFER 0x05 multihash, multihash, address Update a message. The first parameter MUST be the multihash of the message to be updated. The second parameter MUST be the multihash of the updated message. The third parameter MUST be an address referenced by the message. Clients MUST update the message list to show the updated message and they MUST track the reference to the specified account. This can be useful to invite the owner of the referenced account to read this specific message.
ENDORSE 0x06 multihash Endorse a message identified by the specified multihash. The parameter MUST be the multihash of the message to be endorsed. Clients MUST record and track the endorsement for that specific message. Think it as a like, a retwitt, etc.
REMOVE ENDORSEMENT 0x07 multihash Remove endorsement to the message identified by the specified multihash. The parameter MUST be the multihash of the message. Clients MUST remove the endorsement for that specific message.
DISAPPROVE 0x08 multihash Disapprove a message identified by the specified multihash. The parameter MUST be the multihash of the message to disapprove. Clients MUST record and track the disapproval for that specific message. Think it as a I don't like it.
REMOVE DISAPPROVAL 0x09 multihash Remove disapproval of a message identified by the specified multihash. The parameter MUST be the multihash of the message. Clients MUST remove the disapproval for that specific message.
ENDORSE & REPLY 0x0A multihash, multihash Endorse a message and reply to it. The first parameter MUST be the multihash of the message to reply to. The second parameter MUST be the multihash of the message. Clients MUST insert a new message in the message list and they MUST preserve the relationship with the referenced message. Clients MUST also record and track the endorsement for that specific message.
DISAPPROVE & REPLY 0x0B multihash, multihash Disapprove a message and reply to it. The first parameter MUST be the multihash of the message to reply to. The second parameter MUST be the multihash of the message. Clients MUST insert a new message in the message list and they MUST preserve the relationship with the referenced message. Clients MUST also record and track the disapproval for that specific message.

Rationale

Ethereum is account based, so it's good to be identified as a single source of information.

It is also able of doing notarization very well and to impose some restrictions on transaction's structure, so it's good for commands.

IPFS, Swarm or other CANs (Content Addressable Networks) or storage methods are good to store a lot of information. So, the union of both worlds it's a good solution to achieve the objectives of this message standard.

The objective is also to avoid in the first place any kind of scam and malicious behaviors, so MOM don't allow to send transactions to other accounts and the value of a MOM transaction is always 0.

Why not using a smart contract?

MOM wants to be useful, easy to implement and read, error proof, fast and cheap, but:

Without a specific smart contract to rely on, the MOM standard can be implemented and used right now in any existing networks, and even in future ones.

Finally, if you can achieve exactly the same result without a smart contract, you didn't need a smart contract at the first place.

Why not storing messages directly on-chain?

There's no benefit to store static messages on-chain, if they are not related to some smart contract's state or if they don't represent exchange of value. The cost of storing data on-chain is also very high.

Why not storing op codes inside the message?

While cost effectiveness is a very important feature in a blockchain related standard, there's also a compromise to reach with usability and usefulness.

Storing commands inside the messages forces the client to actually download messages to understand what to do with them. This is very inefficient, bandwidth and time consuming.

Being able to see the commands before downloading the content, it allows the client to recreate the history of all messages and then, at the end, download only updated messages.

Creating a structure for the content of the messages leads to many issues and considerations in parsing the content, if it's correct, misspelled, and so on.

Finally, the content must remain clean. You really want to notarize the content and not to refer to a data structure, because this can lead to possible false-negative when checking if a content is the same of another.

Why multihash?

Multihash is flexible, future-proof and there are already tons of library supporting it. Ethereum must be easily integrable with many different platforms and architectures, so MOM standard follows that idea.

Backwards Compatibility

You can already find few transactions over the Ethereum network that use a pattern similar to this EIP. Sometimes it's done to invalidate a previous transaction in memory pool, using the same nonce but with more gas price, so that transaction is mined cancelling the previous one still in the memory pool. This kind of transactions can be easily ignored if created before the approval of this EIP or just checking if the payload follows the correct syntax.

Test Cases

A MOM-compliant client can be found and tested on GitHub.

You can use the latest version of MOM client directly via GitHub Pages or via IPFS (see the client repo for the latest updated address).

Implementation

You can use an already working MOM JavaScript package on GitHub Packages or npmjs. The package is already used by the MOM client above, and you can use it in your ÐApps too with:

npm install @internetofpeers/mom-js

Transaction 0x8e49485c56897757a6f2707b92cd5dad06126afed92261b9fe1a19b110bc34e6 is an example of a valid MOM transaction already mined on the Main net; it's an ADD message.

Security Considerations

MOM is very simple and it has no real security concerns by itself. The standard already considers valid only transactions with 0 value and where from and to addresses are equals.

The only concerns can come from the payload, but it is more related to the client and not to the standard itself, so here you can find some security suggestions related to clients implementing the standard.

Parsing commands

MOM standard involves parsing payloads generated by potentially malicious clients, so attention must be made to avoid unwanted code execution.

Messages

Default content-type of a message following the MOM standard is Markdown text in UTF8 without BOM. It is highly recommended to disallow the reading of any not-text content-type, unless expressly acknowledged by the user.

Because content multihash is always stored into the chain, clients can download that content from Content Addressable Network (like IPFS or Swarm) or from central servers. In the latter case, a client should always check the integrity of the received messages, or it must warn the user if it cannot do that (feature not implemented or in error).

Copyright

Copyright and related rights waived via CC0.