Re: Factor

While looking into the multibase group of self-identifying base encodings, I discovered Base256Emoji which is an encoding format described by an emoji to use for each byte of an input buffer. This spec is implemented, for example, in both Go and Rust

Despite replacing each byte with a typically 3-byte or 4-byte UTF-8 sequence – which is unusual for byte encodings (they often seek to reduce the length of an input sequence) – there are some nice use cases.

We’re going to implement this in Factor and then discuss some variants.

Implementation

First, we define a 256 item sequence of emojis, one for each byte:

CONSTANT: base256>emoji "🚀🪐☄🛰🌌🌑🌒🌓🌔🌕🌖🌗🌘🌍🌏🌎🐉☀💻🖥\
💾💿😂❤😍🤣😊🙏💕😭😘👍😅👏😁🔥🥰💔💖💙😢🤔😆🙄💪😉☺👌🤗💜😔😎😇\
🌹🤦🎉💞✌✨🤷😱😌🌸🙌😋💗💚😏💛🙂💓🤩😄😀🖤😃💯🙈👇🎶😒🤭❣😜💋\
👀😪😑💥🙋😞😩😡🤪👊🥳😥🤤👉💃😳✋😚😝😴🌟😬🙃🍀🌷😻😓⭐✅🥺🌈😈\
🤘💦✔😣🏃💐☹🎊💘😠☝😕🌺🎂🌻😐🖕💝🙊😹🗣💫💀👑🎵🤞😛🔴😤🌼😫⚽🤙\
☕🏆🤫👈😮🙆🍻🍃🐶💁😲🌿🧡🎁⚡🌞🎈❌✊👋😰🤨😶🤝🚶💰🍓💢🤟🙁🚨💨\
🤬✈🎀🍺🤓😙💟🌱😖👶🥴▶➡❓💎💸⬇😨🌚🦋😷🕺⚠🙅😟😵👎🤲🤠🤧📌🔵💅🧐\
🐾🍒😗🤑🌊🤯🐷☎💧😯💆👆🎤🙇🍑❄🌴💣🐸💌📍🥀🤢👅💡💩👐📸👻🤐🤮🎼🥵\
🚩🍎🍊👼💍📣🥂"

And then we compute the reverse mapping:

CONSTANT: emoji>base256 $[ base256>emoji H{ } zip-index-as ]

With those two data blocks, we can define words to convert into and out of base256emoji:

: >base256emoji ( bytes -- str )
    [ base256>emoji nth ] "" map-as ;

: base256emoji> ( str -- bytes )
    [ emoji>base256 at ] B{ } map-as ;

You can try it out:

IN: scratchpad "Hello, Factor!" >base256emoji .
"😄✋🍀🍀😓💪😅💓🤤💃😈😓🥺👏"

IN: scratchpad "😄✋🍀🍀😓💪😅💓🤤💃😈😓🥺👏" base256emoji> "" like .
"Hello, Factor!"

Use Cases

The most interesting use case for base256emoji seems to be the ERC-7673: Distinguishable base256emoji Addresses proposal for Ethereum. This proposal seeks to “address spoofing attacks [that] have mislead tens of thousands of ether, and countless other tokens” by using visual emoji-based strings – a similar justification to the Drunken Bishop algorithm.

We can try base256emoji out with checksums to yield a similar benefit, displaying 16 emojis instead of a 32-byte hex-string:

IN: scratchpad "resource:license.txt" md5 checksum-bytes
               bytes>hex-string .
"ebb5ab617e3a88ed43f7d247c6466d95"

IN: scratchpad "resource:license.txt" md5 checksum-bytes
               >base256emoji .
"💌💨🤨🤤😠✨😹🥀😏🎼🤠🤩⬇💓🌷🤙"

Visual Collisions

Given a desire to use visually dissimilar emojis in identities, it would be useful to think about the chosen emoji set and how that might be interpreted in a visual context. Some criticism of this particular group of emojis, which are sometimes magnified by smaller font sizes, might focus on the visual similarity of:

1. Similar “globe” emojis (🌍 🌏 🌎)
2. Similar “face” emojis (🙂 😐 😑 🙁)
3. Similar “kiss” emojis (😙 😚 😗 😘)
4. Similar “star” emojis (⭐ 🌟)
5. Similar “grin” emojis (😀 😃 😄 😁 😆 😅)
6. Similar “heart” emojis (💔 💗 💕 💖 💘 💙 💜 💚 💛 🖤)
7. Similar “hand” emojis (🤞 👋 ✋ 👊 ✊ 🤝 🤲)

It might be desirable to choose emojis not based on community membership or common usage, but on their most dissimilar visual identity. to make it even harder for scammers to deliberately pick lookalike emojis and rely on small text sizes, platform font differences, or user inattention.

A couple of other emoji sets can be found, for example @Equim-chan/base256 (which uses one of my favorite emojis: 👾), npm/base-emoji, or even the KittenMoji crate.

Fun!

This is available on my GitHub.