As the leading independent NFT marketplace on the Theta blockchain, OpenTheta has been searching for the best rarity ranking system our users can rely on when trading NFTs. Should we develop our own? Or is there a standardized system being adopted across many blockchains that provides unbiased, transparent, and consistent ranking of one NFT’s attributes compared to another in the same collection?
What is OpenRarity?
OpenRarity is a collaboration between Curio, icy.tools, OpenSea, and PROOF that aims to provide a transparent, mathematically sound, and objective rarity calculation that is entirely open-source and reproducible by anyone. It is currently being used by some of the biggest NFT marketplaces in the crypto space, such as OpenSea, Coinbase, LooksRare, and Rarible.
What sets OpenRarity apart is its use of the Surprisal Ranking Algorithm, which is based on Information Content. This method expresses probability in terms of surprise, providing a highly accurate and consistent rarity ranking across all platforms.
At OpenTheta, we believe in the importance of transparency and consistency in building trust in the NFT industry. That’s why we have decided to integrate OpenRarity into our website, providing our users with a reliable and unbiased way to evaluate the rarity of NFTs.
Let’s briefly explore the inner workings of the OpenRarity algorithm’s mathematically-grounded methodology, known as the Surprisal Ranking Algorithm and compare it to that of other ranking systems.
Surprisal Ranking Algorithm with Double Sort
OpenRarity’s Surprisal Ranking Algorithm uses a concept called Information Content. IC is an alternative way of expressing probabilities taking into account how surprised someone would be upon discovering a trait combination. This method has several advantages over other ranking systems, including that it is valid to perform linear operations on information, but not on raw probabilities, making it perfectly suited for assessing rarity.
What about the custom 1/1 NFTs that are part of so many Theta NFT collections? Won’t these score much lower than those NFTs with fewer unique attributes? Fortunately, OpenRarity has implemented Double Sort which performs two scorings to inform the rank: Information Content and Unique Trait Count. The ranking is then performed by first sorting by unique trait count and then by Information Content. This method works on the basis that 1 of 1s are more rare than non-1 of 1s as expected by the broader community yet does not compromise the mathematical grounding of the Information Content calculation.
However, the current algorithm does not consider all characteristics of an NFT in its calculations. Specifically, attributes that have numerical values are disregarded and as a result, do not contribute any information to the final ranking.
Comparing Rarity Algorithms
Comparing ThetaPunk #5949 and ThetaPunk #7523
Both rank third but in different rankings. The ThetaPunk #5949 ranks third by using the Surprisal Ranking Algorithm from OpenRarity and ThetaPunk #7523 ranks third in the legacy ranking system. Let’s look at the attributes and compare.
When examining ThetaPunk #7523, which is rank 3 in the legacy ranking, we can see that the main focus for ranking this Punk highly was because of its Type (Alien). Besides this attribute, this NFT also has a rare Attribute 3 (Medical Mask) and no Skin Tone.
However, when looking at ThetaPunk #5949, we can directly discover that Attribute 5 and Attribute 6 are each much rarer than the Type of Punk #7523 (Alien). Additionally, ThetaPunk #5949 has three other Attributes that it shares with less than 1% of other Punks.
This comparison highlights the different ways these two system’s algorithms approach NFT ranking. The legacy ranking system looks at specific Attributes and weights them differently then the OpenRarity ranking, which looks at how surprised someone would be upon discovering a trait combination.
It is important to note that OpenRarity can only accurately calculate a representative ranking if all the information needed is stored in the attributes of each NFT. This empowers creators as they are ultimately responsible for the on-chain metadata used for these rarity calculations.
It is important to note that OpenRarity can only accurately calculate a representative ranking if all the information needed is stored in the attributes of each NFT. This empowers creators as they are ultimately responsible for the on-chain metadata used for these rarity calculations.
Rarity on OpenTheta
On OpenTheta, eligible collections will feature a numerical rarity ranking on both the item and collection pages. To easily find the rarest NFTs in a collection, you can sort by rarity.
Each item will display its corresponding rank on the top right. It’s important to note that OpenRarity rankings directly reflect the creator-published attributes in the metadata. So for collections that utilize dynamic NFTs or are not yet fully minted, the ranking is simply a snapshot of the current state and subject to change.
Conclusion
In summary, there are many different algorithms to rank NFTs. OpenRarity’s Surprisal Ranking Algorithm stands out as an open-source and reproducible system being adapted across the space. By adding OpenRarity to our platform we are joining many other successful NFT marketplaces, such as OpenSea, Rarible and LooksRare!
We are excited to be a part of the OpenRarity community and contribute to the growth, development, and standardization of the NFT industry. By providing accurate and transparent rarity rankings on the Theta blockchain, we hope to empower our users to make informed trading decisions.
