More unsig rarity

Using Chat GPT’s newest features of data analysis, I’ve been making some headway on determining rarity for an NFT project that is notoriously difficult.

unsigned_algorithms are an NFT project on the Cardano blockchain.

Said differently, they are the first ‘secured object’ art project that’s fully self contained on the blockchain where it can never be deleted or forged.

Of the 31,119 objects there is a first one, named unsig00000 that contains an algorithm. Then, Properties are contained in the individual unsig that come after number 00000. There are from one to six Properties possible. unsig with one property are the fewest.

Each Property is in four parts: 1 | Red | CDF | 180

1 is a multiplier, Red is one of Red green or Blue, DCF is the distribution, and 180 is rotation of that layer by180º.

In this way, regular patterned, sometimes intricate - images are produced.

unsig00236 a two-property unsig: 2 | Red | CDF | 0 and 2 | Blue | CDF | 0

This unsig having a multiplier of 2 and rotation of 0 creates a second line in the middle and the layered colors blend from black to cyan.

Here’s one with six properties:

unsig30462 a six property unsig

Properties:
1 | Red | CDF | 90
1 | Red | CDF | 180
1 | Blue | CDF | 0
1 | Blue | CDF | 90
2 | Green | CDF | 180
4 | Green | CDF | 90

When the same color and distribution are at 0º and 90º it makes a clean ‘horizontal’ diagonal line here with blue. The same thing happens with Red at 90º and 180º making a ‘vertical’ diagonal.

The curves are created when multiples of rotations do not overlap.

RARITY

So, how can we assess rarity for such a large NFT collection with so many possible combinations of parameters? Luckily vector databases and AI have emerged.

For a few weeks, I have been trying to discover rarity ranking using GPT 3.5 and 4, then something great happened… November 2023 saw the introduction of advanced data capabilities and features. Diving right in, I asked GPT to help me get a vector database up and running.

Formerly, there was an obvious rarity measure; how many Properties.
This chart shows the number of Properties and how many unsig have that many.

0 property: 1
1 property: 15
2 properties: 246
3 properties: 2,164
4 properties: 9,553
6 properties: 4,565
5 properties: 14,575

Then we get into how often each Property appears in the whole.

A Property can even repeat within the same unsig. As an example, a two Property pattern could occur three times in a six Property unsig. Also each individual unique Property will occur only so many times in the whole dataset, making another layer of rarity.

Then, when we look at the shapes created by the various layers, we see that patterns persist through color changes and create some ‘named’ types like “windmills” shown below.

unsig12251 a very nice “windmill”

There are hundreds of possible Forms but there will not be a growing number of forms because there are only so many parameters and many rotations add up to the same shape.

I’ve been working to put the data into a graph database and discover unknown possible relations between groupings of unsig.