Today (July 22nd) is the mint day of @OnChainAllStars, a fully on-chain collection that we are releasing with @0x8BitArt, exclusively on @opensea Time to share more technical details in the thread bellow:

GM. The narrative you’re being sold today will change over time. Myth: @v1punks are just cheaper versions of the real thing. Reality: In my professional opinion, drawing from my collecting experience, the long-term outlook for V2s may face challenges🧵

#AuctionResult Lot 83 'Stagnant Elixir's Sweet' by Obvious in our July First Open online auction exceeded its high estimate, realizing $35,280! Congratulations to all parties and artists involved! @obv_ious

Off-Chain Art auction has been finalized an acquired by @thearchivist! Thank you ❤️ I'm now working on a simple frontend and the standard for everyone to use this framework

1/ On UNP (Unified NFT Physics) Or: “So Few NFTs, So Many NFTs” wif Math We will go with a single post from the beginning this time, the way the big boss now wants us to do it I am happy with the spirit of the discussion we had this week about how few good NFTs there are vs fungible tokens. I am less happy about our attempts to categorize them. They are low resolution, incomplete and do not capture disagreements well. Today I will propose a unified mathematical framework that I think can capture the full breadth of anyone's views about NFTs (or art) in a way that is transparent, comparability and supportive of enriching discussions, not reductive ones. 2/ NFT Collectors Do Know Some Things Serious NFT collectors have a quite sophisticated understanding of NFT importance at any given point in time, how it changes over time, and how different collections serve different purposes. Even when we disagree among ourselves about any specific piece or any specific collection, it is a limited disagreement within a broader agreed framework. It is not like we are living in universes with different laws of physics. 3/ Spreadsheets: Good and Bad The spreadsheets that people are making are good first attempts but they are at least 2 orders of magnitude lower resolution than how I think about it. So very logically people are saying “wait, what about this and what about that and you forgot this and it is not that simple” and obviously that is all true. And the spreadsheet does not have a straightforward / computable way to express disagreements within a broader agreed physics. 4/ All Concepts In One Framework So, I did a little thinking yesterday about how we can integrate all the concepts into a unified physics where you can disagree about any specific value but do not disagree about how to do math. The Concepts: a) Collection power laws b) Individual NFT power laws c) Fractal power laws (global, national/regional/industry, local) See here: nitter.app/punk6529/status/151833… d) 1/1s vs Editions e) PFPs vs Generative vs 1/1s of various types vs even utility and fun NFTs. They all fit in here just fine. f) Works for any chain The idea is that we can argue about values, but we do not argue about the physics. 5/ Objections! Let me also deal with all the objections upfront: a) People will disagree about which NFTs are important and which collections and artists are important! True! Of course people will disagree. I disagree with myself half the time! The idea here is to allow you to express what you believe in a way that is legible to someone else. b) The values will change over time! No kidding, everything in the whole world changes over time. Why would NFTs be an exception? c) Not everything can be quantified! True! But a lot of things can be roughly quantified. d) Art is not about lists, come on, this is so crypto/tech bro stuff! Well, yes and no. I don’t have lists for everything, there are many things I can’t quantify about my art feelings. But if anyone tells you they have no actual preferences, no ranking, everything is a random function, including how they feel about trad art and NFTs, well that is a lie. 6/ Which One Would I Pick The way I think about it is very simple. I do pairwise selection. I think about two NFTs and think which “one I would prefer to see in my wallet”. The reasons that I prefer one may be purely qualitative, emotional, ineffable, but I obviously have preferences. Everyone has preferences. Some people collect Monet, some people collect Picasso. That is a preference. Whether you write it down or not, it is an objective fact about the world that you either have Picassos or Monets on your wall or Rare Pepes or Cath Simards in your wallet. So the goal here is not to replace our views on what you like but make them more legible, more comparable, allow us to express our agreements and disagreements in a more precise way than Pepe Bad, Aversano Good! 7/ First Draft Consider this a first draft and then we can plug some actual figures into it and see how it works. Some of the formulas need values for the parameters to determine the slope of the various curves. There will be some trial and error to figure those out, but I think they are broadly right. Don’t be intimidated by the math formulas. They look scarier than they are. Once we get to the point that they are final, we will explain how they work and make tools. Also, X does not support LaTex so the formulas look ugly written out. 8/ Two-Layer Scores We need to have two-layer scores – One for Collections and One for Individual NFTs within collections. Collection-Level Score Let C_Cj(t) be the Importance Consensus of collection Cj at time t, constrained to the range [0,1]. A high C_Cj(t) close to 1 means near-universal agreement that this collection is historically/culturally “important/blue-chip/socially constructed” A new or unknown project might have C_Cj(t) around 0.01. The ungrouped 1/1s of a specific artist can also be considered grouped as a single Collection for this purpose NFT (Token-Level) Score Within a Collection Within collection Cj, each token n has a local “importance” factor r_n(t), also in [0,1]. For a typical “floor” piece, r_n(t) might be about 0.2. For a best piece in the collection, r_n(t) might be > 0.9. 9/ Overall “Importance” In plain English: - for a 1/1 or 1/1/X, we multiply Collection Score x Token Score So if a collection has a collection score of 0.5 and the Token has an in-collection score of 0.3, the token score is 0.15. - for a 1/X (edition) we reduce the score but not linearly because I don’t think it works that way. The reduction in importance for editions is much less harsh than linear and this is best described with an example. Imagine that I think that the Rare Pepes have a Collection Score of 0.95 and the Nakamoto Pepe has an in-collection score of 0.98. This would mean that Nakamoto Pepe has a score of 0.931 before accounting for edition size. Given the Naka has an edition size of 300, if we divided equally, the score for each individual edition token would be 0.931/300 = 0.003103. This is obviously wrong. 0.003103 is a very low score and individual Naka editions are very valuable and desirable and, yes, important. The score should be something like 0.7 instead. 10/ Overall Importance (Formulas) Same thing with a bit of math - define each NFT’s total score as: C_tilde(n, t) = C_Collection(n)(t) * r_n(t) * E_factor(n) where: • C_Collection(n)(t) is the collection’s brand score in [0,1]. • r_n(t) is the token’s local rarity in [0,1]. • E_factor(n) = 1 / [1 + alpha * ln( edition_size(n) )]. Explanation of E_factor(n): • If the edition_size(n) = 1 (a true 1/1), then ln(1) = 0, so E_factor(n) = 1 (no dilution). • If the edition_size(n) is large (e.g., 300), E_factor(n) = 1 / [1 + alpha * ln(300)]. The parameter alpha > 0 controls how steeply the factor drops for bigger editions (can be tweaked for a larger or smaller penalty The alpha parameter determines the penalty. For the Naka in my example to be 0.7, the alpha parameter should be 0.058. I think the right answer for editions is in that range – I need to plug in a few more to see. 11/ Infinite Minting but Power-Laws Hold Obviously there is no limit on how many collections and how many NFTs will be minted Let M_coll(t) be the number of collections minted by time t. Let M_token(t) be the total NFT tokens minted by time t. As t goes to infinity, both can grow arbitrarily large. 12/ Distribution Assumptions (Power-Law Tail) Collection Distribution: C_Cj(t) is distributed according to some heavy-tailed distribution F_{C_t}(x), x in [0,1]. In plain English: “Most collections have low Importance Consensus, but a tiny fraction approach household-name status.” Token Rarity Distribution (within a collection): r_n(t) is distributed according to G_{Cj,t}(r), r in [0,1]. Typically also skewed, though often less extreme than the collection-level distribution. This can be analyzed per collection because it is different in one collection vs the other. Punks absolutely have highly skewed distribution. Incomplete Controls have almost no skew at all. You can think of collection-level skew as: “how happy are you to be allowed to pick your specific NFT from a collection vs receiving a randomly selected one”. Overall Score C_tilde(n, t) = C_{Collection(n)}(t) * r_n(t): When the product (multiplication) is near 1, that event is extremely rare (a remarkable NFT from a remarkable collection or remarkable artist). 13/ Local–National–Global Tiers I believe only 5 to 50 NFT artists or collections or pieces will become truly global memes will make it into the global pantheon of cultural memes (there are only a xxx to low xxxx truly global memes) but thousands / tens of thousands / hundreds of thousands will make it up to the national, local, niche level. This is represented by thresholds. tau_local(t): threshold for “local/niche success.” tau_national(t): threshold for “regional/national brand.” tau_global(t) ~ 0.95 or 0.99: near-universal brand recognition. At time t, an NFT is in the global pantheon if: C_tilde(n, t) > tau_global(t). Likewise for local/national thresholds. Because of the power-law tail, the probability of crossing something like 0.95 is extremely small, so only a handful of tokens or collections achieve it. Or in plain English, we have tiers for the scores. We can also have category tiers as some categories may be less developed and have lower scores. So the most important gaming NFT will be less important than most important art NFT but would still be interesting to now. 14/ Time & Lindy-Like Stickiness Lindy’s Law: The Longer Something Survives, The Longer It Will Continue To Survive This absolutely applies to NFT collections. The more years an NFT collection remains recognized at a high tier, the more years it is likely to remain at that tier. This is true for chains, it is true for tokens, it is true for many things in life. The longer it survives, the longer its future survival will be. We can model this via a survival probability that increases with “time in tier.” 15/ Time in Tier Formulas Time in Tier, Delta_tier(n, t) Define Delta_tier(n, t) as the continuous amount of time that NFT n has alreadyspent above threshold tau_tier. For example: Delta_global(n, t) = integral from s = t0 to s = t of [ 1 if C_tilde(n, s) > tau_global else 0 ] ds, where t0 is the time it first crossed into that tier. In plain English, how long has it been at a certain score. Hazard Function / Stability Once you are in a tier, define a “hazard rate” for dropping out: lambda_tier(Delta) = lambda_0 / (1 + alpha(hazard) * Delta), where lambda_0 and alpha are positive constants. This means the longer you have been in that tier (the bigger Delta is), the lower the hazard rate. In simpler terms: “If a collection has already spent 2 years as top-tier, it’s more likely it’ll remain top-tier for at least 2 more.” This is the fair way to express that Rare Pepes or Punks are more Lindy than any new collection, but any new collection might emerge and if it also stays strong over time, it becomes Lindy too. It also is a way to account for the pump-y flash-in-the-pan nature of most crypto assets, including most NFTs. They won’t spend much time in tier. 16/ The Concepts Again Infinite Supply: Both collection-level and token-level minting are unbounded. Two-Layer Score: Each NFT’s “importance” = (collection brand) * (local rarity). Getting near 1 on both factors is very unlikely, with a reducer penalty for the individual tokens of editions. Local–National–Global Tiers: We have thresholds tau_local < tau_national < tau_global. The measure of NFTs that cross tau_global ~ 0.99 is infinitesimal, explaining the “5 to 50 global memes,” “3,000 apex tokens,” etc. Lindy Stickiness: Once a token or collection stays in a high tier for months/years, the hazard of losing that status declines. The set of “Impossibly Rare” or “Global Pantheon” pieces gets increasingly locked in if they can keep their cultural relevance Slow Supply Expansion at the Apex: New good collections do appear, but the road to the highest tiers is tough, and the road to stay in the highest tiers is tougher (needs time). Consumer vs. Apex: Meanwhile, billions of “Fun NFTs” never approach tau_global. They function like fan merch or ephemeral mints — great use cases, but not “global culture-grade” assets. There is no hard breaks, just a continuous expansion of collections and individual tokens as you go to the “lower” levels. 17/ Equation Summary The equations in one place Collection Score Distribution (heavy tail near 1): P( C_C(t) > x ) ~ k * (1 - x)^alpha, with alpha > 1. Token Local Score: r_n(t) ~ G_{C,t} (Likewise heavy-tailed but usually less extreme than the above.) Overall Score: C_tilde(n, t) = C_{Collection(n)}(t) * r_n(t). (Near 1 is super rare.) Edition Factor: E_factor(n) = 1 / [1 + alpha * ln( edition_size(n) )] (This penalizes large editions but not as severely as dividing by the full edition size.) Tier Threshold: Suppose tau_global(t) is about 0.99. Then the expected number of tokens crossing that threshold is: M_token(t) * P( C_tilde > 0.99 ). Lindy Hazard: lambda_global(Delta) = lambda_0 / (1 + alpha * Delta), (The older your apex status, the less likely you lose it.) 18/ Next Steps This is a much more accurate mathematical expression of how I think people think about NFTs. What are next steps: a) I want to read reactions, comments, ideas and sleep on them for a few days b) I will write this up formally in a “white paper” that is implementable by anyone who wants to do this c) I have already spoken to someone who may be able to do a proof of concept for us to play with In a few days, we can play around with something specific. Look forward to everyone’s thoughts

Following recent discussions by @punk6529 , @batsoupyum and @CozomoMedici , this just came to mind:

My strategic crypto reserve. What's yours?

1/ On Fiat Culture NFTs are the scarcest cultural objects ever made, the only cultural objects with pristine provenance. They are not BTC, but they will behave over time in ways that emotionally parallel BTC for the $100T in global intangibles.

"Hope's Horn" just join my personal collection. This baby triceratops horn (~67Myo) shows healing from a large pathology caused by an adult T-Rex's bite, indicating 'Hope' survived an encounter with Earth's most dangerous predator to date—an exceptionally rare testament.

If you don’t think the jpeg on the left is infinitely more appealing than the jpeg on the right then you are truly lost, there is no hope

AI Artist Robbie Barrat And Painter Ronan Barrot Collaborate On “Infinite Skulls” @DrBeef_ artnome.com/news/2019/1/22/a…