Pre-AI Is a Date. Not a Halo.

Justin Trimble posted the thesis the timeline wanted:Pre-AI model algorithmic art on Art Blocks will age like fine wine because the skill was verifiably harder and it is timestamped.Pindar Van Arman answered with the cultural split: there will be a pre/post-model worldview. He cannot read new non-fiction anymore. Everything sounds too AI.Someone else sharpened it: fully on-chain pre-AI algorithmic art, timestamped, inspectable. Oraand on a Commodore 64 emulator stored on Ethereum in 2022.The argument is useful. The romance is sloppy. That is the opening.

What the tweet gets right

Periodization is coming. Collectors will ask under what constraint a work was made. That is not nostalgia. That is literacy.A mint in 2021 does not care about your current prompt pack. The chain has a clock. Art Blocks from the pre-diffusion years has code, hash, output, time. You can replay the procedure. That survives fashion.“Timestamped” is the strong word.

What the tweet gets wrong

“Verifiably harder” is the weak word.A good Fidenza is hard because Hobbs authored a system with taste. Not because Midjourney was late to the party. A bad Art Blocks drop from the same season is still a bad drop. The calendar does not ferment mediocre code into wine.Post-model work can be harder, not cheaper: own data, own training, own renderer, own machine on-chain. Difficulty moved. It did not disappear.Van Arman’s fatigue is real. It is not a valuation method. If your only test is “this sounds like a model,” you will punish living writers and overpay dead dates.

Three shelves, not one church

Do not smash these into one slogan.

1. Authored algorithms
Art Blocks, Autoglyphs, long-form systems. The claim is the code.

2. Inspectable machines
Oraand-style work: the object is also a running system. The claim is that you can look inside.

3. Early image hybrids
Photo plus machine before the prompt factory. The claim is source, tool, small supply, first mint.Trimble is talking about shelf one. The C64 example is shelf two. A March 2021 OpenSea mint of 100 photo-mix works is shelf three.Shelf three is not Fidenza. It does not need to be. It needs proof, not a borrowed halo.

The collector question

Same question as last time, new costume.If the speech dies tonight, what remains?For Art Blocks: code you can run. For on-chain toys: a machine you can inspect. For early photo-AI: a mint date plus sources the artist can prove without handing the internet the negatives.

If all that remains is “this was before AI, so it is serious,” you bought a period film. Not a work.

How early image-AI should be disclosed

Show the date on-chain. Show the supply. Name the tool. Show a crop, a corner, a timestamped scrap. Keep the master files private. Hash them.

Produce them for a real provenance check.

Publishing the full source photos is not bravery.

It is giving strangers the raw material.

Provenance is not a download pack.

Copycats can remake a look after 2022 anyway.

They cannot remake your wallet’s mint from March 2021.

The new liturgy to refuse

Pre-AI = authentic.

Post-AI = slop.

That is the same spell as “museum wall = 11 ETH.” A date is origin. Origin is not quality. A custom model trained on one painter’s archive can still be scaled liturgy.

A 2021 algorithm can still be empty. A 2026 system can still be a work.Use the date as a filter. Then look at the object.

If you are new

Do not buy “pre-AI” as a ticker.

Ask which shelf you are on.

Ask what is inspectable.

Ask whether the edition matches the labor.

Ask whether the chorus arrived before the work or after it.Art Blocks will keep some of that wine. Not every bottle. On-chain systems will keep the ones you can open.

Early hybrids will keep the ones whose makers still have the sources and the transaction.The rest is speech.Class dismissed. Watch the wallet, not the era.

Pedro Jose
Stack of Truths

Edited with Grok

(P.S. No financial advice. A timestamp is evidence. It is not a floor price.)