A Design Fiction on Visual-Meta
I’ve been spending time with a lovely set of people on the internet lately. One of the perks of searching for the rest of your life is that you discover tiny niches of people doing groundbreaking work. Frode Hegland is one of them, and he shares that work openly with anyone interested a few times each week. The conversations revolve around the future of text and computing and a swath of multicultural references. It’s delightful.
That’s your preface for a bit of design fiction below. It’s a mini read in the simplest sense — a fictional artifact of the future.
The simple bit of magic referred to is known as Visual-Meta, which exists today from Frode with influence and construction from a host of his discussion partners. It’s not more than a few words at the end of a document, but it enables a host of functions locked within PDFs, just by the way it’s written. The simple ones always catch on, because anyone can understand them. I think anyone can understand Visual-Meta.
It so happens that Vint Cerf, one of the inventors of the internet, is writing a piece regarding the impact Visual-Meta may have on supercomputers of the future, which brought the discussion to life. What would that look like?
Ten million documents, interconnected. What are they? Where have they come from?
So, this is a playful dreaming on Vint's prompt. It's floaty, a touch unrefined, and surely took a few turns I didn't expect. So, have fun with it. Then learn about Visual-Meta.
Excerpt from Personal Computing in the 21st Century
It started with academia. They aligned on a protocol to make documents accessible and computational. The first day in 2021 — when ACM signed off on Visual-Meta — was a win for interoperability fans everywhere. Things opened quickly from there. Tens of thousands of documents were annotated with Visual-Meta's signature, a short set of text that doubles as computational metadata.
The academics flocked first. Researchers copied sentences directly from PDFs and citations autogenerated alongside. "Woah, did you see that?" The best technology always feels like magic the first time. Professors constructed trails through hundreds of texts for student exploration — all from source data. Students wrote their theses, finding hidden insights from the related texts they'd come across. Publishing accelerated. The mid 2020's saw twice the rate of published papers in the HCI and IS fields.
Because the tools were built into Author and Reader, new users flooded in as well, at least until every PDF reader had embedded Visual-Meta into its software. That lasted for a few years, until the browsers had no choice but to adopt the same framework. Websites were next. The people had spoken. They refused to live without the benefits of an interconnected web. In late 2025, a million sites were online. A year later, ten million. Were we headed to Ted Nelson's delight?
The real growth boomed from VMQL — a query language that worked inside and alongside the text documents. Consumers had Google, but thinkers quickly realized the combination of Visual-Meta and Roam was giving way to a new economy. A reasonably smart human from anywhere on the planet could find whatever cutting edge piece of insight they wanted, backed by the research sourced alongside. Within a few keystrokes, they could start writing a sentence, hit cmd+M, and a typeahead search generated the perfect words to complete their synthesis. The way it worked people thought it was magic, like acing a fill-in-the-blank exam. We saw another rush of publishing.
We can't quite compare the before and after numbers, because publishing went micro shortly after. Our reasonably smart human was rewarded for sharing their new synthesis, even if it was just a paragraph. Whenever someone sourced that material, our friend was rewarded with a few tokens. The internet was still free, but entering the realm of publishing — which nearly everyone chose to do — allowed the insightful to generate a new revenue stream. Everyone knew something unique that they could share.
When any phrase can be added to another, chaos is bound to ensue. Anyone seeing a Google search autocomplete a phrase can tell you that much. Where VMQL thrived was the frame it could put on any search. Backed by a block-base within each document, PDFs (and later, webpages) shrunk down to movable parts. As a primitive design of nearly every information tool in the mid-21st century, writing ideas often started at the atomic source level and quickly became new construction. The advances in Natural Language Processing and computing power made predictions backed by academic research exceptionally generative. Distributed economics enabled anyone in the world to earn tokens for their thoughts.
Golf tournaments in the early 2000s often showed the number of golfers using Titleist golf balls. Their logo would sit alongside the golfer's name on the leaderboard. It was rare to see a name that didn't accompany Titleist's lovely cursive font. By 2030, Visual-Meta owned the same stronghold on source data. Near everyone used the standard. It made everything easier, and many were making money sharing their own insights. Some might say it enabled the era of building blocks that we're in today.
It's rather remarkable that within 10 years the shift to computational text had flipped entirely. A writer in the year 2020 wouldn't recognize the tools used to write in 2030. Their process would be of another era. Visual-Meta enabled the first wave of a new knowledge economy, but what came next was emergence.
Next - The 2030s...
Thanks to Adam Wern, Alan Laidlaw, Frode Hegland, Mark Anderson, Peter Wasilko, and Rafael Nepo for influencing this to begin with.