@nyewarburton.eth

Nye's Digital Labhttps://paragraph.com/@nyewarburton.eth Weekly scribbles on creativity in the age of rapid change.Sun, 19 Jul 2026 02:23:17 GMThttps://validator.w3.org/feed/docs/rss2.htmlhttps://github.com/jpmonette/feedenNye's Digital Labhttps://storage.googleapis.com/papyrus\_images/f0e5f7860776c98c7d8500a2ee6e9784270bad9bd4ecab95cbf543f4614e3088.jpghttps://paragraph.com/@nyewarburton.eth

All rights reserved<![CDATA[Learning From Zero]]>https://paragraph.com/@nyewarburton.eth/learning-from-zero 3107u3Q88npBhqh2mApaSun, 19 Jul 2026 00:00:00 GMT

Nye's Digital Lab is a weekly scribble on creativity in an age of rapid change.

I'm starting from Zero... again.

Essay #89

Coffee

Generic Coffee Logo, Nano Banana

I was bad at sales for most of my life.

Not shy-bad. I think I can talk to anyone. Competent-bad. The kind where you know the work is good and the pitch is true and the client needs exactly what you make, and still you can't get your mouth to do the part where you ask a stranger for money. I'd walk to the edge of the ask and hear myself say "anyway, let me know if you ever need anything." That's not a close. No wonder they never called.

It took me years to see what I was doing. I was protecting myself. If I never asked, nobody could say no, and the good opinion I had of my own work could stay sealed in its case. Nice and safe.

This went on until it caught up with me. I hit a bad stretch (the specifics don't matter, everybody gets one) and my friend Gavin sat me down. Gavin is a salesman, an actual one, the kind who's delighted to be told no because a no is data and data is Tuesday. He didn't give me a speech. He handed me a legal pad and told me to write down fifty names. I said I didn't know fifty people. He said I knew fifty people. I wrote eleven fast. The other thirty-nine came out of the corners of my brain like coins out of a couch: old classmates, the guy who ran comp at a place I'd left on bad terms, somebody's cousin, a woman I'd sat next to on a plane. Then Gavin gave me the whole method, which took nine seconds.

Ask them for coffee. Say this is what I do... is there anyone I should talk to?

Write down whatever they say. Go down the list.

No funnel, no deck, no pipeline. A legal pad and two sentences.

I did all fifty and it took months. Most were nothing. Some were humbling; one guy checked his phone the entire time, said "cool, cool, cool," and left, and I sat there afterward conducting a full inventory of my life choices. But some gave me a name, and the names gave me names, and by the end of my run in LA I was running a small shop and landing deals. That's a sentence that would have made me laugh out loud in 2003.

Months of pounding pavements, Flow State/Leonardo.ai

Amateur

The fifty cups of coffee are not a sales technique. They're a machine for turning humiliation into arithmetic.

Every ask felt like a referendum on my worth as a person, which is a lot of weight to hang on a flat white in a Santa Monica coffee shop. The list dissolves that. Once fifty names exist on paper, a bad coffee isn't a verdict, it's just name "seventeen." The failure rate is pre-approved. You can't be destroyed by an outcome you already wrote into the plan.

For a long time I was in my head, and I filed my 50 cups of coffee under "amateur."

Real salespeople had training and language and systems that came from somewhere. I had a legal pad from Gavin. But it's the same with everything else I ever built because nobody would tell me the process. None of it came from a book. All of it got invented at the moment of need by a guy who didn't know what he was doing and couldn't afford to say so. I thought that was the shameful part.

Ha. Look at that. It was the job.

I look back at my history and see real work. Projects, relationships, growth. They're what got left behind. The actual career. The only thing that traveled with me from one collapsing industry to the next is a stack of improvised systems I built while standing in rooms where I didn't belong yet.

I'm not a portfolio. I'm fifty cups of coffee and a collection of forty other short hand frameworks I made up under duress. Because that's what it is.

The modern 50 cup of coffee pipeline, Stable Diffusion

Zero

I'm starting a new chapter now, pointed at robotics and AI, and I know exactly what this is.

I guess it's the difference between fifty and thirty. The feeling hasn't changed:

Twenty-five years in, and I'm functionally a beginner. At thirty that meant you're about to be found out. Now I recognize it on sight. It's not a warning. It's the feeling of the first week. I've had this exact week forty times, and every time it was followed by a second week.

It's confidence about the interval between not knowing and knowing, the only part of my career that ever repeated. I'm good in that gap (I guess?). I've spent my whole life there. Improvisation isn't what you fall back on when the plan fails. It's what works when there's no plan yet, which is most of the time.

Which is convenient, because everyone is standing at zero. This is fact.

The tools that matter didn't exist when the current experts were forming their expertise. The gap between me and the person who intimidates me isn't thirty years of wisdom, it's eighteen months of reading and considerably more nerve. Nobody has a deep bench; there hasn't been time to build one. Everybody's improvising.

So I'm doing it again. Fifty names on a friggin' legal pad. (Thanks Gavin) I'm going after researchers, founders, weirdos with good github repos, a few who'll never write back. Same two sentences. Write down the answer. Go down the list. It felt amateur at thirty. It doesn't now, because waiting until you're qualified is a slower way of never starting, and the qualification was always the same thing:

Zero isn't a hole you climb out of. It's where everyone is standing right now, mostly pretending otherwise.

Bring a pad. And... Make it Happen.

Nye Warburton is an educator and adventurer from Savannah, Georgia. These essays begin as improvisational sessions with Otter, are ingested and revised in Obsidian, and then sweetened with Claude's fancy suite of agentic tools.

A collection of last year's essays are in a printed edition are available to purchase at https://nyewarburton.com/book/

More from the Lab

]]>nyewarburton.eth@newsletter.paragraph.com (Nye )learningsalesmotivationessay<![CDATA[Private and Public Networks, Part 2 - Common Ground]]>https://paragraph.com/@nyewarburton.eth/private-and-public-networks-part-2-common-ground Tm1btCLhtgHQy8ws1bGZSun, 12 Jul 2026 00:00:13 GMT

Nye's Digital Lab is a weekly scribble on creativity in an age of rapid change.

Last week, in part one of this two-parter, I lost six hours of my life to the private networks that run my phone, my baseball, and increasingly my everything.

This week: the other kind of network! The kind nobody owns, and why I think it's where the promise is.

The Detour Down Abercorn

My wife and I went for a drive to Port Royal, South Carolina. It's a really lovely little town, worth your day. We called up Google Maps, and Google, knowing our part of Savannah, did what Google does: straight onto the 16, over the bridge, optimized, efficient, correct.

My wife said no. "I want to go down Abercorn."

There was absolutely no reason to go down Abercorn. It's slower. But at this time of year the trees along Abercorn are growing in lush and full, the oaks doing that thing Savannah oaks do, and she wanted to see them.

So we took the dumb route, and it was the right route.

A fairly decent simulation of Abercorn's Oak Trees by Google Gemini.

I keep coming back to that drive, because it's a small version of a question I think we should be asking constantly right now:

what exactly are we automating away?

The routing algorithm optimized for time.

My wife optimized for trees. And that's why we like public parks.

Public parks were invented for precisely this human instinct. New York and San Francisco carved green squares out of the most expensive land on earth because people need somewhere to walk, gather, and be unproductive together. Savannah's squares are the anchor of everything lively and vibrant about this city. Nobody's monetizing your stroll through them. That's the point of them.

A park is a network too. It's just a network whose business model is you being fine.

You must pay the "Road Guy" to drive. Leonardo.ai/Flow State

Nobody Is "The Road Guy"

Here's the thought experiment I've been trying out.

Imagine founding a town, and one guy declares himself the road guy. He builds the roads, he owns the roads, and he charges you at the end of your driveway. Of course this is done with surge pricing, a loyalty program, and a mandatory RoadPlus subscription to access the left lane.

This is silly. Every town needs roads, so we build the roads together. Same with the electric grid, the water system, the schools. Yes, these are centralized entities paid through taxes, and yes, government runs them with all the charm of a DMV, but please notice what they don't do. The road doesn't lock you in. The grid doesn't upsell you. Your kid's school doesn't require a Disney account. (Or it shouldn't!) These are public networks: boring, universal, and quietly load-bearing for everything else, including every private fortune ever built on top of them.

So as more of life moves into digital worlds, and as AI becomes something closer to a utility than a gadget, the question writes itself:

where are the digital public networks?

Where are the roads in the AI future?

Because the cost curve is coming. If private AI networks keep enshittifying on schedule, which means more lock-in, more tiered pricing, more tollbooths between you and your own data, then a lot of people are going to get priced out of what will feel, by then, like a basic service. Not a luxury. The thing your job application gets filtered through, the thing your kid's homework assumes, the thing your doctor's office runs on.

When electricity became essential, we didn't shrug and let the road guy handle it. There was a collective mechanism, because there had to be.

We've succeeded at this before, spectacularly, with technology no fancier than paper.

It's called civics, baby. We wrote constitutions. We broke government into branches and factions, set them against each other on purpose, and built reasonable consensus between systems that don't trust each other. That's a public network protocol that's been running, with patches, for centuries. It's slow and it's ugly and it sucks, but man, it works, well ... sometimes... ok maybe not now, but we have to believe.

The failure of one implementation isn't the failure of the idea — it's a reason to keep iterating on the architecture. Especially as the other direction looks pretty bad.

Collision of Networks, Leonardo.ai/Flow State

Something in the Middle

Pick whichever one you fear, be it the socialist agenda or the aggressive capitalist one, the both complicate all of this equally. I'm not arguing for nationalizing the feed. Pure public systems have their own rot: bureaucracy, capture, the DMV of it all. And part one already covered what pure private systems become.

What I actually think is that these two concepts have to collide, and something has to emerge in the middle. There needs to be a hybrid that lets us sustainably distribute and fractionalize whatever comes next.

Maybe I can finally buy my one Dodgers game individually because it travels over a communal network of individual things, a market inside a commons, the way a food truck operates on a public street.

Do those markets work together? I honestly don't know. But the shape I keep sketching looks something like local-first creative sovereignty.

I don't fully know. I told you this was a meditation, not a manifesto.

But I know which instinct I trust. The algorithm wanted the highway. My wife wanted the trees. Like the Lorax warned us, the networks we build next should leave room for the trees.

Source: Random Facebook-stream via Google search

Make it Happen.

That’s it for this time. I do this every week; if you vibe to the ideas I express, consider subscribing or sharing with friends. If you like tech-detoxing with a book like I do, I crammed some of last year’s best essays into a printed collection.

My essays begin as an improvisation on a morning walk, that become a voice note in Otter.ai, take shape in Obsidian, and are finished in collaboration with several agentic versions of Claude's fancy AI suite.

For more about me: https://nyewarburton.com

We’ll see you next time.

If this resonated, here are some other scribbles from the Digital Lab that are worth your time:

Subscribe

]]>nyewarburton.eth@newsletter.paragraph.com (Nye )publicnetworksdesigntrees<![CDATA[Private and Public Networks, Part 1]]>https://paragraph.com/@nyewarburton.eth/private-and-public-networks-part-1 GkTu1lizws3wUEdsPvNDSun, 05 Jul 2026 00:00:00 GMT

Nye's Digital Lab is a weekly scribble on creativity in an age of rapid change.

This week kicks off a two-parter I've been chewing on while traveling:

private and public networks.

Basically, what the difference is, and what it might mean as AI wires itself into everything.

Part one is about the private side, which means part one is a complaint. (A useful one, I promise.)

There is a special place in hell for wireless stores. Leonardo.ai/Flow State.

I Just Wanted a Phone Number

My wife and I decided to switch carriers. AT&T to Verizon. This should have been the easiest transaction in the history of commerce. Here is a customer walking toward a company, money already in hand, saying please, take it, let me in.

We had every advantage. Our son's Gizmo watch was already on Verizon. The phones were unlocked Apple hardware. This was a layup.

It took six hours.

An hour and a half in the Apple Store just to buy the phones. A trip to AT&T to extract some magic account information. Then Verizon, where my number got stuck in a liminal porting hell somewhere between two multi-billion-dollar networks, tangled up (I swear this is true) in my son's watch plan. Nobody could tell us where the number went. Two competent adults, one trivial task, six hours.

This is just what my life is now. This is all of our lives now.

Increasingly, my world runs on resetting passwords, or begging Apple to send a message to myself so I can prove I'm me to some other system. I wanted to watch a single Dodgers game last month. One game. That required the MLB app, which has been ingested by the ESPN app, which upgraded me into a Disney account, which is billed through Apple because that's where my apps live. I am now four subscriptions deep into a corporate org chart, and I genuinely do not know how to get out.

None of this is a bug. It's the F'N architecture.

The Decay Is the Business Model

Cory Doctorow has a word for this: enshittification.

The pattern is consistent enough to state plainly. A private network starts out genuinely good to its users, because it needs you. Then you get locked in. This means your contacts, your data, your habits, your kid's watch are locked in and it starts extracting value from you. Eventually it extracts from everyone, because by then your exit costs more than your patience.

Corey Doctorow's writing is definitely worth your time:

https://craphound.com/

I want to be fair here. A capitalist private network is a perfectly good way to build something satisfying and make a lot of money. This is where the business models are. The early walled garden is usually a lovely garden. The problem comes when these networks scale and lose focus, when the person standing at the counter stops being the point. The friction I felt in that Verizon store wasn't designed by a villain. It's just what's left over when convenience stops being the goal.

Lock-in isn't a side effect of the private network. Lock-in is the product.

Once you see the pattern, you see it everywhere. The streaming service that raised prices three times and added ads to the tier you already pay for. The airline that turned a seat assignment into a la carte revenue. The printer that bricks itself over third-party ink. Different industries, same math:

Every one of these started as a genuinely better deal than what it replaced.

Ok, so here it is.

Now add AI to this picture, because everyone is about to.

If you think bolting AI onto these bureaucratic mega-networks makes them better, I have a bridge on the 16 to sell you. AI doesn't reverse the incentives; it turbocharges them. More data gathered into centralized actors.

More surface area for tiered pricing. The cheap onboard to get your own information in, pay again to get it out. More clever, personalized, machine-learned ways to make the exit ramp just a little longer than your patience. The same companies that can't port a phone number in under six hours are now promising that an agent will handle my life for me.

Handle it where, exactly? Inside whose garden? Under whose terms of service?

Lock it in. Leonardo.ai/Flow State

You Can't Unilaterally Leave a Network

This is kind of what I am driving at with private networks.

We can't fully opt out.

The infrastructure of ordinary life has been quietly rebuilt on top of these private networks. The school PTA runs on Facebook. The kids' tournament schedule posts exclusively to Instagram. Your two-factor code goes to a phone number that belongs, functionally, to a carrier that treats it like a hostage.

Oh yes.

I've fantasized about the clean break. I am a sucker for the Light Phone, the local-first stack, the landline and the filing cabinet. But you can't unilaterally exit a network that everyone around you is still standing inside. Deleting your account doesn't free you.

When the platform becomes the town square, leaving doesn't make you free. It makes you invisible to the town.

As much as I loath a certain social network, I need to log in to see where to drive my kid for practice.

That's the real trap of private networks at scale. Not that any single company is evil. Most of the humans inside them are kind, competent, and structurally forbidden from helping you. The support rep who couldn't find my number wasn't lazy; she'd been designed into powerlessness, because giving her real authority costs the company money. The trap is that things which used to be genuinely public like the bulletin board, the little league schedule, the phone number itself are now pulled into having a user ID and password to a company database.

So the short-term forecast is not great. Private networks will keep consolidating, AI will make them stickier and more expensive, and more of us will get priced out or locked in (probably both at once.) The pursuit of money isn't going to pause and reconsider. These networks will keep doing exactly what they're built to do.

Which raises the question I actually want to ask, the one this whole six-hour ordeal was scratching at:

What's the alternative?

Because we already have networks that don't work this way. You drove on one today. It didn't ask you to link your Disney account. Nobody in your town is "the road guy," charging a toll at the end of your driveway. There is another model which is older, less glamorous, and weirdly resilient.

I think it has something important to say about what we build next.

That's part two!

Next week: public networks, my wife's detour down Abercorn, and why the trees might matter more than the traffic.

Make it Happen

Thanks for reading. I do this every week, (even when I am on vacation - Go Dodgers) If you vibe to the ideas I express, consider subscribing or sharing with friends. We'll see you next time!

These essays start in Nye's Digital Lab Brain Vault, which is a collection of research and improvisations on a local hard drive. From here, ideas are refined and developed with Anthropic Claude, and sweetened with old fashioned human labor.

For more information visit: https://nyewarburton.com

More from the Digital Lab:

]]>nyewarburton.eth@newsletter.paragraph.com (Nye )networkspublicprivatecorporations<![CDATA[The Engine Is the Next Frontier]]>https://paragraph.com/@nyewarburton.eth/the-engine-is-the-next-frontier 7s0lnjEXcR1zcQMuWMxuSun, 28 Jun 2026 00:00:15 GMTNye's Digital Lab | #86

June 2026

Nye's Digital Lab is a weekly scribble on creativity in an age of rapid change.

Last week at Unreal Fest in Chicago, Epic shipped Unreal Engine 5.8 with an experimental, first-party MCP plugin. If you don't know what that is, it's an open protocol Anthropic introduced to route AI agents into software.

This means that you can now point Claude at the Unreal Editor and have it spawn actors, set lighting, and run tests from inside the engine. This is an "officially" licensed and endorsed plugin.

To the game development universe, the outcry was deafening.

I. Is it official?

People weren't mad about a plugin. They were mad about what it meant.

For years, AI has crept into game engines through side doors. Game AI is decades old. A hundred neural networks are buried under buttons in all parts of the engine.

Epic stamping the MCP protocol officially was pretty jarring.

It was the enterprise saying, out loud to the world, "this is it, everyone." This was especially rough on the fans of Blueprints, which is a successful visual scripting in the engine, but I think that it cracked open an old wound:

what even counts as "real" programming?

Every anxiety about where game development is heading got packaged into one feature and shoved into view.

What fascinates me is how split the reaction is.

I am terrified of generative AI replacing imagery yet completely unbothered by the machine-learning baked into engine's rendering. I find students in hallways asking ChatGPT what to say next. AI is increasingly in everyone's workflow; that's obvious. What's left is signaling or defending your "artisthood."

I believe that it will be a while before any of this has cultural acceptance.

Once Unreal declares AI "accepted," it hands the community license to run amok. I can think of no worse case than connecting models, local or centralized, that accelerate the theft of datasets, design tropes, and assets we already can't police on Steam or in publishing.

But this is where we are.

Front page of the Game Dev Universe, Screen Shot

II. The real frontier is simulation

I but I believe that the plugin fight is the small story.

The big story is what an engine becomes once AI lives inside it.

When I first experimented with AI, what completely changed my life was the discovery of reinforcement learning. You should think of this as letting a system run a simulation over and over until it learns. It's never been easily accessible to successfully accomplish. I have mere experimental prototypes at best.

Quite bluntly, it's near impossible for a non-technical professional.

Tools like Isaac Sim demand a mountain of dependencies and compute just to set up, let alone run an engine through millions of cycles. AI is about to abstract that pain away. And once training is cheap, the engine stops being a place you decorate and becomes a place you teach.

The obvious payoff is robotics. But it's likely to be hospitals, theme parks, live events, virtual events, drone swarms, and a lavish, far away, shopping island paradice.

It's Cartesian space. It's a virtual one-to-one model of the real world, rendered in coordinates that manifest from latent space. Soon we'll scan environments in near real time, quantify what's in them, and decide inside that interactive space. Embed a little intelligence, (i.e. notice a table sits three feet off the wall in any room I drop it in,) and the design process accelerates itself.

Better still, it could become a single source of truth. A firm and its subcontractors start collaborating on one model: a digital thread running end to end.

That's always been the promise of engines in enterprise.

The blockers were complexity, slow rendering, version control, and the programming most people never learned.

AI just absolutely will flatten the difficulty curve and stands to drop the cost of design by billions across verticals — cars, aerospace, cities —

All of it by letting us simulate before we build.

Instant Worlds in the Engine, Leonardo.ai/Flow State

III. It's just the awareness of the elephant

So my takeaway is simple. MCP isn't the story; it's the industry waking up to it.

The story that I think this will actually cause is reinforcement learning and simulation handing creators a new kind of leverage. (That's if we can afford the tokens, but that's another story.)

So to the adventurers amongst you, my advice:

The engine was always the interactive layer. AI just made it accessible to the rest of us.

Make it Happen.

Nye Warburton is an educator from Savannah, Georgia, who has spent too much time playing with technology. This essay was crafted through improvisational sessions using Otter.ai, it was edited and refined using Claude Sonnet 4.6, and spiced with a bit of old-fashioned hand-key human labor.

You may find Interesting:

More from the Digital Lab

]]>nyewarburton.eth@newsletter.paragraph.com (Nye )gamedevenginesaigamessimulations<![CDATA[It's All in the Collection]]>https://paragraph.com/@nyewarburton.eth/its-all-in-the-collection ARS6ngeVXFwcAy5eoglQSun, 21 Jun 2026 00:00:00 GMTNye's Digital Lab is a weekly scribble on creativity in an age of rapid change.

Essay #85

It's what you collect.

Asking me what AI I'm using is like asking whether I animated in Maya, Max, or Blender. Who cares. It's not the ones who know how to use what model. They're the ones who walked in with something worth saying.

You don't produce ideas on command. You can't.

It quietly infuriates me how often someone says just come up with an idea, as if it were a vending-machine transaction. An original idea is the residue of having paid attention for a long time.

For years that meant sketchbooks. A flash of something will sling by my brain. This can be a phrase overheard, a shape, a half-formed argument. Great ideas have a half life decay and I needed to get it down before they evaporate.

That instinct to capture is the muscle that matters.

Sorting my writing in the digital lab. Screen Capture.

Lately the sketchbook is Obsidian, and I collect as much as is humanly possible.

A meeting, a question someone asks sideways, a thought in the shower, a tweet, a podcast. Really, any of it might carry the seed of an essay, a class, an entire initiative. The seeds don't announce themselves. They're found later, and only if they were kept.

You cannot connect dots you never wrote down.

So the work is to pay attention to all of it. Combine ideas like an alchemist; use AI to fill the gaps. I can leap from economics to networks to psychology on a guess.

I work in games and tech. That's no reason to learn nothing from interior design, or fibers, or graphic design. There's an artist in every one of those rooms, and one will hand me a connection I couldn't have reached from inside my own discipline. The failure mode is deciding in advance their work has nothing for you.

The better move is to find the bridge between their world and yours. And it's usually right where the great idea lives.

Painting a masterpiece. Screenshot.

The future is a collection

The future of nearly everything is different collections.

We're already drifting toward recording all of it. We record every conversation, interaction, transaction. All of it because the value is too obvious to resist. Capture everything and you can measure everything, then manufacture insight on demand.

The data is inert.

A pile of recordings is not an idea, and a perfectly indexed life is not a meaningful one.

The value doesn't live in the collection. It lives in the take. This means the human act of deciding what a thing means, which dots are worth connecting, what's worth keeping and what's just noise dressed up as signal.

The collection is the casino. The take is the bet only you would make.

So when people ask which AI I use, it's the wrong question. I could say Claude in the terminal inside my Obsidian, a little Hermes, some ComfyUI experiments. But getting good was never about learning models. I keep what I notice, and I bring a point of view no amount of recording can generate on its own. (At least, as a human, I hope so.)

Get that right and the model is almost incidental.

Get it wrong and the best model in the world only helps you produce (albeit faster, more fluently) things that were never worth making.

The frontier everyone's watching is the machine.

But I'm still gambling on you collecting something interesting first. It's what you collect, and what you make of it. Start collecting anything that you find interesting. The frontier that decides whether any of this is worth doing is still you.

Make it Happen.

Nye Warburton is an artist and educator from Savannah, Georgia.

These essays start as improvisations. Writing is shaped with personalized data sets, Claude agents, and finally edited by hand in Obsidian. Collected essays are available as a printed book at nyewarburton.com/book.

]]>nyewarburton.eth@newsletter.paragraph.com (Nye )collectionwritingaicreativity<![CDATA[The Graphics Hack]]>https://paragraph.com/@nyewarburton.eth/the-graphics-hack HrbQ2LgARzISRhvTbfpRSun, 14 Jun 2026 00:00:00 GMT

In computer graphics... we cheat.

Cheating is the job. The people you admire most in graphics are not the ones who do the most work. They are the ones who figure out how to do the least.

Nye's Digital Lab is a weekly scribble on creativity in an age of rapid change.

Essay #84

Watch the hands of a Noob

Here is a test I used to run in my head.

Watch someone work in any graphics application. Watch them select an object and drag it across the screen. Then watch them go back, select a second object, and drag that one too.

That person just told you everything about how they think.

Because they could have grouped both objects and dragged them once. One move instead of two. It sounds trivial. It is not. Scale that decision across ten thousand operations and you get the difference between an artist who ships and an artist who drowns.

This is the graphics hack. The trick. The shortcut that is not lazy but smart.

Graphics rewards the person who refuses to do anything twice.

Coffee-driven development

When I worked in previs, I had a tool that changed my mornings.

I could select a character, point it at a mocap library, and set an auto-retargeter loose. Then I would go get coffee.

By the time I came back, half the scene was up. Characters moving, blocking roughed in, the skeleton of a shot on screen. It was rough, but it was work. All, while I stood in line at the starbucks across the street.

I felt like a genius. Not because I had worked hard. Because I had worked sideways.

Same with water. The honest way to simulate water is particles. Millions of them, physics on every one, render farms groaning on and on through the night. The hack is to skip all of it. Use animated textures. Scroll some noise, fake the speculars, sell the illusion. Quick and dirty. The audience never knows. The audience never cares.

Hacking is freedom in graphics.

The cheats were always the glory. Every artist I respected had a private arsenal of them, traded like contraband, defended like trade secrets.

And then the machine learned to cheat

So here is my problem...

AI is the ultimate graphics hack. It is every shortcut I ever built, compressed into a single move. Why group two objects when you can describe the whole scene? Why fake water with textures when you can type the word "water?"

I should be thrilled. This is the endgame of everything I believed. The cheat to end all cheats.

Instead I feel weird about it. And I have been trying to figure out why.

I think it is this. My hacks were earned.

I dragged objects one at a time for years before I learned to group them. I built particle sims before I learned to fake them. The shortcut only felt like glory because I knew exactly what I was skipping. The cheat had a price, and I had paid it.

The new cheat skips the part where you learn what you are skipping.

A lifetime of cheating taught me that the cheat is the reward for understanding the system. Now the system understands itself. We did not just get cleverer.

We may have "clevered" ourselves out of the loop entirely.

Choose Wisely, Nano Banana

Two doors

So I see two doors, and I am honestly standing between them.

Door one: slow down. Celebrate process.

Make students drag the objects one at a time, on purpose, the way a chef still learns knife skills in a world full of food processors. Earn the cheat before you use it. I have built a whole pedagogy on this idea.

Boss fight first, automation after.

Door two: screw it! Generate the whole thing.

Accept that the craft was always in service of the image, never the other way around, and the image is now free. Stop mourning the particles. The audience never knew.

And likely the audience never cared.

Most days I believe in door one. Some days, late at night, watching a model conjure in seconds what took my team a month, door two looks pretty damn good.

Here is where I land, at least this week. The graphics hack was never really about saving time. It was about the feeling of beating the system, and you cannot beat a system you never entered. The drag, the group, the retargeter, the fake water. Those were moves inside a game I knew cold.

If you want the glory of the cheat, you still have to play the game first.

Then ... cheat like hell.

Make it Happen.

Nye Warburton is an artist and educator from Savannah, Georgia.

]]>nyewarburton.eth@newsletter.paragraph.com (Nye )graphicshackingaiprocessartcreativity<![CDATA[I'll Take my Data Share, Thanks]]>https://paragraph.com/@nyewarburton.eth/ill-take-my-data-share-thanks TY8wMrmJ50bjDQTu2zIUSat, 06 Jun 2026 00:00:00 GMT

Nye's Digital Lab is a weekly scribble on creatives in an age of rapid change.

This week I'm looking at Bernie Sanders' plan to tax the AI giants and why I'd rather have a receipt than a check.

A flat 50% tax treats my creative work like oil. A big pile of stuff to be split. What I actually want is a provenance system that can prove a specific idea was mine and route its value back to me.

Bernie's warning shot

On June 1, 2026, Bernie Sanders again did what he's pretty good at. He turned a complicated problem into a one-liner.

In a New York Times op-ed he announced the American A.I. Sovereign Wealth Fund Act: a one-time 50% tax on OpenAI, Anthropic, and xAI, paid in stock, dropped into a public fund that eventually mails Americans a check.

"A.I. is built on our collective intelligence," he wrote "our books, songs, artwork, journalism... fed into their A.I. models without permission, without acknowledgment, without compensation."

I'm a creative. I know that somewhere in those training sets is the residue of work I made by hand. This can be drawings, code, the way I do something and what hurts, is that nobody asked me. So when Bernie says they took our stuff, we want our cut, I'm with him. I want mine too.

Then I look at the number, and I think about it.

This is not the right way to go about it.

The Problem With a Number

I'm sure this is how they are approaching it.

50% tax treats my data like oil.

Oil is a lump you tax once and split evenly. My work isn't. My work is labor and it's specific to me. If a sketch I made taught a model to animate a character more believably, that one contribution might be doing real work inside the output. A flat tax can't see that. It scoops everything into a "government pile", promises fairness later, and asks me to trust whoever happens to hold the White House with the slush fund.

And we already know the pile doesn't trickle. When the music platforms built "opt-in" royalty pools for artists, the payout landed around four cents per track, per quarter. Four cents, dude. Um... not liking that outcome.

So no, I don't want to be a line item in someone's fund. I appreciate this conversation, and I am glad we are raising the alarm on our data in AI. For me, as a creative, I want my fair share!

Build the Thing That Proves It Was Ours

Don't tell me we don't have the capability to track things. We already built the surveillance system. We just aimed it at the wrong people.

Adobe logs every prompt you type. Company logs can prove you used AI even when you swear you didn't. The C2PA "Content Credentials" standard (backed by Adobe, OpenAI, and Google) now stamps cryptographic origin data onto generated images, and this August the EU starts requiring it.

I watched a hacker conference on youtube where blockchain engineers showed system maps that blew my mind. We can trace a single satoshi, one hundred-millionth of a Bitcoin, anywhere on Earth.

So, and let me get this right... we can prove who typed a prompt and where a coin went, but not whose work taught the model?

We can start doing this now. And I mean now.

That's the thing I actually want built. Not a tax... sorry I don't think we have the capability to manage that.

I think we need a provenance system.

When my work gets ingested, an immutable receipt rides along with it, and when it generates value downstream, a fraction finds its way home. To me. By name. Cryptographic Secure number that says "Nye made this thing I just stole and put in my AI."

I've been told... It's hard today's metadata falls off the moment you change a file's format, and the blockchain ecosystem just doesn't have it's act together. Fine. Invent something sturdier. We have AI now; point it at this. We can solve this problem, and if we don't. Well. Here's the thing. We won't have an economy, folks.

Bernie did the one thing only Bernie can do: he slammed the table and made the country look up. Appreciate you man, thanks. Fist Bump

But, that's the opening scene, not the ending. The rest is up to the economists, the IP lawyers, and the technologists. Hurry Up!

I don't want 50% of Anthropic. I want the teenie tiny latent space node that was genuinely mine, and a system honest enough to go find it, and then give me credit for it.

Someone, let's build this.

Make it Happen.

Nye Warburton is an artist and educator from Savannah, Georgia.

These essays start as improvisations. Writing is shaped with personalized data sets, Claude agents, and finally edited by hand in Obsidian. Collected essays are available as a printed book at nyewarburton.com/book.

Sources

Links, quotes, references

You might like...

If you are down the rabbit hole on IP tracking, you might like some of my other scribbles on it.

]]>nyewarburton.eth@newsletter.paragraph.com (Nye )<![CDATA[The Recruiting Trap]]>https://paragraph.com/@nyewarburton.eth/the-recruiting-trap KGSdZHLjgxnM5myeIJAJSun, 31 May 2026 00:00:00 GMT

Nye's Digital Lab is a weekly scribble on creativity in an age of rapid change.

This year's distilled advice for the outgoing crop of creative graduates.

Nye's Digital Lab | #82

May 2026

Recruiting

Recruiting, as it exists inside any sufficiently large company, is an outsourcing function.

A hiring manager said I need talent, but I don't have time to find talent, and that need got handed off to a department whose job is to amass and filter candidates at scale.

Note: I believe that good recruiters are valuable. This is not a knock on recruiters.

The ones I've worked with are sharp, networked, and good at their jobs. But their job is not (and structurally cannot be) understanding what makes a great game developer. Or a great technical artist. Or a great anything in a creative-technical field. That understanding takes years of doing the work. Recruiters don't have years of doing the work. They have a req, a timeline, and a stack of resumes that they use AI to process.

So they default to #tags .

Must have Unreal Engine experience.

5+ years shipped AAA. Maya, ZBrush, Substance.

The tags become a proxy for capability, and the proxy becomes the gate.

A developer who has shipped on three other engines but never touched Unreal will get filtered out, while someone who has Unreal on their resume because they took one class will get through. The tag is a terrible signal for the thing it's pretending to measure. Versatility, quick thinking, adaptability. None of it fits in a keyword filter.

This is the recruiting trap today. It might change, but for now, the system is optimized for legibility, not for value.

And if you build your career around being legible to that system, you will spend years being slightly the wrong shape for every door you try to walk through.

Can't see

What the trap cannot see is the person who goes beyond the brief.

The student who, on a weekend, vibe-codes a tool that cuts a pipeline step from eight hours to twenty minutes. The junior who notices that the rig everyone hates could be rebuilt, rebuilds it, and quietly hands it back to the team. The artist who writes a camera-shake script and open-sources it onto the studio's GitHub because the cameras needed shake and nobody had gotten around to it.

These people are absurdly valuable!

They are also completely invisible to the funnel, because the funnel is asking:

ahem, do you have the tags?

and they are answering a question the funnel never thought to ask:

Can you see the problem and fix it ?

That second question is the only one that matters, and it is the question every actual hiring manager is desperate to find an answer to. They just can't get the answer through the system that's supposed to deliver it to them.

Avoid the Trap

So here is the move to avoid the recruiting trap:

Stop applying to jobs.

Start solving problems.

If you find a company whose work you respect, whose aesthetic, mission, or culture pulls you in, don't check their careers page. Check their output. Play the game. Read the postmortem. Watch the GDC talk. Find the seams. Every shipped product has seams. Every pipeline has friction. Every team has something they can't get to because the urgent keeps eating the important.

Then go to them with the seam. Not with a resume. With an observation, a prototype, a pitch, a fix.

I had a student who did exactly this. He went into an interview with a game he'd built, and the technical director on the other side of the table. He started asking him the things they have tried, and the approaches they were working on. That interview wasn't an interview anymore. It was the first day of work. He was already inside the culture of building, and the company recognized it immediately, because the culture of building is what they were starving for.

You will hear that this approach is annoying. That you'll be a pest. That cold-pitching solutions is presumptuous. Some of that is true. If your ideas are bad, you won't get anywhere. And yes, you will annoy some people. I seem to be ok, and I've definitely done that.

You will also, every so often, hit the exact nerve of someone who has been staring at a problem for six months with no bandwidth to solve it, and they will look up and say who are you? I know because sometimes those people just appear in front of me, with exactly the ideas I am looking for.

That single hit is worth a thousand silent applications.

Thinking in Systems

The thing under all of this is systems thinking.

The ability to look at a working operation and see not just what it does, but where it strains. Where the load concentrates. Where one good intervention would unlock ten downstream effects.

Most people operate inside systems.

They take the inputs, do their part, hand off the outputs. Systems thinkers see "the matrix." They see why the inputs are shaped the way they are, why the handoff happens at that particular seam, why the autonomous vehicle takes the left instead of the right. And when something breaks, they don't just patch the symptom. They route through the chain.

This is the skill I'm trying to build in every student who I interact with. Not Unreal proficiency. Not Python. Not any specific tool. Tools change. The job I have is to make you the best creative-technical problem solver I can, because problems are where opportunity lives. When something is working, the people who built it own all the upside. When something breaks, the door opens for whoever can see the fix.

Let's make sure we get this, because I think it's important.

When things work, nobody pays for them.

When things break, money flows toward whoever can make the problem go away.

Be that person. Be relentlessly that person. Make the problems go away.

Here's a thought for you.

You will be wrong about some of them. That's fine, being wrong in specific is more valuable than being right in general, because being wrong in specific is a conversation, and a conversation is a door.

But the work will be there. Yes, even with the F'n robots coming, the work will be there. Because the problems are everywhere, and the people who can see them are rare.

That's the game. Find the problems, solve them, stop labeling the jobs.

Have a great career.

Make it Happen.

Nye Warburton is an artist and educator from Savannah, Georgia.

These essays start as improvisations. Writing is shaped with personalized data sets, Claude agents, and finally edited by hand in Obsidian. Collected essays are available as a printed book at nyewarburton.com/book.