Paul Stamp (Stampy)

@paulstamp

I make software tangible. Senior game dev @nonatomicgames. ServiceKit + VSM2 on GitHub. Focus: location-based gaming venues end-to-end. AI tooling on the side.

Cheshire

Joined May 2008

2.7K Following

1.5K Followers

2.5K Posts

Pinned Tweet

Paul Stamp (Stampy) @paulstamp

10 days ago

Guilt by automation Sometime in the last year, the burden of proof flipped. It used to be that you made a thing and people decided whether it was any good. Now you make a thing and, before anyone will engage with whether it's good, you have to prove you made it at all. That a machine didn’t do it for you. That you sweated. That it counts. I felt the flip land last week. I'd written a reflective piece called *The Reload Cost*, about how AI has quietly removed the cost of switching between tasks, and what that does to the old virtue of single-minded focus. It's a quiet, personal essay. I'm proud of it. I also wanted more people to read it, so I boosted it onto X to drive some traffic back to the site. That's the honest bit I should get out of the way first: I put it in front of strangers on purpose. One of those strangers replied that it was AI slop. Not "I disagree." Not "the focus argument doesn't hold up." Just the label, dropped like a verdict. And here’s what surprised me. It didn’t make me angry. It made me defensive in a way I didn't enjoy noticing. Within about thirty seconds I was assembling evidence. The draft history. The paragraphs I'd rewritten more times than I'd admit. The turns of phrase too clumsy to be anyone's but mine. I was building a legal defence for the crime of having published, and it took me another minute to catch what was wrong with the exercise. I was trying to prove I'd suffered enough to be allowed an opinion. As if visible effort were the thing in question. As if any quantity of documented sweat could answer someone who had decided, before reading a word, that the work didn't count. That presumption is the thing I want to name. Guilt by automation. The assumption that anything fluent, finished, or produced quickly is suspect until proven hand-made. It's a tax, and everyone who makes things in public is now paying it. I want to be fair to the other side, because I'm on it too. Let me concede the strong version of the argument up front, because it's true. There is an enormous amount of one-shotted slop on the web now. Content with no second draft, no point of view, no human who'll stand behind it. Generated, posted, forgotten. I dislike it as much as anyone. It clogs the feeds. It makes search worse. It buries the people who actually have something to say. If the complaint were only "there's too much thoughtless machine-made content," I'd be nodding along. But the slop label has stopped being a judgement about quality and become a judgement about method. And those are not the same thing. A lot of genuinely good work now exists because of AI, not in spite of it. Work that wouldn't exist at all otherwise, because the person who made it was short on time, short on a particular technical skill, or short on the confidence to start. For people who don't spend their working lives inside these tools, AI assistance isn't a shortcut they're embarrassed about. It's the thing that let them make something they couldn't have made before. That's not slop. That's access. The reflex to sneer at it is, more than anything, a failure of imagination about other people's constraints. The bit I don't like admitting. Here's the harder part. I love to write, and I find it frightening. I'm an introvert. Everything I publish carries a low hum of anxiety with it. The risk of getting it wrong in public. The small specific dread of being embarrassed in front of people whose opinion I care about. Lately there's a newer flavour: the risk of public confrontation with someone who's decided that undermining your work is a hobby. I want to be precise about this, because it would be easy to mistake what I'm saying for "I can't take criticism." I can. I want genuine critique: tell me the argument's thin, tell me the ending's soft, tell me I'm wrong about focus. That kind of attention is a gift; someone took my work seriously enough to push back on it.… https://t.co/huDe0AG9g9

paulstamp's tweet photo. Guilt by automation

Sometime in the last year, the burden of proof flipped. It used to be
that you made a thing and people decided whether it was any good. Now you
make a thing and, before anyone will engage with whether it's good, you have to prove you made it at all. That a machine didn’t do it for you. That you sweated. That it counts.

I felt the flip land last week. I'd written a reflective piece called *The
Reload Cost*, about how AI has quietly removed the cost of switching
between tasks, and what that does to the old virtue of single-minded focus.
It's a quiet, personal essay. I'm proud of it. I also wanted more people to
read it, so I boosted it onto X to drive some traffic back to the site.
That's the honest bit I should get out of the way first: I put it in front
of strangers on purpose.

One of those strangers replied that it was AI slop.

Not "I disagree." Not "the focus argument doesn't hold up." Just
the label, dropped like a verdict. And here’s what surprised me. It didn’t make me angry. It made me defensive in a way I didn't enjoy noticing.
Within about thirty seconds I was assembling evidence. The draft history.
The paragraphs I'd rewritten more times than I'd admit. The turns of phrase
too clumsy to be anyone's but mine. I was building a legal defence for the
crime of having published, and it took me another minute to catch what was
wrong with the exercise. I was trying to prove I'd suffered enough to be
allowed an opinion. As if visible effort were the thing in question. As if
any quantity of documented sweat could answer someone who had decided,
before reading a word, that the work didn't count.

That presumption is the thing I want to name. Guilt by automation. The
assumption that anything fluent, finished, or produced quickly is suspect
until proven hand-made. It's a tax, and everyone who makes things in public
is now paying it.

I want to be fair to the other side, because I'm on it too.
Let me concede the strong version of the argument up front, because it's
true. There is an enormous amount of one-shotted slop on the web now.
Content with no second draft, no point of view, no human who'll stand
behind it. Generated, posted, forgotten. I dislike it as much as anyone.
It clogs the feeds. It makes search worse. It buries the people who
actually have something to say. If the complaint were only "there's too much thoughtless machine-made
content," I'd be nodding along.

But the slop label has stopped being a judgement about quality and become a
judgement about method. And those are not the same thing.

A lot of genuinely good work now exists because of AI, not in spite of
it. Work that wouldn't exist at all otherwise, because the person who made
it was short on time, short on a particular technical skill, or short on
the confidence to start. For people who don't spend their working lives
inside these tools, AI assistance isn't a shortcut they're embarrassed
about. It's the thing that let them make something they couldn't have made
before. That's not slop. That's access. The reflex to sneer at it is, more
than anything, a failure of imagination about other people's constraints.

The bit I don't like admitting.
Here's the harder part. I love to write, and I find it
frightening. I'm an introvert. Everything I publish carries a low hum of anxiety with it. The risk of getting it wrong in public. The small specific dread of being embarrassed in front of people whose opinion I care about. Lately there's a newer flavour: the risk of public
confrontation with someone who's decided that undermining your work is a
hobby.

I want to be precise about this, because it would be easy to mistake what
I'm saying for "I can't take criticism." I can. I want genuine critique:
tell me the argument's thin, tell me the ending's soft, tell me I'm wrong
about focus. That kind of attention is a gift; someone took my work
seriously enough to push back on it.…

https://t.co/huDe0AG9g9

127K

Paul Stamp (Stampy) @paulstamp

13 days ago

@AndyLxNen Fair enough, Andrew. All the best.

Paul Stamp (Stampy) @paulstamp

16 days ago

The Reload Cost Deep focus wasn't a virtue. It was a response to a cost that AI has quietly removed. The weather has turned, which in the UK counts as an event. I'm in the garden with a coffee and a book, and for once the book is winning. It's The Only Way Is West, a walk along the Camino de Santiago, the better part of five hundred miles across northern Spain with everything you need carried on your back. I love this stuff. Travel writing, walking memoirs, long linear journeys with a start, a finish, and not much in between except the next step. I've always filed it under escapism. Sitting here in the sun, I'm starting to think it's more specific than that. Behind the book, quiet but never off, is the pull to work. I don't resent it. I like it, mostly. That constant urge to build and tinker is the thing that made me a software engineer in the first place, and it's the thing clients come back for. I'm a builder of digital doodads who always delivers. That reputation wasn't an accident and I'm not shy about it. But the urge comes with a watchful eye that never quite closes. It scans for new trends, curious technologies, the shiny thing two clicks away, and it leaves a trail. I have a long, slightly embarrassing spectrum of unfinished git repos, each one an experiment that reached the point where I understood the thing and then stopped, never quite becoming something worth sharing with the world. A graveyard, if you're feeling dramatic. A junk drawer, if you're being honest. Here is the part I keep circling. The walk on the page is the exact opposite of how I work. One path, one direction, west. You cannot flit on a pilgrimage. You cannot open six tabs and leave them humming. There is the road and the next step and the cathedral at the end, and the whole appeal is that you are doing precisely one thing with your whole self. I did this once for real. My brother Adam and I walked Hadrian's Wall, east from Bowness-on-Solway, wild camping the eighty-four miles to Wallsend, fifteen kilos on my back and eighteen on his. We set off in November, which was a decision. Cold, wet, the days so short we spent hours walking in the dark by headtorch, the world shrunk to the small bobbing circle of light in front of my boots. One afternoon we were low on water, soaked through and leaning into the wind, when we found a tap set into a wall. We were thrilled. It didn't work. We came round the next corner and there was a cafe, open and warm, and I am not exaggerating when I say the soup, crusty bread and hot coffee rearranged my understanding of luxury. That is what the narrowness does. Shrink the world to the next step and a bowl of soup becomes enormous. The scenery helped, the wall riding the hills with horses and sheep and cattle for company. But the thing I carried home wasn't the view. It was the quiet of having exactly one thing to do. On the approach to Carlisle we hit a flooded underpass, the path simply gone under a temporary lake. I still have the photo of Adam, bandana on, slumped at the edge of the water that had closed the path. I was scanning it for a way through like a bug newly surfaced in code review, turning it over, hunting the clean fix. We dreamed of a kayak washing up at the water's edge, the simple solution, but the November days were too short for dreams and wishes. So we went with brute force, hauling our combined thirty-three kilos of backache over gate after gate on a three-kilometre detour. In hindsight the lake was probably an overstated puddle and we should have waded it. No matter. I've made my peace with the fence; standing in the water weighing it up gets you nowhere, and the only move that counts is the one that goes forward. Even a flooded path has a direction. The desk doesn't. The quiet of one thing at a time is gone.… https://t.co/1EauwqZLBC

paulstamp's tweet photo. The Reload Cost

Deep focus wasn't a virtue. It was a response to a cost that AI has quietly removed.

The weather has turned, which in the UK counts as an event. I'm in the garden with a coffee and a book, and for once the book is winning. It's The Only Way Is West, a walk along the Camino de Santiago, the better part of five hundred miles across northern Spain with everything you need carried on your back. I love this stuff. Travel writing, walking memoirs, long linear journeys with a start, a finish, and not much in between except the next step. I've always filed it under escapism. Sitting here in the sun, I'm starting to think it's more specific than that.

Behind the book, quiet but never off, is the pull to work. I don't resent it. I like it, mostly. That constant urge to build and tinker is the thing that made me a software engineer in the first place, and it's the thing clients come back for. I'm a builder of digital doodads who always delivers. That reputation wasn't an accident and I'm not shy about it.

But the urge comes with a watchful eye that never quite closes. It scans for new trends, curious technologies, the shiny thing two clicks away, and it leaves a trail. I have a long, slightly embarrassing spectrum of unfinished git repos, each one an experiment that reached the point where I understood the thing and then stopped, never quite becoming something worth sharing with the world. A graveyard, if you're feeling dramatic. A junk drawer, if you're being honest.

Here is the part I keep circling. The walk on the page is the exact opposite of how I work. One path, one direction, west. You cannot flit on a pilgrimage. You cannot open six tabs and leave them humming. There is the road and the next step and the cathedral at the end, and the whole appeal is that you are doing precisely one thing with your whole self.

I did this once for real. My brother Adam and I walked Hadrian's Wall, east from Bowness-on-Solway, wild camping the eighty-four miles to Wallsend, fifteen kilos on my back and eighteen on his. We set off in November, which was a decision. Cold, wet, the days so short we spent hours walking in the dark by headtorch, the world shrunk to the small bobbing circle of light in front of my boots. One afternoon we were low on water, soaked through and leaning into the wind, when we found a tap set into a wall. We were thrilled. It didn't work. We came round the next corner and there was a cafe, open and warm, and I am not exaggerating when I say the soup, crusty bread and hot coffee rearranged my understanding of luxury. That is what the narrowness does. Shrink the world to the next step and a bowl of soup becomes enormous. The scenery helped, the wall riding the hills with horses and sheep and cattle for company. But the thing I carried home wasn't the view. It was the quiet of having exactly one thing to do.

On the approach to Carlisle we hit a flooded underpass, the path simply gone under a temporary lake. I still have the photo of Adam, bandana on, slumped at the edge of the water that had closed the path. I was scanning it for a way through like a bug newly surfaced in code review, turning it over, hunting the clean fix. We dreamed of a kayak washing up at the water's edge, the simple solution, but the November days were too short for dreams and wishes. So we went with brute force, hauling our combined thirty-three kilos of backache over gate after gate on a three-kilometre detour. In hindsight the lake was probably an overstated puddle and we should have waded it. No matter. I've made my peace with the fence; standing in the water weighing it up gets you nowhere, and the only move that counts is the one that goes forward.

Even a flooded path has a direction. The desk doesn't. The quiet of one thing at a time is gone.…

https://t.co/1EauwqZLBC

412K

Paul Stamp (Stampy) @paulstamp

13 days ago

@AndyLxNen Ironically, you've inspired me to write my next post on the guilt of automation and the branding of AI slop. I hope this doesn't land unwanted in your feed.

Who to follow

Andrea Sancio

@asancio

Technical Director @NMGames // @Zynga // @BAFTA Games Committee // Prev: @ChineseRoom (VTM: Bloodlines 2), @DoubleElevenLtd (RimWorld CE)

Roberto Dillon

@rdillon73

Author and Professor @JCU_Singapore. (ISC)2 Member, IEEE Senior Member. Into #gamedev and #cybersecurity. Views expressed are my own. 😎

Develappr

@develappr

AI Dev, Decision Intelligence, Crypto Trading.

Paul Stamp (Stampy) @paulstamp

13 days ago

@AndyLxNen You’d think that, as a Senior Lecturer in Computing, you’d understand that you are the algorithm. By interacting with this post, you’ve created a self-fulfilling prophecy. And come on, Andrew, trolling the posts of people in your own industry is surely an act of self-sabotage.

Paul Stamp (Stampy) @paulstamp

13 days ago

@AndyLxNen Perhaps read the article first before making unsubstantiated assumptions.

paulstamp retweeted

Jack Stebbins

@jackstebbs

15 days ago

technology has gotten too powerful

113

974

394K

Paul Stamp (Stampy) @paulstamp

15 days ago

@wheeliethebin Glad I didn't make any grandiose claims about my attention to detail 😅

Paul Stamp (Stampy) @paulstamp

15 days ago

@wheeliethebin haha!! how did I not see that?

Paul Stamp (Stampy) @paulstamp

16 days ago

Deep focus wasn't a virtue — it was a response to a cost that AI has quietly removed, and now we're all stuck with the junk drawer of our scattered attention. https://t.co/joCBraYm7m

264K

paulstamp retweeted

🅶🅾🆁🅺

@Raclure03

30 days ago

SHUT UP AND TAKE MY MONEY 💲💲💲💲💲

494

28K

Paul Stamp (Stampy) @paulstamp

about 1 month ago

"Vibe coding" and "agentic engineering" miss what AI-assisted programming actually feels like — it's closer to being a magazine editor who commissions, reviews, and ships code they're accountable for. https://t.co/kxIhircS5M

Paul Stamp (Stampy) @paulstamp

about 1 month ago

"Your GitHub README Is Broken for UK Developers (and You Haven't Noticed)" by Paul Stamp #DEVCommunity https://t.co/gBOmSflpex

Paul Stamp (Stampy) @paulstamp

about 2 months ago

Introducing https://t.co/6Q5NXAHtzu Directory of Unity packages and projects that surfaces popular dependencies. Not sure about the green though, what do you think?

Paul Stamp (Stampy) @paulstamp

about 2 months ago

ServiceKit V2 is now out. A service locator for Unity, on purpose. No framework lock-in, attribute-based registration, and async resolution that finally fixes the startup-order headache. https://t.co/wrrmnsCyGM #Unity3D #gamedev

118

Paul Stamp (Stampy) @paulstamp

6 months ago

Nano Banana still showing some political bias. Prompt "draw an image of a vending machine full of politicians faces as masks in the style of Stefan Große Halbuer"