Skip to content

Faster and Worse

Design & technology research, ideas, & opinions by Stephen Farrugia

Tag: ai

Reverse Vapourware.

by

Vapourware claims to solve a real problem in a way that seems impressive for its time. Your money is gone before the truth comes out: the purpose is real but the product is vapour. But vapourware doesn’t work with the subscription model of the Software-as-a-Service market. Our perception of software as a product has changed as well. Marketing for software emphasises features, properties and potential rather than any concrete purpose. The products are real but the purpose is vaporising.

The internet, the web, and email give us unprecedented access to other people, information, and entertainment. They serve their purpose. That’s why they’re irresistible. In contrast, algorithmic content feeds induce engagement to supplement their purpose. They are irresistible by design.

Criticism of harmful tech should always be aware of this distinction. When ignored the criticism is easy to dismiss as anti-progress. Paul Graham does exactly this in his 2010 essay The Acceleration of Addictiveness.

“It’s the same process that cures diseases: technological progress. And when progress concentrates something we don’t want to want—when it transforms opium into heroin—it seems bad. But it’s the same process at work.”

Paul Graham – The Acceleration of Addictiveness, July 2010

The conflation of addictive usefulness and designed addictiveness strengthen his dismissive stance. It’s also important to recognise his arguments hinge on technology and not products. AI doomerism fuelled by proprietary AI product releases does the same thing. Capitalist owned AI products are not getting an inch out of control if no one is paying for them. But call them technologies and the business accountability vaporises.

“You can’t put the genie back in the lamp” builds on the technology generalisation to create a sense of inevitability. It implies we are the ones who need to adjust and adapt, not the genie. It implies that these things will hurt us but only if we don’t learn how to protect ourselves from them. Debates over abstract concepts like addictive technology or existential AI risk distract us from foul play on a product level.

Gamification and manipulative engagement techniques allow products to thrive without a concrete purpose. Marketing for Notion uses broad purposes like “productivity” and “collaboration”. People love using Notion but they have to define their own purposes that the software can serve.

Marketing for reverse vapourware contains no trace of purpose at all. Web3 may be the best example of this. Web3 marketing, CEOs, and VCs rarely claim a concrete purpose. If they do it’s either dependent on some future event, described as unrealised potential, or doesn’t hold up to five minutes of critical thought.

We define our own purpose and for a product which can cause harm, finding a purpose is the fool’s errand. The businesses behind the products choose the purposes they endorse and distance themselves from the ones they oppose.

They can define the criteria for their own success and they free themselves from any criteria for failure.

Everything is Beautiful All of the Time.

by

The wait for the successor to OK Computer was a tough time for Radiohead dorks like me. Coldplay and Muse may not be here if it weren’t for the superficial itch-scratching they delivered to us desperately impatient fans. But when Radiohead finally released an album in October of 2000 it was a shock to anyone who enjoyed the one before it. The guitars were replaced by electronic tones and distorted vocals. Bleeping, blooping, bullshit.

After I let go of my expectations and gave it more time those bleeps and bloops began coming together. Kid A grew on me and became a favourite album, one I listen to over twenty years later. My perception of Radiohead changed from a band that was comparable to other alt-rock groups into something else. Other music I enjoyed before began to feel thin and limited. My mind had been opened a notch more than before.

Perseverance after the unexpected

What if, before Radiohead released the album, an AI system generated a bunch of albums as possible successors to OK Computer and fans were asked to choose the ones they preferred. What are the chances they would pick something as confronting and unconventional as the one Radiohead created?

This isn’t an attempt to define art, nor is it an attempt to raise one form of art above another. The variations in how things are made are unquantifiable; unique to each maker. The same goes for how those things are received.

This is about what happens when a creative work gives us more after we give it more; and what makes us do that.

It has nothing to do with what prompts us to give something another chance; instead it’s about what we require in order to believe that it’s possible for more to be there. A friend might tell us about the subtext we missed in a boring film but we’re not going to rewatch it without a thought of the director or writer and their intent.

Nick Cave’s response to a song generated by AI in his style may be dramatic and overly fond of suffering as a prerequisite but he makes a strong point:

What makes a great song great is not its close resemblance to a recognisable work. Writing a good song is not mimicry, or replication, or pastiche, it is the opposite. It is an act of self-murder that destroys all one has strived to produce in the past. It is those dangerous, heart-stopping departures that catapult the artist beyond the limits of what he or she recognises as their known self.

Nick Cave – The Red Hand Files, January 2023

Cave confines his view to the interests of the artist but it applies as much to the receiver of the art. We give more because we hope to get something back and we hope it will be something new that catapults us forward.

Superficial beauty is pleasant but it can only hold you until it is replaced by something else. Superficial beauty can be automated because it follows conventions of composition. We can’t break conventions of composition without a genuine intent to communicate something in a way that is shaped by what is being communicated.

Sometimes we dig into a creative work but find nothing; an empty void. It doesn’t mean we failed to dig enough or that the artist failed in their communication. As long as we know something was made to express something other than superficial beauty, we’re left to like it or not; we can’t invalidate it as failure because that denies the artist of the integrity in their intent. An unconventional composition created without any intent to communicate something is noise; any beauty would be accidental.

We don’t need to reduce artistic expression to a product of pain and suffering. If that were true we’d share our galleries with the works of elephants, tigers, and snails. Sympathy is not empathy.

When a chimpanzee throws its shit at a tree we don’t consider it a remarkable expression of the chimpanzee condition. Conversely, if we frame that shit and display it in the Louvre the chimpanzee won’t feel any more understood. We don’t know how a chimpanzee would express themselves through art because we don’t know what a chimpanzee wants to express and how they would express it.

Art communicates through a kinship between the artist and the observer. This translates into a relationship of trust because we have to believe that the artist created the work with genuine intent. Artists need to be observers as much as the observers they expect to appreciate their art. How can we trust the intent of an artist that can’t relate to a person’s appreciation of an old hand-woven rug over a mass-produced one?

A computer can’t synthesise that human artistic expression without synthesising the trust required for a human to appreciate it. That is a dishonest relationship which survives by keeping everything as conventional as possible. The result is infinite variations of aesthetic appeal which mask the fact that every variation is composed within a set of rules that gained legitimacy in the past. (See “Corporate Memphis“)

Consider how much the web has been visually standardised to the point it creates an almost unreasonable yearning for something different. That’s a symptom of the synthesised trust that conventional composition techniques result in. Jony Ive’s flattening of the iOS gui; Google’s Material Design project; Facebook’s uncustomisable profile pages. They all contribute to a pool of automatable resources that can make a crypto exchange appear legitimate despite almost any suggestion to the contrary.

That yearning for something different is a yearning for those unconventional compositions and it feels unreasonable because we know deep down that they won’t make any sense if their purpose is purely ornamental.

Ineffective Automation.

by

I would give anything to be able to pay $28,500 for my computer. Anything! And I can’t, I can only buy these frigging $1000 computers that don’t do what computers are really good for. If you think about it, this is completely fucked up. People are valuing their cars more than their computer? They don’t have any idea what a computer is, they’re just using it to play movies. If you think about this, this is the corruption of consumer electronics and that the computer is basically for convenience rather than for actually doing primary needs.

Alan Kay interviewed by Adam Fisher in Palo Alto, August 2014 – Valley of Genius podcast Season 1, Episode 5 (https://twit.tv/shows/valley-of-genius/episodes/5)

It’s hard to ignore the persistence and flourishing of printed books despite an abundance of digital alternatives. Since computers have had screens they’ve had text. Yet MP3s killed CDs and only the most dedicated buy their movies on discs instead of online.

This difference between books, music, and movies, poses a question; what makes digital automation effective for us?

The easy answer is that automation makes things easy. It takes the laborious bits away so you can enjoy the bits you want.

The easy answer makes sense for music and movies because most of the laborious bits of physical media are peripheral to the listening and watching part, not part of it. But reading a physical book is different, it’s not quite as passive. The physical attributes of the printed book are intertwined with the reading part.

Here’s what I mean by that; to listen to recorded music, physical media or not, you use your hands—or voice—to start and stop the music, but your ears are all you need in between.

Printed books, though, need to be held up and open and you can’t stop holding them until you’re finished reading. So the digitisation of physical media for music and movies is effective for us because its only change to the listening and watching part is the potential for higher quality audio and visuals. It automates away the peripheral bits between us and the good stuff because that’s where the physical media was most problematic.

For reading there’s an effectiveness of the printed book that no amount of paper-like display and skeuomorphic interface has managed to fully capture. E-Readers still need to be held like a book. They have pages like a book, and can be bookmarked too. You can highlight passages and write notes. They also do things books can’t do like having internet connectivity, instant purchases, and the ability to hold hundreds of books at a time; none of which have much to do with reading.

Consider the fact that it’s very difficult to do something else while reading a physical book because your hands are required at all times. It’s a binding that makes reading a deliberate activity, you can’t do a Sudoku while reading a book. Yet almost every feature an E-Reader brings to the party is an invitation to do something other than read.

So, what if that binding to the book is what makes the printed book effective for so many people?

This is where automation—or computation—can be seen from a different perspective, a counter-intuitive one. It’s what I think Alan Kay is referring to in his frustrated quote. What reason does the average person have to be interested in the potential of a computer when hardware upgrades are prompted by software that tells you to upgrade, either explicitly or implicitly. Outside of video game and 3D graphics, the computing power of a device does not correlate with any increasing or decreasing amount of tangible effectiveness. We are being sold speed but without any concrete reason for why we need it.

It’s easy to say that automation is what makes things easy because it’s a lazy way to describe the potential for computation. It ignores the potential for automation to make things more involved, more purposeful, more effective. It opens friction to being something more than a UX dichotomy of being good or bad and instead as being varying degrees of effective.

This doesn’t mean that E-Readers should have a spring loaded cover that requires extra effort to hold them open. What it means is to recognise any exertion of energy as being everything it is in addition to being an exertion of energy. A fish swimming against the current could save plenty of energy by turning around, but they aren’t thinking about the swimming.

The effort exerted while reading a printed book minimises awareness and access to any potential outside of what is being read. The book wants your attention, it doesn’t work without it. We don’t long for some device that will hold the book and turn the pages for us because our hands are effective. We’re only aware of things that interfere with whatever we’re trying to do; obvious problems.

It’s not about having a sentimental attachment to the way things have been done. Far from it. This is a design concern; a way to design with a purpose constrained by the circumstances that are relevant rather than the ones that seem relevant.

Printed books live on because digital books automate all the things that seem relevant about reading a book. E-Reader technology is a collection of features that satisfy an array of purposes related to reading—the peripheral concerns—instead of having a purpose of making reading more effective. If we strip away the internet connectivity, bookmarking, highlighting, and dictionary lookups, we’re left with an inferior imitation of a printed book that needs to be recharged.