Antonym Google=Kodak Edition
"Dude. Your Mom does NOT have a Gutenberg bible lying around the house."
Dear Reader
Even Google AI thinks I shouldn’t use Google AI for search.
Fair enough.
I. The Dilemma
In 1997, Harvard business school professor Clay Christensen published a book that framed how businesses think about new inventions for the next 25 years. The Innovator’s Dilemma explained how big companies failed when small innovators eroded their advantage with new products. The big companies had smart people and could see the upstarts, but were trapped by the incentives of their existing business models, first fooling themselves that the new competitors were irrelevant, too niche or too different to affect them.
GM was humbled by Toyota. Kodak and Polaroid by digital cameras. Blackberry and Nokia by the iPhone. Companies that seemed unassailable in their heyday, disappeared or were sidelined.
Also in 1997, two Stanford students were working on a search engine that they would launch the following year as Google. Over the next decade the product first took the markets of internet rivals, then quickly started eviscerating the fat profit margins of newspapers, TV stations and radio.
Google knew all about the innovator’s dilemma. They lived it, rode the disruption wave. But they were also haunted by its violent logic – one day, they knew, the disruption would come for them.
After beating off social media and mobile revolutions, their Kodak moment has arrived in the form of generative artificial intelligence. The company’s employees invented it. It has spent billions acquiring the world's top AI companies. And yet, still they are not safe.
II. New tech in old shapes
So far, generative AI doesn’t fit well into pre-AI shaped tools. The monolithic, gargantuan money machine that has been Google search advertising for the last 20 years might not ever play nicely with generative AI tools.
Kodak tried to make digital cameras fit into their analogue-shaped products.
Like today’s big tech companies, the company attracted brilliant minds in the 1970s and 80s. It wasn’t stingy with the research budgets either, hence it was where the digital camera, which would be its undoing, was invented. It didn’t bury the invention, it attempted several times to create digital camera products, with varying success.
The problem was photographic film. It just made them so much money. They couldn’t help but try and use the new digital technology – destined to make physical film the preserve of enthusiasts and hobbyists – to keep the film business alive. Money, the streams of film revenue wouldn’t let them look away and imagine a future without it.
Here’s a diagram from 1990 New York Times article about Kodak’s new multimedia/digital photo system. Look at the desperate effort to invent the future in a Kodak film-friendly digital format:
Ten years later, Kodak again tried to embrace digital but keep camera film revenue.The Advantix Preview was a digital camera which let you select the images you wanted to commit to film. This concept did not last long.
Image: A Kodak Advantix Preview from 2000 Bigalid
So is this AI Overview feature a Kodak Advantix Preview moment for Google?
I’ll tell you one thing – apart from the insistence I should consider other search engines – there’s not a single ad in view for now, and not even a reason to go clicking on the SEO-drenched search results below…
III. Your Mom doesn’t have a Gutenberg bible
Yesterday, ChatGPT blessed me with 12 magical hours of its new Mac desktop app, complete with the fabled new voice mode, and the ability to see the screen I was working on. It was very cool, very useful. And then this morning the AI gods are angry with me and I’m back to just having access to the normal-for-now ChatGPT 4 and 4o models. Sucks to be me.
As a notional start up, ChatGPT can muck around like this. Sometimes it works, sometimes it doesn’t – you are using the most powerful thinking machine in the history of humanity, suck it up. For Google, the world’s less forgiving. Its attempt to integrate AI into its search engine results for users in the US has become the latest in a long line of product launch fails.
No doubt they are learning. But even if they are, it might not save them. They are trapped by their business model.
Users of Google in the US have found that it sometimes takes satirical articles as fact (it gave advice based on an article in The Onion that said geologists recommended “eating one small rock a day” as part of a healthy diet).
Although Google search’s AI feature recommended I use Perplexity as a search engine, that tool has similar accuracy issues. Yesterday Perplexity told me that a Gutenberg Bible was estimated to be worth $25 million if it came to auction today but that there had also been estimates of more than $100 million. The lower number had citations that checked out, but the $100 million number led to a Reddit post which had been deleted. The comments remained, giving some clues as to the expertise of the poster: “Dude, I seriously doubt your Mom has a copy of the Gutenberg bible lying around the house.”
IV. How to fail your way to the top
Learning to fail well is a skill which Google is not mastering. At least not publicly. But becoming an “expert failure practitioner”, as Amy Edmondson puts it, is key to learning and taking action in this turbulent age of AI revolution.
Ajit Jaokar of the Oxford University Said Business School, posted a summary of the UK Government CTO David Knott’s talk at an AI conference this week.
Human in loop appears a nice idea - but is not always useful nor pragmatic
[For example, the] Horizon scandal in the UK had a human in the loop but they believed in the machine (which was programmed in a flawed manner) and human in the loop negates the value of self driven cars (how could it be different from driving)
3) We should publish negative results (failure cases). I really loved this. It's actually so different from the marketing driven views we see today
(The Horizon scandal, for non-UK readers, is the tragic case of a computer system implemented by Fujitsu that caused many owners of small Post Offices to be falsely convicted and imprisoned for fraud, resulting in at least one suicide.)
The idea of publishing failure cases is excellent. Imagine if Fujitsu had done that?
I've read and re-read Amy Edmondson's book Right Kind of Wrong (out in paperback May 30, by the way) over the last six months. It's about how failures drive progress, and reads to me as a fieldbook for us all at the dawn of the age of AI. It describes the strength of being able to share failures well and become "expert failure practitioners''. It feels like something that should be – I wonder how much it will be – a key principle for us all right now.
Also failure related - a really inspirational expert failure practitioner popped up in my TikTok feed this week from Dr Michele Veldsman.
Tiktok failed to load.Enable 3rd party cookies or use another browser
An author who aimed for 100 failures a year inspired a consultant neuroscientist, Dr. Veldsman. This approach involves actively seeking opportunities without fearing rejection (because you’re aiming to clock up more rejections), which led to unexpected successes. If you don’t want to view the TikTok I’ve attached a transcript of the video to the bottom of this email.*
Aiming for rejection is similar to a technique that some salespeople use – I recall reading about a door to door rep who realised that 1 in 10 people said yes to their offer, which earned them $100. So each time someone said no they would walk away saying to themselves “thanks for the $10”.
Veldsman’s 100 fails a year is a great idea. But will I do it? Turning an insight like this into action isn’t simple. There are all sorts of cognitive biases nudging me away from actually embracing failures like this.
So, reader – I promise to report back next week on trying to notch up five failures and will report back on the results.
What did we do that we will do again? (I.e. what worked well.)
What would we do differently next time?
What did we learn?
What still puzzles us?
To emphasise the failure part, we might adjust these to:
What failed well?
How would we fail differently?
What did we learn from failure?
What still puzzled us about the failures?
VI. Recommendations
It’s been a long time, so here’s some things that are fantastic.
Everybody In The Place: An Incomplete History of Britain 1998-1992 by Jeremy Deller (YouTube)
This is a lovely watch when you have an hour. An artist – Deller won the Turner prize – telling a class of A level students the history of the UK in the 80s / early 90s through the story of rave music. It’s packed with ideas and explains the 1980s - 90s in fascinating ways. And even though I lived through it in my late teens (I turned 20 in 1992, at Glastonbury the year that dance music landed with full force at that festival) I didn’t have the whole history – as a white Londoner I’d not realised how the scene started all over the country, nor how deep its roots ran through the club and traveller communities). It’s not a nostalgia trip – it’s so much more than that – but the end scene of kids dancing at a club in Stoke-on-Trent made my heart ache with recognition for the days when me and my friends were lost in dance.
A lost Roman Alleluia
One more TikTok video. (I didn’t spend much time there recently, but boy its algorithm gets me.) A complete surprise, a recording of the Idrisi Ensemble, a polyphony choir, singing a song in - I think - the style of Corsican Polyphony. I’d never heard anything like it. But desperately want more.
That’s all for this week…
Thank you for reading. I hope something was interesting for you. I love writing these letters and sharing the things in my head. If you liked it, a “like” or a share is always highly appreciated.
Antony
* Transcript of Dr Michele Veldsman’s TikTok video about failure:
A very well-renowned Cambridge professor of neurology came up to me at a conference once and said, "Hey, I know you. I teach my students about your failure." And I actually absolutely loved that. A colleague of mine had featured me on her blog about how I basically count failures.
In academia, you get rejected all the time. You get rejected for jobs, for PhD places, for funding. Any paper you write just gets torn apart, probably by your colleagues and then by peer reviewers before it goes to publication. Awards, everything. So it's just constant rejection. I had seen some advice from an author who said that they basically count failures. They try to get 100 a year. In doing so, it forces them to apply to lots of things that they will get rejected from. I absolutely love this idea. So I started doing it. I started just actively counting rejections and failures, which aren't really failures.
The two things it did for me were: first, it made me apply for things that I normally wouldn't apply for because I wasn't actually waiting on the acceptance. I was waiting on the rejection. More often than not, I would forget that I'd applied for things. So when I actually got things, it was the nicest surprise. Secondly, I would also get things that I never actually would have applied for. I had a couple of instances where I applied for something, thinking I would never get it, and then got offered something even better. Those things have been such amazing drivers in my career.
The reason I thought of this today is because I applied for a position on a symposium organising committee for the World Health Organization. Again, it was like, no chance here. I got an email back to say, "Actually, love your experience. Here's another thing we'd love you to do for the WHO." Then another email to say, "Can you actually facilitate on the thing that you had originally applied for?" It's just amazing how applying for things and taking a chance can lead to some really awesome opportunities.