387: Google's Bad Week, Topicus Q4, Bill Simmons vs Spotify, Young Bill Gates, Reddit IPO, and Ford
387: Google's Bad Week, Topicus Q4, Bill Simmons vs Spotify, Young Bill Gates, Reddit IPO, and Ford
"How is murder different in a world where you can come back?"
Luck favors the persistent. This simple truth is a fundamental cornerstone of successful company builders. The builders of visionary companies were highly persistent, living to the motto: Never, never, never give up.
—James C. Collins and Jerry I. Porras
🎥🎬🍿🧠🕵️♂️ It doesn’t happen to me very often, but I had a film idea (I wrote about a different idea for a fiction short story in the intro of edition #144). It’s not a detailed story, more world-building and some general concepts that I find fun to play with.
Here’s the outline:
This is a future world where minds can be backed up, and if you die, the backup can be implanted in a fully grown clone of yourself (a bit similar to Richard Morgan’s novel Altered Carbon — I know there’s also a TV show, but I haven’t seen it).
The film follows the perspective of the protagonist. You don’t know things they don’t know.
The fun twist is to do a Memento-style Nolan technique where once in a while, the hero goes to their regular backup session *and then immediately wakes up in a special room, disoriented, being told by people that they are a backup copy that was just booted up in a clone*.
This means that they died sometime after this backup session, but before the next. What took place in that missing time? Who could want them dead? I imagine a kind of Deckard from Blade Runner trying to piece it together from conversations, security camera recordings, etc. But who can you trust?
You end up with a detective working on their own death’s case, and with killers out there who remember quite well what happened and may be planting false information or manipulating the hero into going after the wrong people (Memento’s Teddy? Don’t trust his lies!). There’s definitely a Total Recall vibe.
The identity of the killer is a mystery, but things could be made more suspenseful by giving the viewer reasons to have doubts about multiple characters close to the hero. This works especially well if you discover that backups could be hacked to remove or add information or even slightly alter people’s personalities. After you wake up as a clone, you can’t be 100% sure you can trust your own memory.
A neat B-plot may be to look at the impact of the hero’s violent death on their family and loved ones, even if they are still alive. How is murder different in a world where you can come back..?
You could re-use the death/backup device a few times in the film to make things really weird… You see them do another backup session — blink — they wake up as a new clone, and the date on the calendar is 8 days later or whatever.
Optional: You could have different models of clones, with different physical and mental abilities, and some of these abilities could be useful in solving the mystery.
You can increase the stakes by having the backup company contact the hero and say that someone tried to access the facility and destroy your consciousness’ backup, etc. You aren’t immortal, and shady people are after you AND your backup…
Optional: As part of the final reveal, the hero could meet a clone of themselves — who’s the “real” one, has the other’s mind been modified through the backup process? Have I been modified? What is the nature of consciousness and identity in a world where this is possible?
Is this any good?
I have no idea ¯\_(ツ)_/¯
But if well-executed, I’d certainly watch something set in that universe.
Villeneuve, Nolan… Looking for a new project? Call me 📞
📲📰 I’ve been experimenting with the Artifact app.
It’s the new baby of the two co-founders of Instagram, Kevin Systrom and Mike Krieger.
The elevator pitch is Tiktok, but for news.
When you first sign up, the app asks you to provide some of your interests. Then you get a feed of news items, and as you click on things, spend more or less time reading things, give it signals with 👍/👎, and answer some of its follow-up questions about things you are interested in, it tweaks the feed to be more personalized.
I don’t have high expectations for it, and I’ve never been interested in something like Apple News, but so far it’s pretty good. It has a good hit rate and presents a good variety of sources.
The real challenge will be to see if it beats a well-customized Reddit homepage 🤔
💚 🥃 🐇 This is the free edition of Liberty’s Highlights with 14,900+ subscribers.
You can get 2 extra editions/week full of juicy stuff + access to the private Discord 🗣🗣🗣 community by becoming a paid supporter (it’s quick & easy).
Paid post since last week 🐇:
If you click the links above, you can see the intros for free and there’s a link to get a 7-day free trial of the paid version. What have you got to lose?
🏦 💰 Liberty Capital 💳 💴
Goldman Sachs chart showing 2022 market returns by strategy/industry 📈📉💸
h/t Gavin Baker
🤕 Maybe it will ultimately be good for Google that they had a bad week 🤔
Google had a pretty bad week…
I know, Bing’s chatbot also makes errors and hallucinates stuff — this should surprise no one since ChatGPT and GPT-3.5 also have this tendency. I think it’ll take a while for this problem to be mitigated (and it may never be *fully* solved).
But nonetheless, the general perception was that Google was caught flat-footed, panicked, and rushed out a series of announcements that kind of underwhelmed because expectations are different for 💪GOOGLE🥇 than for tiny little Bing🥉 over there in the corner…
It came out that employees were angry at Sundar Pichai for this and the way the recent layoffs were mishandled (some employees received less comp than they were promised, etc).
Some important voices in tech even started to wonder out loud if maybe Pichai was the wrong person for the job. Does Google now need a wartime CEO..?
If we zoom out, I doubt all this will matter a year from now — in fact, maybe things will turn out better for Google because they had this terrible period.
It'll probably light a fire under them, make them overcompensate and work even harder at getting it right than if they never felt threatened (even if just perception-wise).
One of the big challenges for large, ultra-successful companies is staying hungry. Maybe an attack on the core of the company will be that wake-up call…
How many Google users know which result links are ads and which aren’t? 🔎📃
I’ve long believed that there’s something vaguely unsavory about Google’s search ads being so similar to organic results. How many people click on ads while thinking that they’re clicking on organic search results?
*Nothing* about Google’s result pages is an accident.
Google could, if they wanted, make search links green or put them in a box with a slightly darker background, or make them stand out in any number of ways.
They very carefully A/B tested them to be just noticeable enough that they can say “ads are clearly labeled” yet be ambiguous enough that a large fraction of the population who doesn’t care about such things and isn’t paying much attention probably has no idea what is going on (people’s eyes are trained to read the blue headlines because that’s what contains the primary info to help find the best link for them — the black “Ad” is probably not registering for most users).
The graph above shows the results of one survey that appears to show that the % of people who don’t know is growing over time (I’m not sure of the source — h/t to my friend MBI (💎🐕) for passing it along).
I don’t know if those ratios held over time, but I wouldn’t be surprised if they did.
🦎 Topicus Q4 results 📊
Friend-of-the-show Leandro has a good thread looking at Topicus’ most recent results.
I won’t summarize the whole thing, you can check it out for all the details, but I want to highlight these graphs since they’re so central to how the company creates value.
Bill Simmons convinced Spotify CEO Daniel Ek to keep podcasts non-exclusive 🎧🔓
I was wondering about this very thing!
I listen to The Rewatchables, which is a Ringer podcast about films.
When they were acquired by Spotify, I assumed they would go exclusive and I would probably stop listening (I previously wrote about how I had almost not heard anything by Rogan since he went exclusive — I didn’t plan it, it’s just that when I want to listen to a podcast, I open Overcast).
But to my surprise, they’re still available outside Spotify!
This is probably the back story:
The Swedish audio streaming giant had spent hundreds of millions of dollars to purchase podcast production companies and big name creators in the hopes of luring new subscribers to the platform. Now, Spotify chief content officer Dawn Ostroff […] was ready to stop many of these creators and companies from sharing podcasts on Apple and Amazon, and keep the content exclusively on Spotify.
Then Bill Simmons sent an email to her boss.
Simmons had sold the sports and pop culture audio empire The Ringer to Spotify a year earlier for $200 million. Now he wrote Spotify CEO Daniel Ek to argue for keeping the Ringer’s mass audience on Apple and its advertising revenue, driven by the explosion of sports betting.
Simmons won the argument. But that 2021 dispute exposed deep questions about the strategy behind Spotify’s billion-dollar bet on podcasting. In January, Spotify pushed out Ostroff and canceled nearly a dozen shows at its highest-profile podcast investment, the studio Gimlet.
I’m glad that the open ecosystem for podcasts seems to be winning.
The last thing I need is to have 4-5 podcast apps that I need to bounce between to listen to all the shows I want, each in their own exclusive silos.
Imagine if you needed different web browsers to access different websites? 😬
🗣️ Interview: Sam Altman
It’s interesting to revisit this extended interview with Sam Altman, CEO and co-founder of OpenAI, from June 2021, with the context of everything that has happened since:
🎧 Ezra Klein Interviews Sam Altman (audio & transcript)
There are things I like about OpenAI, and others I like less, but regardless of any of that, I think that as one of the main players in the space, it’s worth getting familiar with how he thinks about things.
A Young Bill Gates (Podcast) 💾 💾 💾 💾 💾 💾 💾 💰
A *great* episode about the book Hard Drive on Bill Gates’ early years by friend-of-the-show and supporter David Senra (📚🎙️):
David had previously covered the book Overdrive about the period that comes after this book, and I also recommend it:
Ford to build $3.5 billion lithium iron phosphate EV battery plant in Michigan 🔋🔋🔋🔋🔋🔋
This is a partnership with China’s Contemporary Amperex Technology Co. Limited (better known as CATL), the battery gorilla right now.
Ford says that they’ll only provide “knowledge and services”.
The construction project will generate 2,500 jobs for the region, with initial production expected to commence in 2026.
Kinda slow and small compared to Tesla’s multiple battery gigafactories, but gotta start somewhere.
🤖 AI helping you get a job? 👨🏻💼💼🗄️📎📋
An interesting study published by the National Bureau of Economic Research:
There is a strong association between the quality of the writing in a resume for new labor market entrants and whether those entrants are ultimately hired. We show that this relationship is, at least partially, causal: a field experiment in an online labor market was conducted with nearly half a million jobseekers in which a treated group received algorithmic writing assistance. Treated jobseekers experienced an 8% increase in the probability of getting hired.
Reddit apparently wants to go public in the second half of 2023
Oh boy, that'll be something!
Social media companies are always surrounded by drama, and being public just magnifies that as everybody — especially journalists looking for a grabby headline — will find any craziness on the site and put a spotlight on it.
🧪🔬 Liberty Labs 🧬 🔭
🤖☔️🌧️Google MetNet-2 Deep Learning model is best at predicting rain 12 hours in advance
Here’s a very interesting paper in Nature:
Existing weather forecasting models are based on physics and use supercomputers to evolve the atmosphere into the future. Better physics-based forecasts require improved atmospheric models, which can be difficult to discover and develop, or increasing the resolution underlying the simulation, which can be computationally prohibitive.
An emerging class of weather models based on neural networks overcome these limitations by learning the required transformations from data instead of relying on hand-coded physics and by running efficiently in parallel.
Extracting the patterns directly from the data!
Here we present a neural network capable of predicting precipitation at a high resolution up to 12 h ahead. The model predicts raw precipitation targets and outperforms for up to 12 h of lead time state-of-the-art physics-based models currently operating in the Continental United States. The results represent a substantial step towards validating the new class of neural weather models.
A very concrete way that AI is making everyday life better: Get ready to have much more reliable rain alerts in the near future!
And if you’re a real nerd and want to play with the code, the model has been released under a very permissive Apache license.
🎨 🎭 Liberty Studio 👩🎨 🎥
The Amazing Special FX Tech used for Avatar 2 (doesn’t matter if you haven’t seen it, this is interesting in itself!)
I haven’t seen the new Avatar, but man, the technology they had to invent to make that film is just 🤯
The water simulation, the underwater performance capture, the neural net for facial expressions, how they do depth capture to composite real and virtual characters in the same scenes…
This video is worth watching even if you don’t care one iota about Avatar as a story.
That film idea is great! Thanks for all you do.